Probe E# — E-Sharp AB
Probe is made by E# (E-Sharp AB) in Sweden. Want your own boards reviewed like this? Book a meeting

WREN XIE

OHWR0091 · revision R2

WREN XIE (OHWR0091 R2) is an open-hardware circuit board from its public repository, designed in KiCad EDA: a 14-layer board of 162.5 × 100 mm with 803 components and 1125 nets, published under CERN-OHL-W-2.0.

review 92% walked · 17 findings: 1 blocker · 10 major · 6 minor (113 items)

Open in the interactive viewerRead the full review

WREN XIE (OHWR0091 R2) circuit board, top view: copper, silkscreen and components
Top view, rendered by Probe from the design's own fabrication data.

Design review

Probe recorded 17 findings: 1 blocker · 10 major · 6 minor (113 items) on this design. Each is shown here with its reasoning; the full review links every one to the pins, nets and datasheet pages it rests on.

  1. LedNoCurrentLimitblockerOpen

    LedNoCurrentLimit: 9 components×9

    Checked by the review, stands · Claude walker X1 (Cern session) — LD1, LD2, LD4, LD5, LD9-LD13 are Wurth 1313210530000 WL-ICLED parts, described on CERN's BOM/schematic as 'Right Angle WL-ICLED Integrated Controller within LED': RGB LEDs with an integrated constant-current driver, so VDD straight on 5VREG with no series resistor is the normal connection for this part type. The maker datasheet could not be obtained (we-online.com 403, recorded unobtainable), so the integrated current setting cannot be cited; the finding stays open for the owner.

    At LD1, LD10, LD11, LD12, LD13, LD2 and 3 more

    Each finding, with its own headline
    • LD1LD1 sits straight across '5VREG' and 'GND' with nothing in series

      LD1 is an LED with pin 3 on the supply net '5VREG' (5.0 V) and pin 2 on the ground net 'GND' — nothing sits in series with it, so its forward current is set by the rail's compliance and the die's knee rather than by the board. Structural, read from the netlist and the resolved net domains: a resistor, a driver or a GPIO on either end would put this LED on a different net and this rule would say nothing. If the supply net is in fact a current-limited driver output classified as a rail, refute this on that fact.

    • LD10LD10 sits straight across '5VREG' and 'GND' with nothing in series

      LD10 is an LED with pin 3 on the supply net '5VREG' (5.0 V) and pin 2 on the ground net 'GND' — nothing sits in series with it, so its forward current is set by the rail's compliance and the die's knee rather than by the board. Structural, read from the netlist and the resolved net domains: a resistor, a driver or a GPIO on either end would put this LED on a different net and this rule would say nothing. If the supply net is in fact a current-limited driver output classified as a rail, refute this on that fact.

    • LD11LD11 sits straight across '5VREG' and 'GND' with nothing in series

      LD11 is an LED with pin 3 on the supply net '5VREG' (5.0 V) and pin 2 on the ground net 'GND' — nothing sits in series with it, so its forward current is set by the rail's compliance and the die's knee rather than by the board. Structural, read from the netlist and the resolved net domains: a resistor, a driver or a GPIO on either end would put this LED on a different net and this rule would say nothing. If the supply net is in fact a current-limited driver output classified as a rail, refute this on that fact.

    • LD12LD12 sits straight across '5VREG' and 'GND' with nothing in series

      LD12 is an LED with pin 3 on the supply net '5VREG' (5.0 V) and pin 2 on the ground net 'GND' — nothing sits in series with it, so its forward current is set by the rail's compliance and the die's knee rather than by the board. Structural, read from the netlist and the resolved net domains: a resistor, a driver or a GPIO on either end would put this LED on a different net and this rule would say nothing. If the supply net is in fact a current-limited driver output classified as a rail, refute this on that fact.

    • LD13LD13 sits straight across '5VREG' and 'GND' with nothing in series

      LD13 is an LED with pin 3 on the supply net '5VREG' (5.0 V) and pin 2 on the ground net 'GND' — nothing sits in series with it, so its forward current is set by the rail's compliance and the die's knee rather than by the board. Structural, read from the netlist and the resolved net domains: a resistor, a driver or a GPIO on either end would put this LED on a different net and this rule would say nothing. If the supply net is in fact a current-limited driver output classified as a rail, refute this on that fact.

    • LD2LD2 sits straight across '5VREG' and 'GND' with nothing in series

      LD2 is an LED with pin 3 on the supply net '5VREG' (5.0 V) and pin 2 on the ground net 'GND' — nothing sits in series with it, so its forward current is set by the rail's compliance and the die's knee rather than by the board. Structural, read from the netlist and the resolved net domains: a resistor, a driver or a GPIO on either end would put this LED on a different net and this rule would say nothing. If the supply net is in fact a current-limited driver output classified as a rail, refute this on that fact.

    • LD4LD4 sits straight across '5VREG' and 'GND' with nothing in series

      LD4 is an LED with pin 3 on the supply net '5VREG' (5.0 V) and pin 2 on the ground net 'GND' — nothing sits in series with it, so its forward current is set by the rail's compliance and the die's knee rather than by the board. Structural, read from the netlist and the resolved net domains: a resistor, a driver or a GPIO on either end would put this LED on a different net and this rule would say nothing. If the supply net is in fact a current-limited driver output classified as a rail, refute this on that fact.

    • LD5LD5 sits straight across '5VREG' and 'GND' with nothing in series

      LD5 is an LED with pin 3 on the supply net '5VREG' (5.0 V) and pin 2 on the ground net 'GND' — nothing sits in series with it, so its forward current is set by the rail's compliance and the die's knee rather than by the board. Structural, read from the netlist and the resolved net domains: a resistor, a driver or a GPIO on either end would put this LED on a different net and this rule would say nothing. If the supply net is in fact a current-limited driver output classified as a rail, refute this on that fact.

    • LD9LD9 sits straight across '5VREG' and 'GND' with nothing in series

      LD9 is an LED with pin 3 on the supply net '5VREG' (5.0 V) and pin 2 on the ground net 'GND' — nothing sits in series with it, so its forward current is set by the rail's compliance and the die's knee rather than by the board. Structural, read from the netlist and the resolved net domains: a resistor, a driver or a GPIO on either end would put this LED on a different net and this rule would say nothing. If the supply net is in fact a current-limited driver output classified as a rail, refute this on that fact.

  2. observationOBS-650majorOpen

    R86 "WR DAC bypass" is DNP in the schematic but mounted in CERN's assembly BOM, which would tie IC10 VOUTB to the FPGA-driven WR_DAC.DIN

    R86 (0 Ω, "WR DAC bypass") is crossed out as not fitted on schematic sheet 12, but CERN's released assembly BOM lists it as mounted; built to the BOM, it shorts DAC IC10's output VOUTB onto WR_DAC.DIN, a line the FPGA drives.

    The two documents disagree

    DocumentR86Source
    Schematicdrawn crossed out (DNP), inside a note "WR DAC bypass"EDA-04811-V2-0_sch.pdf (EDMS 3447981 v.2), sheet 12
    Assembly BOMmounted: "R86, R94, R135, R138", 1 A zero-ohm jumper, YAGEO 232270591001LEDA-04811-V2-0_pcb-mat.xls (EDMS 3447985 v.1), "Parts mounted on the PCB"
    Assembly BOMIC10 DAC8562SDSC mounted too; only OSC5, R58, R136, R137, R161 are Not_Mountedsame file, Not_Mounted section

    What R86 connects when fitted

    ``mermaid graph LR FPGA["IC14 XCZU4CG pin AB9 (drives)"] -- WR_DAC.DIN --> DIN["IC10 DAC8562 pin 8 DIN"] FPGA -- WR_DAC.DIN --> R86["R86 0 Ω (DNP on sheet 12)"] R86 --> VOUTB["IC10 pin 2 VOUTB (DAC output)"] VOUTB --> R57["R57 240 Ω → tuning path"] ``

    With both IC10 and R86 fitted, the net carrying the FPGA's serial data into the DAC is also the DAC's own analogue output: two drivers on one net, and the tuning voltage follows the SPI data instead of the DAC code. The bypass only makes sense with IC10 left off.

    What to settle

    Which build is intended for V2-0: DAC fitted with R86 open (as drawn), or the bypass with IC10 not fitted. The assembly BOM and the schematic should then say the same thing.

  3. observationOBS-654majorOpen

    IC65 (SN74LV1T34 at 5 V) cannot see a high from FP_LED_DIN, a 1.8 V bank-65 output: the front-panel WL-ICLED chain gets no valid data

    IC65 is powered from 5VREG, where its VIH is 2.03 V, but its input FP_LED_DIN comes from FPGA ball L6 in HP bank 65, whose VCCO is P1V8: a 1.8 V high never reaches the threshold, so the data into the front-panel LED chain LD1..LD13 is not reliably recognised.

    Levels at IC65 input

    QuantityValueSource
    DriverIC14 ball L6, IO_L13N_T2L_N1_GC_QBC_65 (bank 65)schematic sheet 3 (fpga-hp-banks)
    Bank 65 VCCOP1V8 (VCCO_65 balls H5, J3, L4)netlist
    P1V81.8 VIC30 configuration table, schematic sheet 43
    Highest output high from the bankat most 1.8 V (VCCO)the output cannot exceed VCCO
    IC65 supply5VREG, 5.0 V (IC65.5)netlist; TPS62125 divider R210/R211
    SN74LV1T34 VIH at VCC 4.5-5.0 V2.02 V (25 C) / 2.03 V (-40 to 125 C)SN74LV1T34 datasheet, 6.5 Electrical Characteristics, p6
    Shortfall1.8 - 2.03 = -0.23 Vderived

    ``mermaid graph LR L6[IC14 L6, bank 65, VCCO 1.8 V] -- FP_LED_DIN --> IC65[IC65 SN74LV1T34, VCC 5VREG, VIH 2.03 V] --> LD1[LD1 DIN] --> LD2 --> LDn[... LD13] ``

    Why it matters

    The LV1T34's lowered thresholds are what make it a 1.8 V-to-higher translator, but only when its own VCC is low enough: at VCC = 1.65-2.0 V its VIH is 1.0 V, at 3.0-3.3 V 1.39 V, and at 4.5-5.0 V 2.03 V. Powered at 5 V it needs a 3.3 V-class input. The companion translator IC8 (SFP LED data) works because its input /SFP.LED_STAT_IN comes from ball Y13 in bank 44, which is at P3V3. Moving FP_LED_DIN to a 3.3 V HD-bank pin, or driving IC65 from a 3.3 V input, restores margin.

  4. observationOBS-657majorOpen

    F_PP_OE_N: a 1.8 V HP-bank pin (IC14 N6, bank 65) is pulled up to P3V3 by R40 and drives the 3.3 V OE# inputs of 16 SN74LVTH125 buffers

    F_PP_OE_N comes from IC14 ball N6 in HP bank 65 (VCCO = P1V8), but R40 10k pulls it up to P3V3: the pin sits 1.5 V above its 1.8 V supply, beyond the HP-bank input absolute maximum of VCCO + 0.55 V, and when the FPGA drives it high (1.8 V) it does not reach the 2.0 V VIH of the 16 SN74LVTH125 OE# inputs it controls, so the pp output buffers cannot be reliably disabled.

    Over-voltage on the FPGA pin

    QuantityValueSource
    DriverIC14 N6, IO_L15N_T2L_N5_AD11N_65 (HP bank 65)schematic sheet 3 (fpga-hp-banks)
    Bank 65 VCCOP1V8 = 1.8 Vnetlist (VCCO_65 H5/J3/L4); IC30 table sheet 43
    Pull-upR40 10k to P3V3 (3.3 V), fittednetlist R40; BOM
    HP I/O input voltage abs maxVCCO + 0.55 V = 2.35 VDS925 Table 1 (Absolute Maximum Ratings), p2
    Voltage the pull-up applies3.3 V, 0.95 V over the limit whenever the pin is released (configuration, tri-state, before VCCO_65 is up)derived

    The receivers cannot see the FPGA's high

    QuantityValueSource
    ReceiversOE# (pins 1, 4, 10, 13) of IC37-IC41, IC47-IC49, IC57-IC64 (16 x SN74LVTH125, VCC = P3V3)netlist
    VIH2.0 VSN74LVTH125 datasheet, recommended operating conditions, p3
    FPGA high when drivingat most 1.8 Vbank 65 VCCO

    ``mermaid graph LR P3V3 -- R40 10k --> OE[F_PP_OE_N] N6[IC14 N6, bank 65, 1.8 V] --> OE OE --> B[16 x SN74LVTH125 OE#, VIH 2.0 V] --> PP[PP_OUT0..31 on J6] ``

    Why it matters

    With the FPGA driving, the line only toggles between 0 V and 1.8 V, so the buffers' disable level sits in the undefined band and the pp outputs may stay enabled or oscillate; with the FPGA not driving, R40 forces current into the bank-65 clamp structure and back-powers VCCO_65 from P3V3 during power sequencing. The same OE scheme on a 3.3 V HD bank (as the front-panel OE lines F_FP_OE0/1 use) or a translator would fix both. This is the only 1.8 V-bank pin on the board with a pull-up to 3.3 V.

  5. BomLayoutMismatchmajorOpen

    BomLayoutMismatch: 2 components×2

    2 checked by the review by Claude walker X1 (Cern session) — 2 different reasons

    At HS1, TT1

    Each finding, with its own headline
    • HS1BOM refdes 'HS1' is not placed in the layout.
    • TT1BOM refdes 'TT1' is not placed in the layout.
  6. EsdProtectorPlacementmajorOpen

    EsdProtectorPlacement: 26 components×26

    Checked by the review, stands · Claude walker X1 (Cern session) — Checked against the copper (esdPlacement): the remaining findings are PRTR5V0U4D (D25-D34) on pp-connector J6 lines reached through vias or on 1.1-4.7 mm stubs, and the 0402ESDA-MLP ones are argued in OBS-652. The PRTR5V0U4D datasheet on file gives no placement distance or stub limit (searched: close / layout / place), so the 1 mm limit is layout practice, not a stated requirement; the findings stand for the owner. D4/D6/D7/D8 are answered by OBS-652.

    At D25, D26, D27, D28, D29, D30 and 8 more

    Each finding, with its own headline
    • D25D25.1 on '/pp_connector/PP_LED1_DATA': D25.1 hangs 4.73 mm off the line (limit 1 mm, default

      D25 (PRTR5V0U4D) protects '/pp_connector/PP_LED1_DATA' at J6.34. Along the copper: 8.92 mm of track from the contact, 0 via(s) before it, on a 4.73 mm branch off the line; the line runs 85.49 mm on to IC14.E14; D25 sits 3.05 mm farther from the contact than the nearest place outside J6.

      D25.1 hangs 4.73 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D25D25.3 on '/pp_connector/PP_ID0': D25 joins the line 13.17 mm from J6.35 on a 14.61 mm run to R10.1

      D25 (PRTR5V0U4D) protects '/pp_connector/PP_ID0' at J6.35. Along the copper: 13.17 mm of track from the contact, 0 via(s) before it, the line runs through its pad (flow-through); the line runs 14.61 mm on to R10.1; D25 sits 6.07 mm farther from the contact than the nearest place outside J6.

      D25 joins the line 13.17 mm from J6.35 on a 14.61 mm run to R10.1: nearer the far end than the contact, so the surge travels most of the line before it is clamped.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D25D25.4 on '/pp_connector/PP_ID1': D25 joins the line 9.49 mm from J6.36 on a 14.96 mm run to R28.1

      D25 (PRTR5V0U4D) protects '/pp_connector/PP_ID1' at J6.36. Along the copper: 9.49 mm of track from the contact, 0 via(s) before it, the line runs through its pad (flow-through); the line runs 14.96 mm on to R28.1; D25 sits 2.74 mm farther from the contact than the nearest place outside J6.

      D25 joins the line 9.49 mm from J6.36 on a 14.96 mm run to R28.1: nearer the far end than the contact, so the surge travels most of the line before it is clamped.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D25D25.6 on '/pp_connector/PP_LED0_DATA': 2 via(s) between J6.33 and D25.6

      D25 (PRTR5V0U4D) protects '/pp_connector/PP_LED0_DATA' at J6.33. Along the copper: 17.29 mm of track from the contact, 2 via(s) before it, on a 2.12 mm branch off the line; the line runs 95.28 mm on to IC14.D14; D25 sits 6.24 mm farther from the contact than the nearest place outside J6.

      2 via(s) between J6.33 and D25.6: the strike reaches an inner layer before it is clamped. D25.6 hangs 2.12 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D26D26.1 on '/pp_connector/PP_ID2': 2 via(s) between J6.37 and D26.1

      D26 (PRTR5V0U4D) protects '/pp_connector/PP_ID2' at J6.37. Along the copper: 11.34 mm of track from the contact, 2 via(s) before it, on a 4.5 mm branch off the line; the line runs 6.84 mm on to R33.1; D26 sits 2.04 mm farther from the contact than the nearest place outside J6.

      2 via(s) between J6.37 and D26.1: the strike reaches an inner layer before it is clamped. D26.1 hangs 4.5 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch. D26 joins the line 6.84 mm from J6.37 on a 6.84 mm run to R33.1: nearer the far end than the contact, so the surge travels most of the line before it is clamped.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D27D27.1 on '/io_drivers_pp/PP_OUT0': 2 via(s) between J6.32 and D27.1

      D27 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT0' at J6.32. Along the copper: 12.12 mm of track from the contact, 2 via(s) before it, on a 1.17 mm branch off the line; the line runs 38.32 mm on to R41.2; D27 sits 2.01 mm farther from the contact than the nearest place outside J6.

      2 via(s) between J6.32 and D27.1: the strike reaches an inner layer before it is clamped. D27.1 hangs 1.17 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D28D28.1 on '/io_drivers_pp/PP_OUT4': D28 joins the line 8.13 mm from J6.31 on a 11.81 mm run to R103.2

      D28 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT4' at J6.31. Along the copper: 8.13 mm of track from the contact, 0 via(s) before it, the line runs through its pad (flow-through); the line runs 11.81 mm on to R103.2; D28 sits 1.51 mm farther from the contact than the nearest place outside J6.

      D28 joins the line 8.13 mm from J6.31 on a 11.81 mm run to R103.2: nearer the far end than the contact, so the surge travels most of the line before it is clamped.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D29D29.1 on '/io_drivers_pp/PP_OUT8': 2 via(s) between J6.23 and D29.1

      D29 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT8' at J6.23. Along the copper: 13.49 mm of track from the contact, 2 via(s) before it, on a 1.35 mm branch off the line; the line runs 35.53 mm on to R219.2; D29 sits 5.71 mm farther from the contact than the nearest place outside J6.

      2 via(s) between J6.23 and D29.1: the strike reaches an inner layer before it is clamped. D29.1 hangs 1.35 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D29D29.3 on '/io_drivers_pp/PP_OUT9': D29.3 hangs 1.32 mm off the line (limit 1 mm, default

      D29 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT9' at J6.24. Along the copper: 6.98 mm of track from the contact, 0 via(s) before it, on a 1.32 mm branch off the line; the line runs 21.73 mm on to R223.2; D29 sits 2.01 mm farther from the contact than the nearest place outside J6.

      D29.3 hangs 1.32 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D30D30.1 on '/io_drivers_pp/PP_OUT12': D30.1 hangs 1.4 mm off the line (limit 1 mm, default

      D30 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT12' at J6.18. Along the copper: 12.25 mm of track from the contact, 0 via(s) before it, on a 1.4 mm branch off the line; the line runs 33.64 mm on to R234.2; D30 sits 2.16 mm farther from the contact than the nearest place outside J6.

      D30.1 hangs 1.4 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D30D30.3 on '/io_drivers_pp/PP_OUT14': 1 via(s) between J6.14 and D30.3

      D30 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT14' at J6.14. Along the copper: 6.24 mm of track from the contact, 1 via(s) before it, the line runs through its pad (flow-through); the line runs 21.22 mm on to R244.2; D30 sits 2.01 mm farther from the contact than the nearest place outside J6.

      1 via(s) between J6.14 and D30.3: the strike reaches an inner layer before it is clamped.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D30D30.6 on '/io_drivers_pp/PP_OUT13': 2 via(s) between J6.16 and D30.6

      D30 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT13' at J6.16. Along the copper: 9.57 mm of track from the contact, 2 via(s) before it, on a 1.44 mm branch off the line; the line runs 24.12 mm on to R239.2; D30 sits 2.05 mm farther from the contact than the nearest place outside J6.

      2 via(s) between J6.16 and D30.6: the strike reaches an inner layer before it is clamped. D30.6 hangs 1.44 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D31D31.1 on '/io_drivers_pp/PP_OUT16': 2 via(s) between J6.10 and D31.1

      D31 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT16' at J6.10. Along the copper: 11.7 mm of track from the contact, 2 via(s) before it, on a 4.68 mm branch off the line; the line runs 21.48 mm on to R246.2; D31 sits 3.05 mm farther from the contact than the nearest place outside J6.

      2 via(s) between J6.10 and D31.1: the strike reaches an inner layer before it is clamped. D31.1 hangs 4.68 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D31D31.3 on '/io_drivers_pp/PP_OUT17': D31 joins the line 7.71 mm from J6.8 on a 10.83 mm run to R249.2

      D31 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT17' at J6.8. Along the copper: 7.71 mm of track from the contact, 0 via(s) before it, the line runs through its pad (flow-through); the line runs 10.83 mm on to R249.2; D31 sits 2.74 mm farther from the contact than the nearest place outside J6.

      D31 joins the line 7.71 mm from J6.8 on a 10.83 mm run to R249.2: nearer the far end than the contact, so the surge travels most of the line before it is clamped.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D31D31.4 on '/io_drivers_pp/PP_OUT18': D31.4 hangs 1.62 mm off the line (limit 1 mm, default

      D31 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT18' at J6.6. Along the copper: 5.73 mm of track from the contact, 0 via(s) before it, on a 1.62 mm branch off the line; the line runs 34.77 mm on to R256.2; D31 sits 2.46 mm farther from the contact than the nearest place outside J6.

      D31.4 hangs 1.62 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D31D31.6 on '/io_drivers_pp/PP_OUT19': D31.6 hangs 1.15 mm off the line (limit 1 mm, default

      D31 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT19' at J6.4. Along the copper: 6.98 mm of track from the contact, 0 via(s) before it, on a 1.15 mm branch off the line; the line runs 24.51 mm on to R261.2; D31 sits 2.24 mm farther from the contact than the nearest place outside J6.

      D31.6 hangs 1.15 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D32D32.1 on '/io_drivers_pp/PP_OUT20': 2 via(s) between J6.2 and D32.1

      D32 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT20' at J6.2. Along the copper: 15.23 mm of track from the contact, 2 via(s) before it, the line runs through its pad (flow-through); the line runs 17.69 mm on to R262.2; D32 sits 5.99 mm farther from the contact than the nearest place outside J6.

      2 via(s) between J6.2 and D32.1: the strike reaches an inner layer before it is clamped. D32 joins the line 15.23 mm from J6.2 on a 17.69 mm run to R262.2: nearer the far end than the contact, so the surge travels most of the line before it is clamped.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D32D32.4 on '/io_drivers_pp/PP_OUT23': D32.4 hangs 1.13 mm off the line (limit 1 mm, default

      D32 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT23' at J6.17. Along the copper: 3.37 mm of track from the contact, 0 via(s) before it, on a 1.13 mm branch off the line; the line runs 30.67 mm on to R270.2; D32 sits 1.51 mm farther from the contact than the nearest place outside J6.

      D32.4 hangs 1.13 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D33D33.1 on '/io_drivers_pp/PP_OUT24': D33 joins the line 6.75 mm from J6.15 on a 9.21 mm run to R275.2

      D33 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT24' at J6.15. Along the copper: 6.75 mm of track from the contact, 0 via(s) before it, the line runs through its pad (flow-through); the line runs 9.21 mm on to R275.2; D33 sits 1.55 mm farther from the contact than the nearest place outside J6.

      D33 joins the line 6.75 mm from J6.15 on a 9.21 mm run to R275.2: nearer the far end than the contact, so the surge travels most of the line before it is clamped.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D33D33.4 on '/io_drivers_pp/PP_OUT26': D33.4 hangs 1.24 mm off the line (limit 1 mm, default

      D33 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT26' at J6.11. Along the copper: 7.52 mm of track from the contact, 0 via(s) before it, on a 1.24 mm branch off the line; the line runs 21.64 mm on to R280.2; D33 sits 1.51 mm farther from the contact than the nearest place outside J6.

      D33.4 hangs 1.24 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D33D33.6 on '/io_drivers_pp/PP_OUT27': D33.6 hangs 2.13 mm off the line (limit 1 mm, default

      D33 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT27' at J6.9. Along the copper: 3.43 mm of track from the contact, 0 via(s) before it, on a 2.13 mm branch off the line; the line runs 29.85 mm on to R282.2; D33 sits 1.51 mm farther from the contact than the nearest place outside J6.

      D33.6 hangs 2.13 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D34D34.6 on '/io_drivers_pp/PP_OUT29': 2 via(s) between J6.5 and D34.6

      D34 (PRTR5V0U4D) protects '/io_drivers_pp/PP_OUT29' at J6.5. Along the copper: 7.6 mm of track from the contact, 2 via(s) before it, on a 1.99 mm branch off the line; the line runs 19.93 mm on to R289.2; D34 sits 2.04 mm farther from the contact than the nearest place outside J6.

      2 via(s) between J6.5 and D34.6: the strike reaches an inner layer before it is clamped. D34.6 hangs 1.99 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D4D4.1 on '/fpga/fpga-config/JTAG_TCK': 1 via(s) between J7.6 and D4.1

      D4 (0402ESDA-MLP) protects '/fpga/fpga-config/JTAG_TCK' at J7.6. Along the copper: 9.33 mm of track from the contact, 1 via(s) before it, on a 1.15 mm branch off the line; the line runs 8.18 mm on to R52.1; D4 sits 2.74 mm farther from the contact than the nearest place outside J7.

      1 via(s) between J7.6 and D4.1: the strike reaches an inner layer before it is clamped. D4.1 hangs 1.15 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch. D4 joins the line 8.18 mm from J7.6 on a 8.18 mm run to R52.1: nearer the far end than the contact, so the surge travels most of the line before it is clamped.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D6D6.1 on '/fpga/fpga-config/JTAG_TDI': 1 via(s) between J7.10 and D6.1

      D6 (0402ESDA-MLP) protects '/fpga/fpga-config/JTAG_TDI' at J7.10. Along the copper: 6.59 mm of track from the contact, 1 via(s) before it, the line runs through its pad (flow-through); the line runs 7.99 mm on to R54.1; D6 sits 2.27 mm farther from the contact than the nearest place outside J7.

      1 via(s) between J7.10 and D6.1: the strike reaches an inner layer before it is clamped. D6 joins the line 6.59 mm from J7.10 on a 7.99 mm run to R54.1: nearer the far end than the contact, so the surge travels most of the line before it is clamped.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D7D7.1 on '/I2C.SDA': D7.1 hangs 1.84 mm off the line (limit 1 mm, default

      D7 (0402ESDA-MLP) protects '/I2C.SDA' at J3.2. Along the copper: 1.84 mm of track from the contact, 0 via(s) before it, on a 1.84 mm branch off the line; the line runs 16.13 mm on to R26.1; D7 sits at J3, within 0 mm of the nearest place outside it to the contact: the track to it is mostly the escape out of J3's own pins, which no placement shortens.

      D7.1 hangs 1.84 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

    • D8D8.1 on '/I2C.SCL': D8.1 hangs 1.81 mm off the line (limit 1 mm, default

      D8 (0402ESDA-MLP) protects '/I2C.SCL' at J3.3. Along the copper: 1.81 mm of track from the contact, 0 via(s) before it, on a 1.81 mm branch off the line; the line runs 11.68 mm on to R25.1; D8 sits at J3, within 0 mm of the nearest place outside it to the contact: the track to it is mostly the escape out of J3's own pins, which no placement shortens.

      D8.1 hangs 1.81 mm off the line (limit 1 mm, default: no datasheet figure on file): the line carries the strike past it while the clamp is reached through the branch.

      Measured on the copper the ODB states: tracks, the pads they land on, and the netlist's vias. Not a straight line between centroids.

  7. FloatingInputmajorOpen

    FloatingInput: 17 nets×17

    e.g. “IC18.1 on 'unconnected-(IC18-IN1-Pad1)' has no driver and no pull” · 17 different statements

    At unconnected-(IC18-FB_IN-Pad41), unconnected-(IC18-IN1-Pad1), unconnected-(IC18-IN2-Pad10), unconnected-(IC18-~{FB_IN}-Pad42), unconnected-(IC18-~{IN1}-Pad2), unconnected-(IC18-~{IN2}-Pad11) and 11 more

    Each finding, with its own headline
    • unconnected-(IC18-FB_IN-Pad41)IC18.41 on 'unconnected-(IC18-FB_IN-Pad41)' has no driver and no pull

      Net 'unconnected-(IC18-FB_IN-Pad41)': input(s) IC18.41 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

    • unconnected-(IC18-IN1-Pad1)IC18.1 on 'unconnected-(IC18-IN1-Pad1)' has no driver and no pull

      Net 'unconnected-(IC18-IN1-Pad1)': input(s) IC18.1 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

    • unconnected-(IC18-IN2-Pad10)IC18.10 on 'unconnected-(IC18-IN2-Pad10)' has no driver and no pull

      Net 'unconnected-(IC18-IN2-Pad10)': input(s) IC18.10 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

    • unconnected-(IC18-~{FB_IN}-Pad42)IC18.42 on 'unconnected-(IC18-~{FB_IN}-Pad42)' has no driver and no pull

      Net 'unconnected-(IC18-~{FB_IN}-Pad42)': input(s) IC18.42 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

    • unconnected-(IC18-~{IN1}-Pad2)IC18.2 on 'unconnected-(IC18-~{IN1}-Pad2)' has no driver and no pull

      Net 'unconnected-(IC18-~{IN1}-Pad2)': input(s) IC18.2 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

    • unconnected-(IC18-~{IN2}-Pad11)IC18.11 on 'unconnected-(IC18-~{IN2}-Pad11)' has no driver and no pull

      Net 'unconnected-(IC18-~{IN2}-Pad11)': input(s) IC18.11 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

    • unconnected-(IC18-~{OE}-Pad12)IC18.12 on 'unconnected-(IC18-~{OE}-Pad12)' has no driver and no pull

      Net 'unconnected-(IC18-~{OE}-Pad12)': input(s) IC18.12 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

    • unconnected-(IC19-~{RESET}-Pad3)IC19.3 on 'unconnected-(IC19-~{RESET}-Pad3)' has no driver and no pull

      Net 'unconnected-(IC19-~{RESET}-Pad3)': input(s) IC19.3 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

    • unconnected-(IC21-~{RESET}-Pad3)IC21.3 on 'unconnected-(IC21-~{RESET}-Pad3)' has no driver and no pull

      Net 'unconnected-(IC21-~{RESET}-Pad3)': input(s) IC21.3 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

    • unconnected-(IC3-CTS_ECI-Pad10)IC3.10 on 'unconnected-(IC3-CTS_ECI-Pad10)' has no driver and no pull

      Net 'unconnected-(IC3-CTS_ECI-Pad10)': input(s) IC3.10 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

    • unconnected-(IC3-CTS_SCI-Pad18)IC3.18 on 'unconnected-(IC3-CTS_SCI-Pad18)' has no driver and no pull

      Net 'unconnected-(IC3-CTS_SCI-Pad18)': input(s) IC3.18 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

    • unconnected-(IC3-GPIO.0_SCI{slash}DCD_SCI-Pad24)IC3.24 on 'unconnected-(IC3-GPIO.0_SCI{slash}DCD_SCI-Pad24)' has no driver and no pull

      Net 'unconnected-(IC3-GPIO.0_SCI{slash}DCD_SCI-Pad24)': input(s) IC3.24 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

    • unconnected-(IC3-GPIO.1_ECI{slash}DSR_ECI-Pad14)IC3.14 on 'unconnected-(IC3-GPIO.1_ECI{slash}DSR_ECI-Pad14)' has no driver and no pull

      Net 'unconnected-(IC3-GPIO.1_ECI{slash}DSR_ECI-Pad14)': input(s) IC3.14 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

    • unconnected-(IC3-GPIO.2_SCI{slash}DSR_SCI-Pad22)IC3.22 on 'unconnected-(IC3-GPIO.2_SCI{slash}DSR_SCI-Pad22)' has no driver and no pull

      Net 'unconnected-(IC3-GPIO.2_SCI{slash}DSR_SCI-Pad22)': input(s) IC3.22 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

    • unconnected-(IC3-NC{slash}VPP{slash}DCD_ECI-Pad16)IC3.16 on 'unconnected-(IC3-NC{slash}VPP{slash}DCD_ECI-Pad16)' has no driver and no pull

      Net 'unconnected-(IC3-NC{slash}VPP{slash}DCD_ECI-Pad16)': input(s) IC3.16 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

    • unconnected-(IC3-~{SUSPEND}{slash}RI_ECI-Pad17)IC3.17 on 'unconnected-(IC3-~{SUSPEND}{slash}RI_ECI-Pad17)' has no driver and no pull

      Net 'unconnected-(IC3-~{SUSPEND}{slash}RI_ECI-Pad17)': input(s) IC3.17 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

    • unconnected-(IC3-~{SUSPEND}{slash}RI_SCI-Pad1)IC3.1 on 'unconnected-(IC3-~{SUSPEND}{slash}RI_SCI-Pad1)' has no driver and no pull

      Net 'unconnected-(IC3-~{SUSPEND}{slash}RI_SCI-Pad1)': input(s) IC3.1 with no driver and no pull — floating input (undefined level); add a driver or a pull resistor.

  8. PartLifecyclemajorOpen

    PartLifecycle: 4 part numbers×4

    4 Obsolete · on 7 placements

    At IC1, IC19, TJ1, XJ4

    Each finding, with its own headline
    • IC1SN74CB3T3384PW on IC1: the distributor catalogue reports "Obsolete" for Texas Instruments's SN74CB3T3384PW

      SN74CB3T3384PW on IC1: the distributor catalogue reports "Obsolete" for Texas Instruments's SN74CB3T3384PW — a sourcing/longevity risk to confirm against the MANUFACTURER before acting. That is the maker this board's BOM names (TEXAS INSTRUMENTS). A distributor delisting is not the same claim as the part being unavailable; check the vendor's own lifecycle/product-change notice for the status and any named successor. Re-read the source: GET /api/odb/parts/SN74CB3T3384PW/providers (which provider said it, the maker and part number it matched, its own words, and when we asked). THE LAST BUY DATE HAS PASSED (2024-12-12, 703 day(s) ago), so this is a substitution rather than a buy. The distributor reports 450 left. No approved alternate is recorded for this position.

    • IC19MT25QU512ABB8ESF-0SIT on 2 placements (IC19, IC21): the distributor catalogue reports "Obsolete" for Micron Technology Inc.'s MT25QU512ABB8ESF-0SIT

      MT25QU512ABB8ESF-0SIT on 2 placements (IC19, IC21): the distributor catalogue reports "Obsolete" for Micron Technology Inc.'s MT25QU512ABB8ESF-0SIT — a sourcing/longevity risk to confirm against the MANUFACTURER before acting. That is the maker this board's BOM names (MICRON TECHNOLOGY). A distributor delisting is not the same claim as the part being unavailable; check the vendor's own lifecycle/product-change notice for the status and any named successor. Re-read the source: GET /api/odb/parts/MT25QU512ABB8ESF-0SIT/providers (which provider said it, the maker and part number it matched, its own words, and when we asked). THE LAST BUY DATE HAS PASSED (2026-08-31, 76 day(s) ago), so this is a substitution rather than a buy. The distributor reports NONE left. No approved alternate is recorded for this position.

    • TJ1973028 on 3 placements (TJ1, TJ2, XJ3): the distributor catalogue reports "Obsolete" for TE Connectivity Erni's 973028

      973028 on 3 placements (TJ1, TJ2, XJ3): the distributor catalogue reports "Obsolete" for TE Connectivity Erni's 973028 — a sourcing/longevity risk to confirm against the MANUFACTURER before acting. This board's BOM names ERNI, and that is one line with TE Connectivity Erni: Tyco Electronics renamed itself TE Connectivity in March 2011, and TE completed its acquisition of ERNI Group AG (ERNI Electronics) in September 2021. A distributor delisting is not the same claim as the part being unavailable; check the vendor's own lifecycle/product-change notice for the status and any named successor. Re-read the source: GET /api/odb/parts/973028/providers (which provider said it, the maker and part number it matched, its own words, and when we asked). THE LAST BUY DATE HAS PASSED (2022-11-30, 1,446 day(s) ago), so this is a substitution rather than a buy. The distributor reports NONE left. No approved alternate is recorded for this position.

    • XJ4214443 on XJ4: the distributor catalogue reports "Obsolete" for TE Connectivity Erni's 214443

      214443 on XJ4: the distributor catalogue reports "Obsolete" for TE Connectivity Erni's 214443 — a sourcing/longevity risk to confirm against the MANUFACTURER before acting. This board's BOM names ERNI, and that is one line with TE Connectivity Erni: Tyco Electronics renamed itself TE Connectivity in March 2011, and TE completed its acquisition of ERNI Group AG (ERNI Electronics) in September 2021. A distributor delisting is not the same claim as the part being unavailable; check the vendor's own lifecycle/product-change notice for the status and any named successor. Re-read the source: GET /api/odb/parts/214443/providers (which provider said it, the maker and part number it matched, its own words, and when we asked). THE LAST BUY DATE HAS PASSED (2022-11-30, 1,446 day(s) ago), so this is a substitution rather than a buy. The distributor reports NONE left. No approved alternate is recorded for this position.

  9. CrystalLoadCapsmajorOpen

    Q6 has no load capacitor on its terminal net(s) Net-(IC18-XA), Net-(IC18-XB)

    Q6 (crystal): no load capacitor found on its terminal net(s) Net-(IC18-XA), Net-(IC18-XB) — a passive crystal is specified for a load capacitance; verify the oscillator driver integrates the caps or add them. Q6's load capacitance is not on file — and the datasheet on file for CX3225SB48000D0FPJC1 does not name that exact ordering code, so it describes the family and a load capacitance read from it would be the family's range, not this part's. State it on the BOM row as parametrics.loadCapacitanceF (farads) from the part's own record, or submit parametric 'c_load' for CX3225SB48000D0FPJC1 with a cite. The oscillator load of IC18 is not on file: read the crystal-oscillator table of its datasheet and submit parametric 'c_osc_load' (as the datasheet states it, with a cite).

    At Q6

  10. ReferencePlaneDiscontinuitymajorOpen

    ReferencePlaneDiscontinuity: 7 boards×7

    Checked by the review, stands · Claude walker X1 (Cern session) — Checked: the 54 findings are tracks running along or across power-plane splits (P0V85/P1V8/P1V2/MGT_1V8/VCC_PSPLL on L8/L9) and reference changes at vias between GND and power planes. They are layout practice; no datasheet or interface specification on file states a requirement for these nets, and the nets' edges are only assumed (no IBIS on file), so their effect cannot be quantified. They stand for the owner.

    At along@L11_GND4:GND|P0V85, along@L8_PWR1:P0V85|P1V8, along@L8_PWR1:P1V2|P1V8, along@L9_PWR2:MGT_1V8|VCC_PSPLL, refchange:GND|P0V85, refchange:GND|P3V3, void:routed@BOTTOM_LAYER|L13_GND5

    Each finding, with its own headline
    • along@L11_GND4:GND|P0V852 nets run ALONG the split between the GND and P0V85 planes on L11_GND4 at 2 places, the longest 5.46 mm; the impedance rises from 44 ohm to 45 ohm there at worst, 1.7 % of each edge reflected (fastest: /PXIe-TIMING.DSTARB_P6 at 100 ps 10-90 %, ASSUMED)

      On L11_GND4 the GND and P0V85 planes part, and 2 nets run down the slot between them at 2 place(s), rather than across it. This is not a crossing: the return still follows the track, in the plane on its other side and in the slot's edges, so it does not detour. What changes is the track's own impedance. With the plane gone from under it on this side for that length, its field reaches further, to the slot's edges and to the next plane beyond, so its capacitance falls and its impedance rises; every edge travelling down it is partly reflected where it enters the stretch and again where it leaves, by (Z2 - Z1)/(Z2 + Z1). The longer the stretch against the edge's rise, the more of the edge sees the full step. A crossing is generally the worse fault; a mismatch is not free. Over solid plane each track is 44 ohm; along the slot 45 ohm, reflecting up to 1.7 % of each edge (a method-of-moments field solve of the cross-section, one dielectric; were the slot's edges to carry none of the return, up to 51 ohm). At 1 place(s) the impedance cannot be said: the track's centreline is not over the slot. At 1 of the 2 place(s) the rise is within 10 %, the tolerance controlled impedance is commonly built to (the stackup states none); the rest exceed it or cannot be said, so the finding is a warning. One decision covers the split line's slot: whether these tracks may run down it, or should be moved over solid plane.

      • '/PXIe-TIMING.DSTARB_P6' on L10 at (178.5, -108.62), 3.63 mm, running ALONG the split for 3.47 mm, not across it, impedance not said: the track's centreline is not over the slot; other side: P3V3 on L9_PWR2, 0.11 mm from the track against 0.1 mm to L11_GND4: 52% of the return on L11_GND4
      • '/WR_CLK.SPI_MOSI' on L10 at (162.97, -108), 5.69 mm, running ALONG the split for 5.46 mm, not across it, 44 ohm over solid plane, 45 ohm along it: 1.7 % of each edge reflected; other side: P3V3 on L9_PWR2, 0.11 mm from the track against 0.1 mm to L11_GND4: 52% of the return on L11_GND4

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=along%40L11_GND4%3AGND%7CP0V85 serves each with its own layer_image region; all of them lie in x 161.89..179.61, y -111.81..-104.19 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • along@L8_PWR1:P0V85|P1V83 nets run ALONG the split between the P0V85 and P1V8 planes on L8_PWR1 at 5 places, the longest 23.41 mm; the impedance rises from 44 ohm to 49 ohm there at worst, 5.8 % of each edge reflected (fastest: /WR_DAC.DIN at 100 ps 10-90 %, ASSUMED)

      On L8_PWR1 the P0V85 and P1V8 planes part, and 3 nets run down the slot between them at 5 place(s), rather than across it. This is not a crossing: the return still follows the track, in the plane on its other side and in the slot's edges, so it does not detour. What changes is the track's own impedance. With the plane gone from under it on this side for that length, its field reaches further, to the slot's edges and to the next plane beyond, so its capacitance falls and its impedance rises; every edge travelling down it is partly reflected where it enters the stretch and again where it leaves, by (Z2 - Z1)/(Z2 + Z1). The longer the stretch against the edge's rise, the more of the edge sees the full step. A crossing is generally the worse fault; a mismatch is not free. Over solid plane each track is 44 ohm; along the slot 45 ohm to 49 ohm, reflecting up to 5.8 % of each edge (a method-of-moments field solve of the cross-section, one dielectric; were the slot's edges to carry none of the return, up to 50 ohm). At 3 of the 5 place(s) the rise is within 10 %, the tolerance controlled impedance is commonly built to (the stackup states none); the rest exceed it or cannot be said, so the finding is a warning. One decision covers the split line's slot: whether these tracks may run down it, or should be moved over solid plane.

      • '/io_drivers_fp/TERM_EN4' on L7 at (156.08, -99.3), 1.49 mm, running ALONG the split for 1.02 mm, not across it, 44 ohm over solid plane, 49 ohm along it: 5.8 % of each edge reflected; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/WR_DAC.DIN' on L7 at (179.4, -99.28), 2.38 mm, running ALONG the split for 1.95 mm, not across it, 44 ohm over solid plane, 48 ohm along it: 5.1 % of each edge reflected; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/WR_DAC.SCLK' on L7 at (179.5, -126.35), 8.12 mm, running ALONG the split for 7.7 mm, not across it, 44 ohm over solid plane, 47 ohm along it: 3.4 % of each edge reflected; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/WR_DAC.DIN' on L7 at (167.74, -135.79), 23.41 mm, running ALONG the split for 23.41 mm, not across it, 44 ohm over solid plane, 45 ohm along it: 2.1 % of each edge reflected; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/WR_DAC.SCLK' on L7 at (167.71, -135.5), 23.65 mm, running ALONG the split for 23.32 mm, not across it, 44 ohm over solid plane, 45 ohm along it: 2.1 % of each edge reflected; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=along%40L8_PWR1%3AP0V85%7CP1V8 serves each with its own layer_image region; all of them lie in x 154.3..180.55, y -136.79..-97.15 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • along@L8_PWR1:P1V2|P1V82 nets run ALONG the split between the P1V2 and P1V8 planes on L8_PWR1 at 2 places, the longest 1.16 mm; the impedance rises from 42 ohm to 47 ohm there at worst, 5.1 % of each edge reflected (fastest: /DDR4-PS.DQ12 at 100 ps 10-90 %, ASSUMED)

      On L8_PWR1 the P1V2 and P1V8 planes part, and 2 nets run down the slot between them at 2 place(s), rather than across it. This is not a crossing: the return still follows the track, in the plane on its other side and in the slot's edges, so it does not detour. What changes is the track's own impedance. With the plane gone from under it on this side for that length, its field reaches further, to the slot's edges and to the next plane beyond, so its capacitance falls and its impedance rises; every edge travelling down it is partly reflected where it enters the stretch and again where it leaves, by (Z2 - Z1)/(Z2 + Z1). The longer the stretch against the edge's rise, the more of the edge sees the full step. A crossing is generally the worse fault; a mismatch is not free. Over solid plane each track is 42 ohm; along the slot 46 ohm to 47 ohm, reflecting up to 5.1 % of each edge (a method-of-moments field solve of the cross-section, one dielectric; were the slot's edges to carry none of the return, up to 48 ohm). At 1 of the 2 place(s) the rise is within 10 %, the tolerance controlled impedance is commonly built to (the stackup states none); the rest exceed it or cannot be said, so the finding is a warning. One decision covers the split line's slot: whether these tracks may run down it, or should be moved over solid plane.

      • '/DDR4-PS.DQ12' on L7 at (189.48, -82.84), 1.52 mm, running ALONG the split for 1.16 mm, not across it, 42 ohm over solid plane, 47 ohm along it: 5.1 % of each edge reflected; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/DDR4-PS.DQ13' on L7 at (185.4, -82.74), 1.3 mm, running ALONG the split for 0.9 mm, not across it, 42 ohm over solid plane, 46 ohm along it: 3.7 % of each edge reflected; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=along%40L8_PWR1%3AP1V2%7CP1V8 serves each with its own layer_image region; all of them lie in x 183.88..191.13, y -83.98..-81.68 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • along@L9_PWR2:MGT_1V8|VCC_PSPLL'/WR_DAC.LDAC' runs ALONG the split between the MGT_1V8 and VCC_PSPLL planes on L9_PWR2 at 1 place, the longest 0.88 mm; the impedance rises from 44 ohm to 49 ohm there at worst, 5.9 % of each edge reflected (fastest: /WR_DAC.LDAC at 100 ps 10-90 %, ASSUMED)

      On L9_PWR2 the MGT_1V8 and VCC_PSPLL planes part, and '/WR_DAC.LDAC' runs down the slot between them at 1 place(s), rather than across it. This is not a crossing: the return still follows the track, in the plane on its other side and in the slot's edges, so it does not detour. What changes is the track's own impedance. With the plane gone from under it on this side for that length, its field reaches further, to the slot's edges and to the next plane beyond, so its capacitance falls and its impedance rises; every edge travelling down it is partly reflected where it enters the stretch and again where it leaves, by (Z2 - Z1)/(Z2 + Z1). The longer the stretch against the edge's rise, the more of the edge sees the full step. A crossing is generally the worse fault; a mismatch is not free. Over solid plane each track is 44 ohm; along the slot 49 ohm, reflecting up to 5.9 % of each edge (a method-of-moments field solve of the cross-section, one dielectric; were the slot's edges to carry none of the return, up to 50 ohm). No place stays within 10 %, the tolerance controlled impedance is commonly built to (the stackup states none), so the finding is a warning. One decision covers the split line's slot: whether these tracks may run down it, or should be moved over solid plane.

      • '/WR_DAC.LDAC' on L10 at (175.89, -108.93), 1.36 mm, running ALONG the split for 0.88 mm, not across it, 44 ohm over solid plane, 49 ohm along it: 5.9 % of each edge reflected; other side: P0V85 on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=along%40L9_PWR2%3AMGT_1V8%7CVCC_PSPLL serves each with its own layer_image region; all of them lie in x 174.49..177.28, y -110.56..-107.53 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • refchange:GND|P0V8525 nets change reference between GND and P0V85 at 27 vias; 22 of them partial (fastest: /DDR4-PS.BA1 at 100 ps 10-90 %, ASSUMED)

      At 27 via(s), on 25 net(s), the signal leaves a layer referenced by one of GND and P0V85 for a layer referenced by the other, and no plane references both layers. Different nets: no via can carry the return across; only decoupling between the two planes near the via can. The capacitor carrying the return across was measured at each via, at the knee of the crossing net's edge against 50 ohm assumed: 27 with none. 27 of them at an ASSUMED edge (no edge on file for the net): state one (net_timing_state) and it is re-read. One decision covers the pair: whether GND and P0V85 are decoupled to each other near these vias. A via whose layers have more references than these two is a site of each pair it joins. 22 of them are partial: another reference continues across through a stitching via of its net, so only part of the return has nowhere to go. The pure ones are listed first, then the partial ones with the farthest stitching via first.

      • '/WR_DAC.DIN' at (177.8, -100.9), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P0V85 on L8_PWR1; partial, GND continues through a via of GND 0.8 mm away; stitching none, 146.6 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P0V85 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/WR_DAC.SCLK' at (177.8, -100.1), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P0V85 on L8_PWR1; partial, GND continues through a via of GND 1.13 mm away; stitching none, 144.9 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (1 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P0V85 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/WR_CLK.SPI_MISO' at (163.4, -104.1), TOP_LAYER -> L10: GND on L2_GND1 to P3V3 on L9_PWR2, P0V85 on L11_GND4; pure, no reference continues across; stitching none, 90.3 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (1 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P0V85 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_P7' at (177.8, -104.9), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P0V85 on L8_PWR1; partial, GND continues through a via of GND 0.8 mm away; stitching none, 83.9 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P0V85 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_N7' at (177.8, -105.7), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P0V85 on L8_PWR1; partial, GND continues through a via of GND 0.8 mm away; stitching none, 83.5 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P0V85 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_P6' at (178.6, -104.9), TOP_LAYER -> L10: GND on L2_GND1 to P3V3 on L9_PWR2, P0V85 on L11_GND4; pure, no reference continues across; stitching none, 83.4 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P0V85 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_N15' at (178.6, -106.5), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P0V85 on L8_PWR1; partial, GND continues through a via of GND 1.13 mm away; stitching none, 81.5 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P0V85 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_P15' at (177.8, -106.5), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P0V85 on L8_PWR1; partial, GND continues through a via of GND 0.8 mm away; stitching none, 81.5 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P0V85 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/WR_CLK.MAIN_P' at (162.6, -102.5), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P0V85 on L8_PWR1; partial, GND continues through a via of GND 0.8 mm away; stitching none, 74.4 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (1 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P0V85 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/WR_CLK.MAIN_N' at (162.6, -103.3), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P0V85 on L8_PWR1; partial, GND continues through a via of GND 1.13 mm away; stitching none, 73.8 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (1 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P0V85 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/io_drivers_fp/TERM_EN0' at (160.23, -100.09), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P0V85 on L8_PWR1; partial, GND continues through a via of GND 1.57 mm away; stitching none, 64.2 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (20 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P0V85 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/io_drivers_fp/F_FP_OE0' at (161.01, -100.09), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P0V85 on L8_PWR1; partial, GND continues through a via of GND 0.79 mm away; stitching none, 61.8 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P0V85 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • and 15 more

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=refchange%3AGND%7CP0V85 serves each with its own layer_image region; all of them lie in x 158.2..181.15, y -113.9..-87.7 mm (board frame, Y up).

      The reference on each layer is the first plane the track meets leaving the via, anti-pads and the net's own copper skipped. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • refchange:GND|P3V336 nets change reference between GND and P3V3 at 41 vias; 36 of them partial (fastest: /DDR4-PS.A0 at 100 ps 10-90 %, ASSUMED)

      At 41 via(s), on 36 net(s), the signal leaves a layer referenced by one of GND and P3V3 for a layer referenced by the other, and no plane references both layers. Different nets: no via can carry the return across; only decoupling between the two planes near the via can. The capacitor carrying the return across was measured at each via, at the knee of the crossing net's edge against 50 ohm assumed: 41 with none. 41 of them at an ASSUMED edge (no edge on file for the net): state one (net_timing_state) and it is re-read. One decision covers the pair: whether GND and P3V3 are decoupled to each other near these vias. A via whose layers have more references than these two is a site of each pair it joins. 36 of them are partial: another reference continues across through a stitching via of its net, so only part of the return has nowhere to go. The pure ones are listed first, then the partial ones with the farthest stitching via first.

      • '/PXIe-MGT.RX_N1' at (162.61, -84.91), TOP_LAYER -> L10: GND on L2_GND1 to P3V3 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.79 mm away; stitching none, 109.0 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (3 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-MGT.RX_P1' at (162.61, -85.71), TOP_LAYER -> L10: GND on L2_GND1 to P3V3 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.79 mm away; stitching none, 109.0 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (3 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-MGT.TX_N1' at (217.1, -87.55), TOP_LAYER -> L10: GND on L2_GND1 to P3V3 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 2.01 mm away; stitching none, 96.3 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (3 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-MGT.TX_P1' at (217.1, -88.05), TOP_LAYER -> L10: GND on L2_GND1 to P3V3 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 1.51 mm away; stitching none, 96.3 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (3 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/WR_CLK.SPI_MISO' at (163.4, -104.1), TOP_LAYER -> L10: GND on L2_GND1 to P3V3 on L9_PWR2, P0V85 on L11_GND4; pure, no reference continues across; stitching none, 90.3 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (1 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/WR_CLK.SPI_MOSI' at (162.6, -104.9), TOP_LAYER -> L10: GND on L2_GND1 to P3V3 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 1.13 mm away; stitching none, 86.9 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (1 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/WR_CLK.SPI_SCK' at (161.8, -104.9), TOP_LAYER -> L10: GND on L2_GND1 to P3V3 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.8 mm away; stitching none, 86.4 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (1 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_P6' at (178.6, -104.9), TOP_LAYER -> L10: GND on L2_GND1 to P3V3 on L9_PWR2, P0V85 on L11_GND4; pure, no reference continues across; stitching none, 83.4 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_N6' at (178.6, -105.7), TOP_LAYER -> L10: GND on L2_GND1 to P3V3 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.8 mm away; stitching none, 83.4 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_P13' at (180.2, -105.7), TOP_LAYER -> L10: GND on L2_GND1 to P3V3 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.8 mm away; stitching none, 80.8 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_N13' at (181, -105.7), TOP_LAYER -> L10: GND on L2_GND1 to P3V3 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.8 mm away; stitching none, 80.8 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_P14' at (179.4, -104.1), TOP_LAYER -> L10: GND on L2_GND1 to P3V3 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 1.13 mm away; stitching none, 75.0 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • and 29 more

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=refchange%3AGND%7CP3V3 serves each with its own layer_image region; all of them lie in x 110.3..219.1, y -135.25..-77.95 mm (board frame, Y up).

      The reference on each layer is the first plane the track meets leaving the via, anti-pads and the net's own copper skipped. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • void:routed@BOTTOM_LAYER|L13_GND5Tracks on BOTTOM_LAYER and L13_GND5 cross each other at 1 place, each layer the other's reference with its pour cleared round its own tracks, the longest 0.51 mm (fastest: /io_drivers_pp/PP_OUT2 at 100 ps 10-90 %, ASSUMED)

      BOTTOM_LAYER and L13_GND5 each carry a pour and routed tracks, and each is the other's nearest reference. Wherever a track on one crosses a track on the other, the pour under it is cleared round that other track, so for the width of the cut the return has no plane under it and detours round the cut. 1 place(s) on 2 net(s), 1 of them seen from both tracks (each counted once). One decision covers the layer pair: whether these layers may take a pour their own tracks cut as the reference, or the routing or the stackup should give the tracks a plane that is not cut where they cross. No crossing has a plane on the track's other side, so the whole return meets the break.

      • one crossing of two tracks, seen from each: '/io_drivers_pp/PP_OUT2' on BOTTOM_LAYER at (100.37, -125.06), 0.49 mm over copper of /io_drivers_pp/bank0/OUT3, not a plane; other side: none: the signal layer is outermost / '/io_drivers_pp/bank0/OUT3' on L13_GND5 at (100.37, -125.06), 0.51 mm over copper of /io_drivers_pp/PP_OUT2, not a plane; other side: no plane on L12

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=void%3Arouted%40BOTTOM_LAYER%7CL13_GND5 serves each with its own layer_image region; all of them lie in x 99.2..101.55, y -126.31..-123.8 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

  11. SinglePointNetmajorOpen

    SinglePointNet: 17 pins×17

    e.g. “IC21.3 (RESET#) is an IC input left unconnected” · 16 different statements

    At IC18.1, IC18.10, IC18.11, IC18.12, IC18.2, IC18.41 and 11 more

    Each finding, with its own headline
    • IC18.1IC18.1 (IN1) is an IC input left unconnected

      Pin IC18.1 (IN1) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC18-IN1-Pad1)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

    • IC18.10IC18.10 (IN2) is an IC input left unconnected

      Pin IC18.10 (IN2) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC18-IN2-Pad10)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

    • IC18.11IC18.11 (IN2b) is an IC input left unconnected

      Pin IC18.11 (IN2b) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC18-~{IN2}-Pad11)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

    • IC18.12IC18.12 (OEb) is an IC input left unconnected

      Pin IC18.12 (OEb) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC18-~{OE}-Pad12)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

    • IC18.2IC18.2 (IN1b) is an IC input left unconnected

      Pin IC18.2 (IN1b) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC18-~{IN1}-Pad2)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

    • IC18.41IC18.41 (FB_IN) is an IC input left unconnected

      Pin IC18.41 (FB_IN) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC18-FB_IN-Pad41)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

    • IC18.42IC18.42 (FB_INb) is an IC input left unconnected

      Pin IC18.42 (FB_INb) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC18-~{FB_IN}-Pad42)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

    • IC19.3IC19.3 (RESET#) is an IC input left unconnected

      Pin IC19.3 (RESET#) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC19-~{RESET}-Pad3)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

    • IC21.3IC21.3 (RESET#) is an IC input left unconnected

      Pin IC21.3 (RESET#) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC21-~{RESET}-Pad3)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

    • IC3.1IC3.1 (RI_SCI) is an IC input left unconnected

      Pin IC3.1 (RI_SCI) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC3-~{SUSPEND}{slash}RI_SCI-Pad1)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

    • IC3.10IC3.10 (CTS_ECI) is an IC input left unconnected

      Pin IC3.10 (CTS_ECI) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC3-CTS_ECI-Pad10)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

    • IC3.14IC3.14 (DSR_ECI) is an IC input left unconnected

      Pin IC3.14 (DSR_ECI) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC3-GPIO.1_ECI{slash}DSR_ECI-Pad14)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

    • IC3.16IC3.16 (DCD_ECI) is an IC input left unconnected

      Pin IC3.16 (DCD_ECI) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC3-NC{slash}VPP{slash}DCD_ECI-Pad16)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

    • IC3.17IC3.17 (RI_ECI) is an IC input left unconnected

      Pin IC3.17 (RI_ECI) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC3-~{SUSPEND}{slash}RI_ECI-Pad17)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

    • IC3.18IC3.18 (CTS_SCI) is an IC input left unconnected

      Pin IC3.18 (CTS_SCI) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC3-CTS_SCI-Pad18)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

    • IC3.22IC3.22 (DSR_SCI) is an IC input left unconnected

      Pin IC3.22 (DSR_SCI) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC3-GPIO.2_SCI{slash}DSR_SCI-Pad22)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

    • IC3.24IC3.24 (DCD_SCI) is an IC input left unconnected

      Pin IC3.24 (DCD_SCI) is an IC INPUT the designer flagged no-connect (KiCad net 'unconnected-(IC3-GPIO.0_SCI{slash}DCD_SCI-Pad24)'), but an input normally needs a defined level - confirm it is truly meant to float. If the datasheet defines this input left open (an internal pull, "read as low when left floating", or a tri-level input whose open state is one of its settings), file it on the pin as if_unused (pull_up, pull_down or may_float) with submit_pin_behavior, citing the page, and this finding clears.

  12. observationOBS-652minorOpen

    ESD suppressors D4, D6, D7, D8 sit away from their connector contacts, against the 0402ESDA-MLP placement guidance

    D4 and D6 (JTAG TCK/TDI on J7) are reached through a via and several millimetres of line, and D7/D8 (PMBus SDA/SCL on J3) hang off a 1.8 mm stub: the Eaton datasheet asks for the suppressor at the signal input, ahead of any other component, on a 0-stub pad.

    What the copper shows

    ProtectorLineContactRouted length to protectorVias before itTopologySource
    D4JTAG_TCKJ7.69.33 mm (joins 8.18 mm along an 8.18 mm run to R52.1)1stub, 1.15 mmProbe esdPlacement, copper read
    D6JTAG_TDIJ7.106.59 mm along a 7.99 mm run to R54.11flow-throughProbe esdPlacement, copper read
    D7I2C.SDAJ3.21.84 mm0stub, 1.84 mmProbe esdPlacement, copper read
    D8I2C.SCLJ3.31.81 mm0stub, 1.81 mmProbe esdPlacement, copper read

    What the part's maker asks for

    0402ESDA-MLP datasheet (Eaton Technical Data 4367), Design considerations, p2: the device should be placed as close to the signal input as possible and ahead of any other component, with a "0-stub" pad directly on the signal line.

    ``mermaid graph LR J7_6[J7.6 TCK contact] --> V[via] --> L[8 mm line] --> D4[D4 clamp] --> R52[R52 pull-up / IC14.R19] ``

    Why it matters

    An ESD strike on the JTAG header reaches an inner layer through the via and travels most of the line before D4/D6 clamp it; on D7/D8 the stub inductance adds to the clamp voltage the FPGA I2C pins and the two IRPS5401 PMBus pins see. Moving each suppressor onto the line at the contact (before the via) follows the maker's guidance. Severity minor: these are service headers (debug JTAG, PMBus), not field ports.

  13. observationOBS-653minorOpen

    F1 (750 mA) on PP_VDD can never open: its source 5VREG is a 300 mA buck that current-limits first and takes the whole board down

    F1 (0466.750NR, 750 mA) protects PP_VDD on the pp connector J6, but it is fed from 5VREG, which IC32 (TPS62125) limits to about 300 mA: a short on PP_VDD current-limits IC32 instead of opening F1, and the collapse of 5VREG drops P5V_REG_PG and the VCC of both IRPS5401s, so every rail on the board goes down.

    The fuse cannot reach its opening current

    QuantityValueSource
    F1 rating0.75 A0466.750NR datasheet p1, Electrical Specifications
    F1 opening time at 100 % / 200 % of rating4 h minimum / 5 s maximum0466.750NR datasheet p1
    IC32 output current capability (VIN 6-17 V)300 mATPS62125 datasheet p4, Recommended Operating Conditions
    IC32 output current in short circuit0.5 x I_LIMF, typ. 300 mATPS62125 datasheet p12, 7.4.5
    IC32 high-side switch limit I_LIMF600 / 750 / 900 mA (min/typ/max)TPS62125 datasheet p5

    A fault current of about 300 mA is 40 % of F1's rating, where the fuse is specified to carry current for at least 4 hours: F1 does not open.

    What a short on the connector does instead

    ``mermaid graph LR P12V --> IC32[IC32 TPS62125, 300 mA] --> V5[5VREG] V5 --> F1[F1 750 mA] --> PPVDD[PP_VDD, J6.38/39] V5 --> VCC[IC30/IC31 VCC, VDRV; IC29 VCC; LD1..LD13; IC22 BIAS] IC32 -- PG --> PG[P5V_REG_PG] --> EN[IC30 EN_A..L, IC29 EN] ``

    5VREG is the VCC/VDRV supply of IC30 and IC31, the bias of IC22, and IC32's PG pin drives P5V_REG_PG, which enables every IC30 channel and IC29. A short or overload on the external pp connector therefore browns out the whole board rather than isolating the connector.

    Question for the designer

    Is F1 meant to protect against something other than a hard short (it cannot clear one from this source), and is the PP_VDD load meant to be shared with the board's own 5VREG loads within IC32's 300 mA? A separate current-limited switch for PP_VDD, or a fuse sized below IC32's limit, would keep a connector fault off the board's supply.

  14. observationOBS-656minorOpen

    IC12 (REF5030, 3.0 V) has no input headroom at the bottom of the P3V3 window the designer specifies

    IC12 (REF5030AID, 3.0 V reference for the White Rabbit DAC IC10) needs VIN of at least VOUT + 0.2 V = 3.2 V, and its input P3V3_CLK is P3V3 behind two ferrite beads, whose own specified minimum is 3.2 V: at the low end of that window the reference has zero headroom before the bead drop, and drops out of regulation.

    The numbers

    QuantityValueSource
    REF5030 output3.0 VREF50xx datasheet, Table 4-1, p3
    Minimum input voltageVOUT + 0.2 V = 3.2 VREF50xx datasheet, Recommended Operating Conditions, p5
    P3V3 window as configured3.3 V nom, 3.2 V min, 3.4 V maxIC31 'Infineon Configuration 6' table, schematic sheet 44
    Path P3V3 to IC12.2L13 then L14 (BLM21PG221SN1 ferrite beads)netlist: P3V3, Net-(C15-Pad2), P3V3_CLK
    Headroom at P3V3 = 3.2 V3.2 - 3.2 = 0 V, minus the two beads' IR dropderived

    ``mermaid graph LR P3V3[P3V3, min 3.2 V] --> L13 --> L14 --> CLK[P3V3_CLK] --> IC12[IC12 REF5030, needs >= 3.2 V] --> VREF[IC10 DAC8562 VREF] --> DAC[VOUTB -> OCXO OSC6 tuning] ``

    Why it matters

    A reference in dropout does not fail visibly; its 3.0 V output follows the input down, which shifts the DAC full scale that steers the OCXO. At the nominal 3.3 V there is 0.1 V of margin, so this only bites at the low end of the stated window, which is why it is minor. Feeding IC12 from a rail with more headroom (for example 5VREG, within REF5030's 18 V maximum input) or choosing a 2.5 V reference would remove it.

  15. DecapViaFanoutminorOpen

    DecapViaFanout: 3 components×3

    Checked by the review, stands · Claude walker X1 (Cern session) — Checked: C208 (P1V2), C191 and C192 (P3V3PXIE, 10 uF bulk at IC32/IC24 inputs) have 1.8-3.2 mm runs to their rail vias (0.5-1.4 nH). The 0.5 nH limit is layout practice; no datasheet on file states a fan-out limit for these parts, and for 10 uF bulk input capacitors the extra nH matters little. Stands for the owner.

    At C191, C192, C208

    Each finding, with its own headline
    • C191C191 on 'P3V3PXIE': its rail pad reaches its via over 1.81 mm of track, adding about 0.51 nH to its loop, past Probe's 0.5 nH starting value; put the vias at the pads (a 10 uF bulk capacitor: its inductance matters only far above the band it serves)

      C191 on 'P3V3PXIE': its rail pad reaches its via over 1.81 mm of track, adding about 0.51 nH to its loop, past Probe's 0.5 nH starting value; put the vias at the pads (a 10 uF bulk capacitor: its inductance matters only far above the band it serves). C191.1 (rail) runs 1.81 mm along its copper to its nearest via, 0.81 mm past the 1 mm beside the pad: 0.51 nH; C191.2 (ground) not measured: the pad sits in a pour of its net on its own layer, which has no track length to measure. Each run past the pad as one conductor of a pair 3.00 mm apart (the pad pitch), 0.2 l ln(D/R) nH with R 0.127 mm (the track's width is not read): 0.51 nH in all. Put the vias at the pads (or in them), one per pad at least, the rail's and the ground's close together; the capacitor's own inductance as mounted is 1.38 nH (pad 0.20 nH + no via of the first net within 1 mm of its pad: the study's 0.6 nH per mm of 0.63 mm depth, ASSUMED). 0.5 nH is Probe's starting value, about the whole mounted inductance of a well-mounted 0402.

    • C192C192 on 'P3V3PXIE': its rail pad reaches its via over 3.24 mm of track, adding about 1.42 nH to its loop, past Probe's 0.5 nH starting value; put the vias at the pads (a 10 uF bulk capacitor: its inductance matters only far above the band it serves)

      C192 on 'P3V3PXIE': its rail pad reaches its via over 3.24 mm of track, adding about 1.42 nH to its loop, past Probe's 0.5 nH starting value; put the vias at the pads (a 10 uF bulk capacitor: its inductance matters only far above the band it serves). C192.1 (rail) runs 3.24 mm along its copper to its nearest via, 2.24 mm past the 1 mm beside the pad: 1.42 nH; C192.2 (ground) not measured: the pad sits in a pour of its net on its own layer, which has no track length to measure. Each run past the pad as one conductor of a pair 3.00 mm apart (the pad pitch), 0.2 l ln(D/R) nH with R 0.127 mm (the track's width is not read): 1.42 nH in all. Put the vias at the pads (or in them), one per pad at least, the rail's and the ground's close together; the capacitor's own inductance as mounted is 1.38 nH (pad 0.20 nH + no via of the first net within 1 mm of its pad: the study's 0.6 nH per mm of 0.63 mm depth, ASSUMED). 0.5 nH is Probe's starting value, about the whole mounted inductance of a well-mounted 0402.

    • C208C208 on 'P1V2': its rail pad reaches its via over 1.96 mm of track and its ground pad over 0.00 mm, adding about 0.61 nH to its loop, past Probe's 0.5 nH starting value; put the vias at the pads (a 10 uF bulk capacitor: its inductance matters only far above the band it serves)

      C208 on 'P1V2': its rail pad reaches its via over 1.96 mm of track and its ground pad over 0.00 mm, adding about 0.61 nH to its loop, past Probe's 0.5 nH starting value; put the vias at the pads (a 10 uF bulk capacitor: its inductance matters only far above the band it serves). C208.1 (rail) runs 1.96 mm along its copper to its nearest via, 0.96 mm past the 1 mm beside the pad: 0.61 nH; C208.2 (ground) runs 0.00 mm along its copper to its nearest via, beside the pad. Each run past the pad as one conductor of a pair 3.00 mm apart (the pad pitch), 0.2 l ln(D/R) nH with R 0.127 mm (the track's width is not read): 0.61 nH in all. Put the vias at the pads (or in them), one per pad at least, the rail's and the ground's close together; the capacitor's own inductance as mounted is 1.50 nH (pad 0.20 nH + no via of the first net within 1 mm of its pad: the study's 0.6 nH per mm of 0.84 mm depth, ASSUMED). 0.5 nH is Probe's starting value, about the whole mounted inductance of a well-mounted 0402.

  16. ReferencePlaneDiscontinuityminorOpen

    ReferencePlaneDiscontinuity: 17 boards×17

    Checked by the review, stands · Claude walker X1 (Cern session) — Checked: the 54 findings are tracks running along or across power-plane splits (P0V85/P1V8/P1V2/MGT_1V8/VCC_PSPLL on L8/L9) and reference changes at vias between GND and power planes. They are layout practice; no datasheet or interface specification on file states a requirement for these nets, and the nets' edges are only assumed (no IBIS on file), so their effect cannot be quantified. They stand for the owner.

    At along@L8_PWR1:P12V|P1V8, along@L9_PWR2:MGT_1V8|P3V3, along@L9_PWR2:P2V5|P3V3, refchange:GND|MGT_1V8, refchange:GND|P12V, refchange:GND|P1V2 and 11 more

    Each finding, with its own headline
    • along@L8_PWR1:P12V|P1V8'/PXIe-TIMING.DSTARB_N0' runs ALONG the split between the P12V and P1V8 planes on L8_PWR1 at 1 place, the longest 0.78 mm; the impedance rises from 54 ohm to 59 ohm there at worst, 4.3 % of each edge reflected (fastest: /PXIe-TIMING.DSTARB_N0 at 100 ps 10-90 %, ASSUMED)

      On L8_PWR1 the P12V and P1V8 planes part, and '/PXIe-TIMING.DSTARB_N0' runs down the slot between them at 1 place(s), rather than across it. This is not a crossing: the return still follows the track, in the plane on its other side and in the slot's edges, so it does not detour. What changes is the track's own impedance. With the plane gone from under it on this side for that length, its field reaches further, to the slot's edges and to the next plane beyond, so its capacitance falls and its impedance rises; every edge travelling down it is partly reflected where it enters the stretch and again where it leaves, by (Z2 - Z1)/(Z2 + Z1). The longer the stretch against the edge's rise, the more of the edge sees the full step. A crossing is generally the worse fault; a mismatch is not free. Over solid plane each track is 54 ohm; along the slot 59 ohm, reflecting up to 4.3 % of each edge (a method-of-moments field solve of the cross-section, one dielectric; were the slot's edges to carry none of the return, up to 60 ohm). At every place the rise is within 10 %, the tolerance controlled impedance is commonly built to (the stackup states none), so the finding is the lower tier (info). One decision covers the split line's slot: whether these tracks may run down it, or should be moved over solid plane.

      • '/PXIe-TIMING.DSTARB_N0' on L7 at (219.07, -106.06), 1.16 mm, running ALONG the split for 0.78 mm, not across it, 54 ohm over solid plane, 59 ohm along it: 4.3 % of each edge reflected; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=along%40L8_PWR1%3AP12V%7CP1V8 serves each with its own layer_image region; all of them lie in x 217.9..220.2, y -107.58..-104.6 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • along@L9_PWR2:MGT_1V8|P3V3'/PXIe-MGT.TX_N1' runs ALONG the split between the MGT_1V8 and P3V3 planes on L9_PWR2 at 1 place, the longest 0.12 mm; the impedance rises from 54 ohm to 55 ohm there at worst, 1.1 % of each edge reflected (fastest: /PXIe-MGT.TX_N1 at 3 ns 10-90 %, suggested by its name)

      On L9_PWR2 the MGT_1V8 and P3V3 planes part, and '/PXIe-MGT.TX_N1' runs down the slot between them at 1 place(s), rather than across it. This is not a crossing: the return still follows the track, in the plane on its other side and in the slot's edges, so it does not detour. What changes is the track's own impedance. With the plane gone from under it on this side for that length, its field reaches further, to the slot's edges and to the next plane beyond, so its capacitance falls and its impedance rises; every edge travelling down it is partly reflected where it enters the stretch and again where it leaves, by (Z2 - Z1)/(Z2 + Z1). The longer the stretch against the edge's rise, the more of the edge sees the full step. A crossing is generally the worse fault; a mismatch is not free. Over solid plane each track is 54 ohm; along the slot 55 ohm, reflecting up to 1.1 % of each edge (a method-of-moments field solve of the cross-section, one dielectric; were the slot's edges to carry none of the return, up to 60 ohm). At every place the rise is within 10 %, the tolerance controlled impedance is commonly built to (the stackup states none), so the finding is the lower tier (info). One decision covers the split line's slot: whether these tracks may run down it, or should be moved over solid plane.

      • '/PXIe-MGT.TX_N1' on L10 at (164.59, -87.01), 0.14 mm, running ALONG the split for 0.12 mm, not across it, 54 ohm over solid plane, 55 ohm along it: 1.1 % of each edge reflected; other side: GND on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=along%40L9_PWR2%3AMGT_1V8%7CP3V3 serves each with its own layer_image region; all of them lie in x 163.59..165.61, y -88.07..-85.94 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • along@L9_PWR2:P2V5|P3V3'/DDR4-PS.DQ20' runs ALONG the split between the P2V5 and P3V3 planes on L9_PWR2 at 1 place, the longest 0.2 mm; the impedance rises from 42 ohm to 44 ohm there at worst, 1.3 % of each edge reflected (fastest: /DDR4-PS.DQ20 at 100 ps 10-90 %, ASSUMED)

      On L9_PWR2 the P2V5 and P3V3 planes part, and '/DDR4-PS.DQ20' runs down the slot between them at 1 place(s), rather than across it. This is not a crossing: the return still follows the track, in the plane on its other side and in the slot's edges, so it does not detour. What changes is the track's own impedance. With the plane gone from under it on this side for that length, its field reaches further, to the slot's edges and to the next plane beyond, so its capacitance falls and its impedance rises; every edge travelling down it is partly reflected where it enters the stretch and again where it leaves, by (Z2 - Z1)/(Z2 + Z1). The longer the stretch against the edge's rise, the more of the edge sees the full step. A crossing is generally the worse fault; a mismatch is not free. Over solid plane each track is 42 ohm; along the slot 44 ohm, reflecting up to 1.3 % of each edge (a method-of-moments field solve of the cross-section, one dielectric; were the slot's edges to carry none of the return, up to 48 ohm). At every place the rise is within 10 %, the tolerance controlled impedance is commonly built to (the stackup states none), so the finding is the lower tier (info). One decision covers the split line's slot: whether these tracks may run down it, or should be moved over solid plane.

      • '/DDR4-PS.DQ20' on L10 at (188.33, -80.26), 0.36 mm, running ALONG the split for 0.2 mm, not across it, 42 ohm over solid plane, 44 ohm along it: 1.3 % of each edge reflected; other side: GND on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=along%40L9_PWR2%3AP2V5%7CP3V3 serves each with its own layer_image region; all of them lie in x 187.15..189.46, y -81.29..-79.18 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • refchange:GND|MGT_1V83 nets change reference between GND and MGT_1V8 at 3 vias, each with another reference that continues across (fastest: /WR_DAC.LDAC at 100 ps 10-90 %, ASSUMED)

      At 3 via(s), on 3 net(s), the signal leaves a layer referenced by one of GND and MGT_1V8 for a layer referenced by the other, and no plane references both layers. Different nets: no via can carry the return across; only decoupling between the two planes near the via can. The capacitor carrying the return across was measured at each via, at the knee of the crossing net's edge against 50 ohm assumed: 3 with none. 3 of them at an ASSUMED edge (no edge on file for the net): state one (net_timing_state) and it is re-read. One decision covers the pair: whether GND and MGT_1V8 are decoupled to each other near these vias. A via whose layers have more references than these two is a site of each pair it joins. Every one is partial: another reference continues across through a stitching via of its net, so only part of the return has nowhere to go, and the finding is the lower tier (info). They are listed with the farthest stitching via first.

      • '/PXIe-MGT.TX_N1' at (162.61, -88.11), TOP_LAYER -> L10: GND on L2_GND1 to MGT_1V8 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.79 mm away; stitching none, 96.3 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (3 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and MGT_1V8 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. Measured for reference: C342 (4.7 uF 0402) 1.3 mm away: 0.69 nH in the 0.730 mm cavity + 0.30 nH ESL + 1.12 nH mounting (pad 0.07 nH (vias beside the pads) + 1 via pair(s) at 0.80 mm pitch, 0.20 mm drill, 1.26 mm to the farther plane): 66.1 ohm at the knee, 132 % of 50 ohm (assumed track impedance). None.
      • '/PXIe-MGT.TX_P1' at (162.61, -88.91), TOP_LAYER -> L10: GND on L2_GND1 to MGT_1V8 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 1.13 mm away; stitching none, 96.3 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (3 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and MGT_1V8 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. Measured for reference: C25 (4.7 uF 0402) 1.7 mm away: 0.76 nH in the 0.730 mm cavity + 0.30 nH ESL + 1.12 nH mounting (pad 0.07 nH (vias beside the pads) + 1 via pair(s) at 0.80 mm pitch, 0.20 mm drill, 1.26 mm to the farther plane): 68.3 ohm at the knee, 137 % of 50 ohm (assumed track impedance). None.
      • '/WR_DAC.LDAC' at (171.95, -136.25), L7 -> L10: GND on L6_GND3, P1V8 on L8_PWR1 to MGT_1V8 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 1.77 mm away; stitching none, 44.5 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and MGT_1V8 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=refchange%3AGND%7CMGT_1V8 serves each with its own layer_image region; all of them lie in x 160.61..173.95, y -138.25..-86.11 mm (board frame, Y up).

      The reference on each layer is the first plane the track meets leaving the via, anti-pads and the net's own copper skipped. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • refchange:GND|P12V2 nets change reference between GND and P12V at 2 vias, each with another reference that continues across (fastest: /power-supply-1/P1V8AUX_PG at 20 ns 10-90 %, suggested by its name)

      At 2 via(s), on 2 net(s), the signal leaves a layer referenced by one of GND and P12V for a layer referenced by the other, and no plane references both layers. Different nets: no via can carry the return across; only decoupling between the two planes near the via can. The capacitor carrying the return across was measured at each via, at the knee of the crossing net's edge against 50 ohm assumed: 2 with none. 2 of them at an ASSUMED edge (no edge on file for the net): state one (net_timing_state) and it is re-read. One decision covers the pair: whether GND and P12V are decoupled to each other near these vias. A via whose layers have more references than these two is a site of each pair it joins. Every one is partial: another reference continues across through a stitching via of its net, so only part of the return has nowhere to go, and the finding is the lower tier (info). They are listed with the farthest stitching via first.

      • '/power-supply-1/P5V_REG_PG' at (152.78, -149.76), L7 -> BOTTOM_LAYER: GND on L6_GND3, P12V on L8_PWR1 to GND on L13_GND5; partial, GND continues through a via of GND 3.45 mm away; stitching none, 4.9 mm of track past the via on BOTTOM_LAYER: At the 5 GHz knee of a 100 ps edge (20 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P12V are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/power-supply-1/P1V8AUX_PG' at (158.12, -71.22), L3 -> L7: GND on L2_GND1, GND on L4_GND2 to GND on L6_GND3, P12V on L8_PWR1; partial, GND continues through a via of GND 2.37 mm away; stitching none, 3.0 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (20 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P12V are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=refchange%3AGND%7CP12V serves each with its own layer_image region; all of them lie in x 150.78..160.12, y -151.76..-69.22 mm (board frame, Y up).

      The reference on each layer is the first plane the track meets leaving the via, anti-pads and the net's own copper skipped. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • refchange:GND|P1V221 nets change reference between GND and P1V2 at 39 vias, each with another reference that continues across (fastest: /DDR4-PS.A2 at 100 ps 10-90 %, ASSUMED)

      At 39 via(s), on 21 net(s), the signal leaves a layer referenced by one of GND and P1V2 for a layer referenced by the other, and no plane references both layers. Different nets: no via can carry the return across; only decoupling between the two planes near the via can. The capacitor carrying the return across was measured at each via, at the knee of the crossing net's edge against 50 ohm assumed: 39 with none. 39 of them at an ASSUMED edge (no edge on file for the net): state one (net_timing_state) and it is re-read. One decision covers the pair: whether GND and P1V2 are decoupled to each other near these vias. A via whose layers have more references than these two is a site of each pair it joins. Every one is partial: another reference continues across through a stitching via of its net, so only part of the return has nowhere to go, and the finding is the lower tier (info). They are listed with the farthest stitching via first.

      • '/DDR4-PS.A3' at (194.97, -74.83), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V2 on L8_PWR1; partial, GND continues through a via of GND 1.6 mm away; stitching none, 35.6 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V2 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.A3' at (177, -84.9), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V2 on L8_PWR1; partial, GND continues through a via of GND 1.13 mm away; stitching none, 35.6 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V2 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. Measured for reference: C90 (4.7 uF 0402) 2.5 mm away: 0.75 nH in the 0.630 mm cavity + 0.30 nH ESL + 0.63 nH mounting (pad 0.07 nH (vias beside the pads) + 1 via pair(s) at 0.99 mm pitch, 0.30 mm drill, 0.74 mm to the farther plane): 52.8 ohm at the knee, 106 % of 50 ohm (assumed track impedance). None.
      • '/DDR4-PS.A2' at (177.8, -84.9), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V2 on L8_PWR1; partial, GND continues through a via of GND 0.8 mm away; stitching none, 35.3 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V2 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.A2' at (196.57, -77.23), L3 -> L7: GND on L2_GND1, GND on L4_GND2 to GND on L6_GND3, P1V2 on L8_PWR1; partial, GND continues through a via of GND 0.8 mm away; stitching none, 35.3 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V2 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.A5' at (177, -85.7), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V2 on L8_PWR1; partial, GND continues through a via of GND 0.8 mm away; stitching none, 35.1 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V2 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. Measured for reference: C92 (470 nF 0402) 2.0 mm away: 0.69 nH in the 0.630 mm cavity + 0.30 nH ESL + 1.52 nH mounting (pad 0.07 nH (vias beside the pads) + 1 via pair(s) at 1.79 mm pitch, 0.20 mm drill, 1.26 mm to the farther plane): 79.1 ohm at the knee, 158 % of 50 ohm (assumed track impedance). None.
      • '/DDR4-PS.A5' at (195.77, -74.03), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V2 on L8_PWR1; partial, GND continues through a via of GND 1.13 mm away; stitching none, 35.1 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V2 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. Measured for reference: C182 (470 nF 0402) 2.5 mm away: 0.75 nH in the 0.630 mm cavity + 0.30 nH ESL + 1.14 nH mounting (pad 0.07 nH (vias beside the pads) + 1 via pair(s) at 0.84 mm pitch, 0.20 mm drill, 1.26 mm to the farther plane): 68.9 ohm at the knee, 138 % of 50 ohm (assumed track impedance). None.
      • '/DDR4-PS.BA1' at (194.97, -74.03), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V2 on L8_PWR1; partial, GND continues through a via of GND 0.8 mm away; stitching none, 35.1 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V2 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. Measured for reference: C175 (4.7 uF 0402) 2.5 mm away: 0.75 nH in the 0.630 mm cavity + 0.30 nH ESL + 1.12 nH mounting (pad 0.07 nH (vias beside the pads) + 1 via pair(s) at 0.80 mm pitch, 0.20 mm drill, 1.26 mm to the farther plane): 68.2 ohm at the knee, 136 % of 50 ohm (assumed track impedance). None.
      • '/DDR4-PS.DQ13' at (182.6, -88.1), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V2 on L8_PWR1; partial, GND continues through a via of GND 1.79 mm away; stitching none, 27.6 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V2 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.DQ13' at (186.17, -74.03), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V2 on L8_PWR1; partial, GND continues through a via of GND 0.8 mm away; stitching none, 27.6 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V2 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. Measured for reference: C202 (470 nF 0402) 2.3 mm away: 0.72 nH in the 0.630 mm cavity + 0.30 nH ESL + 1.12 nH mounting (pad 0.07 nH (vias beside the pads) + 1 via pair(s) at 0.80 mm pitch, 0.20 mm drill, 1.26 mm to the farther plane): 67.3 ohm at the knee, 135 % of 50 ohm (assumed track impedance). None.
      • '/DDR4-PS.DM1' at (180.2, -88.9), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V2 on L8_PWR1; partial, GND continues through a via of GND 1.6 mm away; stitching none, 27.6 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V2 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.DM1' at (187.77, -78.83), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V2 on L8_PWR1; partial, GND continues through a via of GND 0.8 mm away; stitching none, 27.6 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V2 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.DQ14' at (186.97, -77.23), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V2 on L8_PWR1; partial, GND continues through a via of GND 0.8 mm away; stitching none, 27.6 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V2 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • and 27 more

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=refchange%3AGND%7CP1V2 serves each with its own layer_image region; all of them lie in x 175..200.86, y -91.7..-59.24 mm (board frame, Y up).

      The reference on each layer is the first plane the track meets leaving the via, anti-pads and the net's own copper skipped. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • refchange:GND|P1V842 nets change reference between GND and P1V8 at 55 vias, each with another reference that continues across (fastest: /DDR4-PS.DQ10 at 100 ps 10-90 %, ASSUMED)

      At 55 via(s), on 42 net(s), the signal leaves a layer referenced by one of GND and P1V8 for a layer referenced by the other, and no plane references both layers. Different nets: no via can carry the return across; only decoupling between the two planes near the via can. The capacitor carrying the return across was measured at each via, at the knee of the crossing net's edge against 50 ohm assumed: 55 with none. 55 of them at an ASSUMED edge (no edge on file for the net): state one (net_timing_state) and it is re-read. One decision covers the pair: whether GND and P1V8 are decoupled to each other near these vias. A via whose layers have more references than these two is a site of each pair it joins. Every one is partial: another reference continues across through a stitching via of its net, so only part of the return has nowhere to go, and the finding is the lower tier (info). They are listed with the farthest stitching via first.

      • '/WR_DAC.SYNC' at (155.25, -124), L3 -> L7: GND on L2_GND1, GND on L4_GND2 to GND on L6_GND3, P1V8 on L8_PWR1; partial, GND continues through a via of GND 0.71 mm away; stitching none, 84.6 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V8 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_P1' at (181, -102.5), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V8 on L8_PWR1; partial, GND continues through a via of GND 1.13 mm away; stitching none, 78.0 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V8 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_N1' at (181, -103.3), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V8 on L8_PWR1; partial, GND continues through a via of GND 0.8 mm away; stitching none, 77.9 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V8 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_P3' at (182.6, -106.5), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V8 on L8_PWR1; partial, GND continues through a via of GND 0.8 mm away; stitching none, 76.6 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V8 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_N3' at (182.6, -107.3), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V8 on L8_PWR1; partial, GND continues through a via of GND 1.13 mm away; stitching none, 76.3 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V8 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_N2' at (182.6, -104.9), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V8 on L8_PWR1; partial, GND continues through a via of GND 1.21 mm away; stitching none, 75.5 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V8 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_P2' at (181.8, -104.9), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V8 on L8_PWR1; partial, GND continues through a via of GND 0.8 mm away; stitching none, 75.5 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V8 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_P4' at (183.44, -111.27), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V8 on L8_PWR1; partial, GND continues through a via of GND 4 mm away; stitching none, 74.7 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V8 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_N4' at (183.44, -111.77), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V8 on L8_PWR1; partial, GND continues through a via of GND 4.27 mm away; stitching none, 74.5 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V8 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_P0' at (182.6, -101.7), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V8 on L8_PWR1; partial, GND continues through a via of GND 1.79 mm away; stitching none, 73.2 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V8 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_N0' at (182.6, -102.5), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V8 on L8_PWR1; partial, GND continues through a via of GND 1.21 mm away; stitching none, 72.9 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V8 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/PXIe-TIMING.DSTARB_P10' at (181.8, -105.7), TOP_LAYER -> L7: GND on L2_GND1 to GND on L6_GND3, P1V8 on L8_PWR1; partial, GND continues through a via of GND 0.8 mm away; stitching none, 72.5 mm of track past the via on L7: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P1V8 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • and 43 more

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=refchange%3AGND%7CP1V8 serves each with its own layer_image region; all of them lie in x 96.05..220.73, y -140.16..-64.1 mm (board frame, Y up).

      The reference on each layer is the first plane the track meets leaving the via, anti-pads and the net's own copper skipped. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • refchange:GND|P2V519 nets change reference between GND and P2V5 at 25 vias, each with another reference that continues across (fastest: /DDR4-PS.A0 at 100 ps 10-90 %, ASSUMED)

      At 25 via(s), on 19 net(s), the signal leaves a layer referenced by one of GND and P2V5 for a layer referenced by the other, and no plane references both layers. Different nets: no via can carry the return across; only decoupling between the two planes near the via can. The capacitor carrying the return across was measured at each via, at the knee of the crossing net's edge against 50 ohm assumed: 25 with none. 25 of them at an ASSUMED edge (no edge on file for the net): state one (net_timing_state) and it is re-read. One decision covers the pair: whether GND and P2V5 are decoupled to each other near these vias. A via whose layers have more references than these two is a site of each pair it joins. Every one is partial: another reference continues across through a stitching via of its net, so only part of the return has nowhere to go, and the finding is the lower tier (info). They are listed with the farthest stitching via first.

      • '/DDR4-PS.DQ23' at (190.99, -64.43), TOP_LAYER -> L10: GND on L2_GND1 to P2V5 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 1.6 mm away; stitching none, 36.4 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P2V5 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.DQS_N2' at (188.59, -66.83), TOP_LAYER -> L10: GND on L2_GND1 to P2V5 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.8 mm away; stitching none, 36.4 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P2V5 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.DQS_P2' at (189.39, -66.83), TOP_LAYER -> L10: GND on L2_GND1 to P2V5 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.8 mm away; stitching none, 36.3 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P2V5 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.DQ17' at (188.59, -64.43), TOP_LAYER -> L10: GND on L2_GND1 to P2V5 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.97 mm away; stitching none, 36.1 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P2V5 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.DQ18' at (190.19, -66.83), TOP_LAYER -> L10: GND on L2_GND1 to P2V5 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.8 mm away; stitching none, 36.1 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P2V5 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.DQ20' at (190.19, -68.43), TOP_LAYER -> L10: GND on L2_GND1 to P2V5 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.8 mm away; stitching none, 36.1 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P2V5 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.DQ21' at (190.19, -63.63), TOP_LAYER -> L10: GND on L2_GND1 to P2V5 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.8 mm away; stitching none, 36.1 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P2V5 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.DM2' at (187.79, -64.43), TOP_LAYER -> L10: GND on L2_GND1 to P2V5 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.8 mm away; stitching none, 36.1 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P2V5 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.DQ16' at (189.39, -68.43), TOP_LAYER -> L10: GND on L2_GND1 to P2V5 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.8 mm away; stitching none, 36.1 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P2V5 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.DQ22' at (190.99, -66.83), TOP_LAYER -> L10: GND on L2_GND1 to P2V5 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.8 mm away; stitching none, 36.0 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P2V5 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.DQ19' at (190.19, -64.43), TOP_LAYER -> L10: GND on L2_GND1 to P2V5 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 1.13 mm away; stitching none, 36.0 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P2V5 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/DDR4-PS.A4' at (194.97, -77.23), L3 -> L10: GND on L2_GND1, GND on L4_GND2 to P2V5 on L9_PWR2, GND on L11_GND4; partial, GND continues through a via of GND 0.8 mm away; stitching none, 35.2 mm of track past the via on L10: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P2V5 are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • and 13 more

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=refchange%3AGND%7CP2V5 serves each with its own layer_image region; all of them lie in x 185.79..199.37, y -79.23..-61.63 mm (board frame, Y up).

      The reference on each layer is the first plane the track meets leaving the via, anti-pads and the net's own copper skipped. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • split@L8_PWR1:P0V85|P1V820 nets cross the split between the P0V85 and P1V8 planes on L8_PWR1 at 21 places, all with a plane on the track's other side (fastest: /DDR4-PS.BA1 at 100 ps 10-90 %, ASSUMED)

      On L8_PWR1 the P0V85 and P1V8 planes part, and 20 nets cross from one to the other at 21 place(s). The return current cannot follow a track across; it detours to wherever the two planes are joined (a stitching capacitor, or the source), and the loop it opens radiates and couples. One decision covers the boundary: whether it is stitched near these crossings, or the nets crossing it are slow enough not to care. Tracks that run ALONG this split, down the slot rather than across it, are a finding of their own: theirs is an impedance step, not a detour. At 21 of the 21 crossing(s) the track's other neighbour carries a plane, which the return can use there. Every crossing has a plane on the track's other side at most twice as far away as the broken one, so at most 48% of the return meets the break and the rest follows the track on that plane (to first order each plane beside a stripline carries a share of the return inversely proportional to its distance from the track). The finding is the lower tier (info).

      • '/DDR4-PS.BA1' on L7 at (179.4, -91.7), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/I2C.SDA' on L7 at (166.96, -90), 0.35 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PS-QSPI.UWR_CLK' on L7 at (179.4, -92.5), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARB_N15' on L7 at (179.43, -108.28), 0.42 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARB_N7' on L7 at (179.4, -107.34), 0.42 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARB_P15' on L7 at (179.4, -108.5), 0.43 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARB_P7' on L7 at (179.4, -107.09), 0.42 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARC_MUX_SEL2' on L7 at (179.36, -100.63), 0.38 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARC_MUX_SEL2' on L7 at (179.4, -109.65), 0.42 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe.PERSTn' on L7 at (166.6, -91.06), 0.42 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/WR_CLK.MAIN_N' on L7 at (156.6, -103.52), 0.42 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/WR_CLK.MAIN_P' on L7 at (156.6, -103.27), 0.42 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • and 9 more

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=split%40L8_PWR1%3AP0V85%7CP1V8 serves each with its own layer_image region; all of them lie in x 155.36..180.55, y -110.8..-88.9 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • split@L8_PWR1:P12V|P1V827 nets cross the split between the P12V and P1V8 planes on L8_PWR1 at 35 places, all with a plane on the track's other side (fastest: /PXIe-TIMING.DSTARB_N1 at 100 ps 10-90 %, ASSUMED)

      On L8_PWR1 the P12V and P1V8 planes part, and 27 nets cross from one to the other at 35 place(s). The return current cannot follow a track across; it detours to wherever the two planes are joined (a stitching capacitor, or the source), and the loop it opens radiates and couples. One decision covers the boundary: whether it is stitched near these crossings, or the nets crossing it are slow enough not to care. Tracks that run ALONG this split, down the slot rather than across it, are a finding of their own: theirs is an impedance step, not a detour. At 35 of the 35 crossing(s) the track's other neighbour carries a plane, which the return can use there. Every crossing has a plane on the track's other side at most twice as far away as the broken one, so at most 48% of the return meets the break and the rest follows the track on that plane (to first order each plane beside a stripline carries a share of the return inversely proportional to its distance from the track). The finding is the lower tier (info).

      • '/PXIe-TIMING.DSTARB_N1' on L7 at (219.05, -108.76), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARB_N10' on L7 at (219.05, -116.84), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARB_N11' on L7 at (219.05, -126.55), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARB_N12' on L7 at (219.05, -133.62), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARB_N15' on L7 at (214.05, -146.09), 0.42 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARB_N2' on L7 at (219.05, -116.37), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARB_N3' on L7 at (219.05, -123.79), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARB_N4' on L7 at (219.05, -131.25), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARB_N7' on L7 at (219.05, -134.27), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARB_N9' on L7 at (219.05, -109.29), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARB_P0' on L7 at (219.14, -105.49), 0.35 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/PXIe-TIMING.DSTARB_P1' on L7 at (219.05, -108.58), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • and 23 more

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=split%40L8_PWR1%3AP12V%7CP1V8 serves each with its own layer_image region; all of them lie in x 118.92..220.2, y -150.59..-81.2 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • split@L8_PWR1:P12V|P3V3_CLK4 nets cross the split between the P12V and P3V3_CLK planes on L8_PWR1 at 4 places, all with a plane on the track's other side (fastest: /WR_DAC.DIN at 100 ps 10-90 %, ASSUMED)

      On L8_PWR1 the P12V and P3V3_CLK planes part, and 4 nets cross from one to the other at 4 place(s). The return current cannot follow a track across; it detours to wherever the two planes are joined (a stitching capacitor, or the source), and the loop it opens radiates and couples. One decision covers the boundary: whether it is stitched near these crossings, or the nets crossing it are slow enough not to care. Tracks that run ALONG this split, down the slot rather than across it, are a finding of their own: theirs is an impedance step, not a detour. At 4 of the 4 crossing(s) the track's other neighbour carries a plane, which the return can use there. Every crossing has a plane on the track's other side at most twice as far away as the broken one, so at most 48% of the return meets the break and the rest follows the track on that plane (to first order each plane beside a stripline carries a share of the return inversely proportional to its distance from the track). The finding is the lower tier (info).

      • '/WR_DAC.DIN' on L7 at (118.87, -145.39), 0.18 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/WR_DAC.LDAC' on L7 at (118.83, -145.75), 0.17 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/WR_DAC.SCLK' on L7 at (118.87, -145.07), 0.18 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/WR_DAC.SYNC' on L7 at (118.87, -144.49), 0.18 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=split%40L8_PWR1%3AP12V%7CP3V3_CLK serves each with its own layer_image region; all of them lie in x 117.74..119.93, y -146.75..-143.42 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • split@L8_PWR1:P1V2|P1V810 nets cross the split between the P1V2 and P1V8 planes on L8_PWR1 at 17 places, all with a plane on the track's other side (fastest: /DDR4-PS.BA1 at 100 ps 10-90 %, ASSUMED)

      On L8_PWR1 the P1V2 and P1V8 planes part, and 10 nets cross from one to the other at 17 place(s). The return current cannot follow a track across; it detours to wherever the two planes are joined (a stitching capacitor, or the source), and the loop it opens radiates and couples. One decision covers the boundary: whether it is stitched near these crossings, or the nets crossing it are slow enough not to care. Tracks that run ALONG this split, down the slot rather than across it, are a finding of their own: theirs is an impedance step, not a detour. At 17 of the 17 crossing(s) the track's other neighbour carries a plane, which the return can use there. Every crossing has a plane on the track's other side at most twice as far away as the broken one, so at most 48% of the return meets the break and the rest follows the track on that plane (to first order each plane beside a stripline carries a share of the return inversely proportional to its distance from the track). The finding is the lower tier (info).

      • '/DDR4-PS.BA1' on L7 at (194.59, -80.43), 0.31 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/DDR4-PS.DM1' on L7 at (183.59, -89.3), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/DDR4-PS.DM1' on L7 at (188.59, -82.83), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/DDR4-PS.DQ10' on L7 at (190.36, -82.83), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/DDR4-PS.DQ11' on L7 at (183.59, -88.9), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/DDR4-PS.DQ11' on L7 at (186.89, -82.87), 0.33 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/DDR4-PS.DQ12' on L7 at (183.29, -89.7), 0.43 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/DDR4-PS.DQ13' on L7 at (183.59, -88.1), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/DDR4-PS.DQ13' on L7 at (184.4, -83.32), 0.38 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/DDR4-PS.DQ13' on L7 at (184.72, -83.12), 0.44 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/DDR4-PS.DQ14' on L7 at (190.97, -82.83), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/DDR4-PS.DQ8' on L7 at (183.59, -88.5), 0.3 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • and 5 more

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=split%40L8_PWR1%3AP1V2%7CP1V8 serves each with its own layer_image region; all of them lie in x 182.07..195.59, y -90.7..-79.27 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • split@L8_PWR1:P1V8|P3V3_CLK2 nets cross the split between the P1V8 and P3V3_CLK planes on L8_PWR1 at 2 places, all with a plane on the track's other side (fastest: /WR_CLK.MAIN_N at 1 ns 10-90 %, suggested by its name)

      On L8_PWR1 the P1V8 and P3V3_CLK planes part, and 2 nets cross from one to the other at 2 place(s). The return current cannot follow a track across; it detours to wherever the two planes are joined (a stitching capacitor, or the source), and the loop it opens radiates and couples. One decision covers the boundary: whether it is stitched near these crossings, or the nets crossing it are slow enough not to care. Tracks that run ALONG this split, down the slot rather than across it, are a finding of their own: theirs is an impedance step, not a detour. At 2 of the 2 crossing(s) the track's other neighbour carries a plane, which the return can use there. Every crossing has a plane on the track's other side at most twice as far away as the broken one, so at most 48% of the return meets the break and the rest follows the track on that plane (to first order each plane beside a stripline carries a share of the return inversely proportional to its distance from the track). The finding is the lower tier (info).

      • '/WR_CLK.MAIN_N' on L7 at (116.94, -130.98), 0.11 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1
      • '/WR_CLK.MAIN_P' on L7 at (116.76, -130.98), 0.11 mm; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=split%40L8_PWR1%3AP1V8%7CP3V3_CLK serves each with its own layer_image region; all of them lie in x 115.76..117.94, y -132.03..-129.92 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • split@L9_PWR2:P3V3|VCC_PSPLL'/WR_DAC.LDAC' crosses the split between the P3V3 and VCC_PSPLL planes on L9_PWR2 at 2 places, all with a plane on the track's other side (fastest: /WR_DAC.LDAC at 100 ps 10-90 %, ASSUMED)

      On L9_PWR2 the P3V3 and VCC_PSPLL planes part, and '/WR_DAC.LDAC' crosses from one to the other at 2 place(s). The return current cannot follow a track across; it detours to wherever the two planes are joined (a stitching capacitor, or the source), and the loop it opens radiates and couples. One decision covers the boundary: whether it is stitched near these crossings, or the nets crossing it are slow enough not to care. Tracks that run ALONG this split, down the slot rather than across it, are a finding of their own: theirs is an impedance step, not a detour. At 2 of the 2 crossing(s) the track's other neighbour carries a plane, which the return can use there. Every crossing has a plane on the track's other side at most twice as far away as the broken one, so at most 48% of the return meets the break and the rest follows the track on that plane (to first order each plane beside a stripline carries a share of the return inversely proportional to its distance from the track). The finding is the lower tier (info).

      • '/WR_DAC.LDAC' on L10 at (175.49, -112.84), 0.1 mm; other side: P0V85 on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2
      • '/WR_DAC.LDAC' on L10 at (177.7, -97.51), 0.16 mm; other side: P0V85 on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=split%40L9_PWR2%3AP3V3%7CVCC_PSPLL serves each with its own layer_image region; all of them lie in x 174.49..178.76, y -113.89..-96.45 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • void:GND@L11_GND4:96.34,-81.54'/power-supply-1/P1V8AUX_PG' crosses where the outline of the GND plane on L11_GND4 and the GND plane on L13_GND5 is drawn back, at 1 place, the longest 0.56 mm; 1 cleared on both sides of the track (fastest: /power-supply-1/P1V8AUX_PG at 20 ns 10-90 %, suggested by its name)

      Along '/power-supply-1/P1V8AUX_PG' on L12, the GND plane on L11_GND4 has 0.56 mm under the track with no copper, at (96.34, -81.54), the same plane on both sides. No single drill's anti-pad and no landing explains it: a clearance is excused only when it holds a drill and is drawn round it, or the track ends in it on the net's own pad or via, or runs through its own via's clearance and no farther. The return current detours round the void. 2 of them run ALONG the bare copper, parallel to its edges, rather than across it (the longest 0.27 mm). No hole cleared it: the outline of the pour at x 73.7..233.5, y -152.39..-57 is drawn back there. At 1 place(s) the track's plane is cleared on both sides of it, the same stretch on each reference layer: it has no plane on either side there, counted once. One decision covers the void: whether these tracks may cross it, or the plane should be filled under them. No crossing has a plane on the track's other side, so the whole return meets the break.

      • one stretch, cleared on 2 reference layers: '/power-supply-1/P1V8AUX_PG' on L12 at (96.34, -81.54), 0.56 mm, running ALONG the bare copper for 0.27 mm, not across it over no copper; other side: no plane on L13_GND5 / '/power-supply-1/P1V8AUX_PG' on L12 at (96.34, -81.54), 0.56 mm, running ALONG the bare copper for 0.27 mm, not across it over no copper; other side: no plane on L11_GND4

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=void%3AGND%40L11_GND4%3A96.34%2C-81.54 serves each with its own layer_image region; all of them lie in x 95.34..97.34, y -82.82..-80.26 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • void:P12V@L8_PWR1:157.18,-71.41'/power-supply-1/P1V8AUX_PG' crosses where the outline of the P12V plane on L8_PWR1 is drawn back, at 1 place, the longest 0.44 mm (fastest: /power-supply-1/P1V8AUX_PG at 20 ns 10-90 %, suggested by its name)

      Along '/power-supply-1/P1V8AUX_PG' on L7, the P12V plane on L8_PWR1 has 0.44 mm under the track with no copper, at (157.18, -71.41), the same plane on both sides. No single drill's anti-pad and no landing explains it: a clearance is excused only when it holds a drill and is drawn round it, or the track ends in it on the net's own pad or via, or runs through its own via's clearance and no farther. The return current detours round the void. 1 of them run ALONG the bare copper, parallel to its edges, rather than across it (the longest 0.44 mm). No hole cleared it: the outline of the pour at x 95.86..232.7, y -154..-56.01 is drawn back there. One decision covers the void: whether these tracks may cross it, or the plane should be filled under them. Every crossing has a plane on the track's other side at most twice as far away as the broken one, so at most 48% of the return meets the break and the rest follows the track on that plane (to first order each plane beside a stripline carries a share of the return inversely proportional to its distance from the track). The finding is the lower tier (info).

      • '/power-supply-1/P1V8AUX_PG' on L7 at (157.18, -71.41), 0.44 mm, running ALONG the bare copper for 0.44 mm, not across it over no copper; other side: GND on L6_GND3, 0.1 mm from the track against 0.11 mm to L8_PWR1: 48% of the return on L8_PWR1

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=void%3AP12V%40L8_PWR1%3A157.18%2C-71.41 serves each with its own layer_image region; all of them lie in x 155.96..158.4, y -72.41..-70.41 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • void:routed@L12|L13_GND5'/fpga/fpga-hp-banks/F_PP_OE_N' on L12 crosses the tracks routed through the GND pour on L13_GND5 at 1 place, the longest 0.58 mm (fastest: /fpga/fpga-hp-banks/F_PP_OE_N at 100 ps 10-90 %, ASSUMED)

      The GND pour on L13_GND5 is cleared round tracks routed through it, and tracks on L12, which take that pour as their reference, cross them: for the width of each cut the return has no plane under it and detours round the cut. 1 place(s) on 1 net(s). One decision covers the layer pair: whether these layers may take a pour their own tracks cut as the reference, or the routing or the stackup should give the tracks a plane that is not cut where they cross. Every crossing has a plane on the track's other side at most twice as far away as the broken one, so at most 52% of the return meets the break and the rest follows the track on that plane (to first order each plane beside a stripline carries a share of the return inversely proportional to its distance from the track). The finding is the lower tier (info).

      • '/fpga/fpga-hp-banks/F_PP_OE_N' on L12 at (100.54, -124.94), 0.58 mm over copper of /io_drivers_pp/bank0/OUT3, not a plane, to the end of the track; other side: GND on L11_GND4, 0.11 mm from the track against 0.1 mm to L13_GND5: 52% of the return on L13_GND5

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=void%3Arouted%40L12%7CL13_GND5 serves each with its own layer_image region; all of them lie in x 99.23..101.85, y -126.25..-123.63 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

  17. ReferencePlaneDiscontinuityminorOpen

    ReferencePlaneDiscontinuity: 4 boards×4

    Checked by the review, stands · Claude coordinator (Cern session) — Re-affirmed after the 2026-10-28 SI5340B-D-GM supply-pin facts (datasheet a3f57dfa p41) staled it: they made P3V3_CLK/P1V8_CLK supply rails, and the finding's substance is unchanged. Checked: the 54 findings are tracks running along or across power-plane splits (P0V85/P1V8/P1V2/MGT_1V8/VCC_PSPLL on L8/L9) and reference changes at vias between GND and power planes. They are layout practice; no datasheet or interface specification on file states a requirement for these nets, and the nets' edges are only assumed (no IBIS on file), so their effect cannot be quantified. They stand for the owner.

    At refchange:GND|P3V3_CLK, split@L11_GND4:GND|P0V85, split@L9_PWR2:MGT_1V8|P3V3, split@L9_PWR2:P2V5|P3V3

    Each finding, with its own headline
    • refchange:GND|P3V3_CLK6 nets change reference between GND and P3V3_CLK at 6 vias, each with another reference that continues across (fastest: /WR_DAC.DIN at 100 ps 10-90 %, ASSUMED)

      At 6 via(s), on 6 net(s), the signal leaves a layer referenced by one of GND and P3V3_CLK for a layer referenced by the other, and no plane references both layers. Different nets: no via can carry the return across; only decoupling between the two planes near the via can. The capacitor carrying the return across was measured at each via, at the knee of the crossing net's edge against 50 ohm assumed: 6 with none. 6 of them at an ASSUMED edge (no edge on file for the net): state one (net_timing_state) and it is re-read. One decision covers the pair: whether GND and P3V3_CLK are decoupled to each other near these vias. A via whose layers have more references than these two is a site of each pair it joins. Every one is partial: another reference continues across through a stitching via of its net, so only part of the return has nowhere to go, and the finding is the lower tier (info). They are listed with the farthest stitching via first.

      • '/WR_DAC.DIN' at (91.52, -136.8), L7 -> BOTTOM_LAYER: GND on L6_GND3, P3V3_CLK on L8_PWR1 to GND on L13_GND5; partial, GND continues through a via of GND 1.02 mm away; stitching none, 3.6 mm of track past the via on BOTTOM_LAYER: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3_CLK are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. Measured for reference: C260 (0.1 nF 0402) 1.7 mm away: 0.22 nH in the 0.210 mm cavity + 0.30 nH ESL + 0.57 nH mounting (pad 0.07 nH + no via of the first net within 1 mm of its pad: the study's 0.6 nH per mm of 0.84 mm depth, ASSUMED): 34.0 ohm at the knee, 68 % of 50 ohm (assumed track impedance). None.
      • '/WR_DAC.SCLK' at (91.42, -136.3), L7 -> BOTTOM_LAYER: GND on L6_GND3, P3V3_CLK on L8_PWR1 to GND on L13_GND5; partial, GND continues through a via of GND 1.46 mm away; stitching none, 1.8 mm of track past the via on BOTTOM_LAYER: At the 5 GHz knee of a 100 ps edge (1 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3_CLK are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. Measured for reference: C260 (0.1 nF 0402) 1.7 mm away: 0.22 nH in the 0.210 mm cavity + 0.30 nH ESL + 0.57 nH mounting (pad 0.07 nH + no via of the first net within 1 mm of its pad: the study's 0.6 nH per mm of 0.84 mm depth, ASSUMED): 34.0 ohm at the knee, 68 % of 50 ohm (assumed track impedance). None.
      • '/WR_CLK.MAIN_N' at (118.92, -138.69), L7 -> BOTTOM_LAYER: GND on L6_GND3, P3V3_CLK on L8_PWR1 to GND on L13_GND5; partial, GND continues through a via of GND 2.39 mm away; stitching none, 1.2 mm of track past the via on BOTTOM_LAYER: At the 5 GHz knee of a 100 ps edge (1 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3_CLK are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/WR_CLK.MAIN_P' at (118.92, -139.24), L7 -> BOTTOM_LAYER: GND on L6_GND3, P3V3_CLK on L8_PWR1 to GND on L13_GND5; partial, GND continues through a via of GND 2.44 mm away; stitching none, 1.2 mm of track past the via on BOTTOM_LAYER: At the 5 GHz knee of a 100 ps edge (1 ns 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3_CLK are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.
      • '/WR_DAC.SYNC' at (91.47, -135.8), L7 -> BOTTOM_LAYER: GND on L6_GND3, P3V3_CLK on L8_PWR1 to GND on L13_GND5; partial, GND continues through a via of GND 1.96 mm away; stitching none, 1.1 mm of track past the via on BOTTOM_LAYER: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3_CLK are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. Measured for reference: C260 (0.1 nF 0402) 2.0 mm away: 0.23 nH in the 0.210 mm cavity + 0.30 nH ESL + 0.57 nH mounting (pad 0.07 nH + no via of the first net within 1 mm of its pad: the study's 0.6 nH per mm of 0.84 mm depth, ASSUMED): 34.4 ohm at the knee, 69 % of 50 ohm (assumed track impedance). None.
      • '/WR_DAC.LDAC' at (90.42, -131.32), L7 -> BOTTOM_LAYER: GND on L6_GND3, P3V3_CLK on L8_PWR1 to GND on L13_GND5; partial, GND continues through a via of GND 2.35 mm away; stitching none, 0.8 mm of track past the via on BOTTOM_LAYER: At the 5 GHz knee of a 100 ps edge (100 ps 10-90 %, assumed: no edge is on file for the net) no capacitor carries the return; only the planes do. GND and P3V3_CLK are not a thin pair (at most 0.2 mm), so their own capacitance does not carry the return. To serve it, change layers across the same plane, or move the signal to a layer referenced to ground. None.

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=refchange%3AGND%7CP3V3_CLK serves each with its own layer_image region; all of them lie in x 88.42..120.92, y -141.24..-129.32 mm (board frame, Y up).

      The reference on each layer is the first plane the track meets leaving the via, anti-pads and the net's own copper skipped. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • split@L11_GND4:GND|P0V854 nets cross the split between the GND and P0V85 planes on L11_GND4 at 5 places, all with a plane on the track's other side (fastest: /DDR4-PS.A0 at 100 ps 10-90 %, ASSUMED)

      On L11_GND4 the GND and P0V85 planes part, and 4 nets cross from one to the other at 5 place(s). The return current cannot follow a track across; it detours to wherever the two planes are joined (a stitching capacitor, or the source), and the loop it opens radiates and couples. One decision covers the boundary: whether it is stitched near these crossings, or the nets crossing it are slow enough not to care. Tracks that run ALONG this split, down the slot rather than across it, are a finding of their own: theirs is an impedance step, not a detour. At 5 of the 5 crossing(s) the track's other neighbour carries a plane, which the return can use there. Every crossing has a plane on the track's other side at most twice as far away as the broken one, so at most 52% of the return meets the break and the rest follows the track on that plane (to first order each plane beside a stripline carries a share of the return inversely proportional to its distance from the track). The finding is the lower tier (info).

      • '/DDR4-PS.A0' on L10 at (178.51, -90.24), 0.28 mm; other side: P3V3 on L9_PWR2, 0.11 mm from the track against 0.1 mm to L11_GND4: 52% of the return on L11_GND4
      • '/WRFLASH_I2C.SDA' on L10 at (162.95, -100.5), 0.11 mm; other side: P3V3 on L9_PWR2, 0.11 mm from the track against 0.1 mm to L11_GND4: 52% of the return on L11_GND4
      • '/WR_CLK.SPI_MISO' on L10 at (162.95, -111.51), 0.16 mm; other side: P3V3 on L9_PWR2, 0.11 mm from the track against 0.1 mm to L11_GND4: 52% of the return on L11_GND4
      • '/WR_DAC.LDAC' on L10 at (171.95, -134.06), 0.11 mm; other side: MGT_1V8 on L9_PWR2, 0.11 mm from the track against 0.1 mm to L11_GND4: 52% of the return on L11_GND4
      • '/WR_DAC.LDAC' on L10 at (178.56, -96.66), 0.16 mm; other side: P3V3 on L9_PWR2, 0.11 mm from the track against 0.1 mm to L11_GND4: 52% of the return on L11_GND4

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=split%40L11_GND4%3AGND%7CP0V85 serves each with its own layer_image region; all of them lie in x 161.89..179.61, y -135.11..-89.14 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • split@L9_PWR2:MGT_1V8|P3V33 nets cross the split between the MGT_1V8 and P3V3 planes on L9_PWR2 at 3 places, all with a plane on the track's other side (fastest: /WR_DAC.LDAC at 100 ps 10-90 %, ASSUMED)

      On L9_PWR2 the MGT_1V8 and P3V3 planes part, and 3 nets cross from one to the other at 3 place(s). The return current cannot follow a track across; it detours to wherever the two planes are joined (a stitching capacitor, or the source), and the loop it opens radiates and couples. One decision covers the boundary: whether it is stitched near these crossings, or the nets crossing it are slow enough not to care. Tracks that run ALONG this split, down the slot rather than across it, are a finding of their own: theirs is an impedance step, not a detour. At 3 of the 3 crossing(s) the track's other neighbour carries a plane, which the return can use there. Every crossing has a plane on the track's other side at most twice as far away as the broken one, so at most 48% of the return meets the break and the rest follows the track on that plane (to first order each plane beside a stripline carries a share of the return inversely proportional to its distance from the track). The finding is the lower tier (info).

      • '/PXIe-MGT.TX_N1' on L10 at (164.55, -87.54), 0.13 mm; other side: GND on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2
      • '/PXIe-MGT.TX_P1' on L10 at (164.55, -88.36), 0.14 mm; other side: GND on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2
      • '/WR_DAC.LDAC' on L10 at (174.08, -123.39), 0.15 mm; other side: P0V85 on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=split%40L9_PWR2%3AMGT_1V8%7CP3V3 serves each with its own layer_image region; all of them lie in x 163.5..175.13, y -124.45..-86.5 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

    • split@L9_PWR2:P2V5|P3V313 nets cross the split between the P2V5 and P3V3 planes on L9_PWR2 at 13 places, all with a plane on the track's other side (fastest: /DDR4-PS.A0 at 100 ps 10-90 %, ASSUMED)

      On L9_PWR2 the P2V5 and P3V3 planes part, and 13 nets cross from one to the other at 13 place(s). The return current cannot follow a track across; it detours to wherever the two planes are joined (a stitching capacitor, or the source), and the loop it opens radiates and couples. One decision covers the boundary: whether it is stitched near these crossings, or the nets crossing it are slow enough not to care. Tracks that run ALONG this split, down the slot rather than across it, are a finding of their own: theirs is an impedance step, not a detour. At 13 of the 13 crossing(s) the track's other neighbour carries a plane, which the return can use there. Every crossing has a plane on the track's other side at most twice as far away as the broken one, so at most 48% of the return meets the break and the rest follows the track on that plane (to first order each plane beside a stripline carries a share of the return inversely proportional to its distance from the track). The finding is the lower tier (info).

      • '/DDR4-PS.A0' on L10 at (195.39, -80.24), 0.11 mm; other side: GND on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2
      • '/DDR4-PS.A13' on L10 at (192.35, -80.24), 0.11 mm; other side: GND on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2
      • '/DDR4-PS.A4' on L10 at (193.57, -80.24), 0.11 mm; other side: GND on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2
      • '/DDR4-PS.DM2' on L10 at (184.69, -65.3), 0.1 mm; other side: GND on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2
      • '/DDR4-PS.DQ16' on L10 at (184.69, -67.21), 0.1 mm; other side: GND on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2
      • '/DDR4-PS.DQ17' on L10 at (184.69, -64.01), 0.1 mm; other side: GND on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2
      • '/DDR4-PS.DQ18' on L10 at (185.81, -80.24), 0.11 mm; other side: GND on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2
      • '/DDR4-PS.DQ19' on L10 at (184.69, -62.47), 0.1 mm; other side: GND on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2
      • '/DDR4-PS.DQ21' on L10 at (184.69, -63.22), 0.1 mm; other side: GND on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2
      • '/DDR4-PS.DQ22' on L10 at (188.99, -80.24), 0.11 mm; other side: GND on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2
      • '/DDR4-PS.DQ23' on L10 at (191.37, -80.24), 0.11 mm; other side: GND on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2
      • '/DDR4-PS.DQS_N2' on L10 at (184.69, -65.94), 0.1 mm; other side: GND on L11_GND4, 0.1 mm from the track against 0.11 mm to L9_PWR2: 48% of the return on L9_PWR2
      • and 1 more

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/96/reference-planes?group=split%40L9_PWR2%3AP2V5%7CP3V3 serves each with its own layer_image region; all of them lie in x 183.64..196.39, y -81.29..-61.47 mm (board frame, Y up).

      Read from the copper as drawn on the reference layer, along each track's centreline at exact breakpoints. The reference is the copper beside the track, never the layer's type or name. Whether it matters depends on each signal's edge rate, which nothing on file states for these nets: a slow static line crosses a split harmlessly, a fast edge does not.

Main parts

Part numberMakerDescriptionWhere
0466.750NRLITTELFUSEVery Fast-Acting Surface Mount Thin Film Fuse, Size 1206, Series 466F1
1888247-1TYCO ELECTRONICS20 Position, Pitch 0.8mm SFP+ Receptacle (Small Form-factor Pluggable)J5
2007198-1TYCO ELECTRONICS1x1 Press Fit Cage for Use with SFP+ Connector (Small Form-factor Pluggable)Cage1
214443ERNI40 Contacts, 5 Rows (a,b,c,d,e) of 8 positions + 4 Shield Contacts, Pitch 2.5mm, Press-Fit, Right Angle Ermet B8 CompactPCI Express Female Connector with 4 Upper Contacts ShieldXJ4
24AA025E48T-I/OTMICROCHIP TECHNOLOGY1.7V-5.5V, 2K I2C Serial EEPROM with EUI-48™ Node Identity; CascadableIC35
24AA64T-I/MSMICROCHIP TECHNOLOGY1.7V-5.5V, 64K (8K x 8-bit) I2C Serial EEPROMIC34
435451019820WURTHSMT Washable Tactile Switch SPST-NO Top Actuated Surface MountPB2
500075-1517MOLEXUSB Mini-B Serie Type 2, Through Hole Straight ReceptacleJ2
742792040WURTH ELEKTRONIK±25% 2A SMD EMI Suppression Chip Ferrite Bead (WE-CBF Series)L2
87832-1420MOLEX2.00mm (.079") Pitch Milli-Grid™ Header, Surface Mount, Vertical, Shrouded, 14 Circuits, with Locking Windows and Centre Polarization Slot, PCB LocatorJ7
AMK325ABJ337MM-PTAIYO YUDEN±20% 4V X5R SMD Multilayer Chip Ceramic CapacitorC33
AOCJY-10.000MHZ-EABRACONCMOS 3.3V, ±10ppb, 15pF, -20°C ..+70°C, SMD Oven Controlled Crystal Oscillator (AICJY Series)OSC6
BLM31PG500SN1MURATA3.5A SMD EMIFIL Chip Ferrite BeadL19
C0402C101J5GACTUKEMET±5% 50V C0G/NP0 SMD Multilayer Chip Ceramic CapacitorC260
C0402C270J5GACTUKEMET±5% 50V C0G/NP0 SMD Multilayer Chip Ceramic CapacitorC48
C1005X7R1C224K050TDK±10% 16V X7R SMD Multilayer Chip Ceramic CapacitorC270
C1210C224F3GACKEMET±1% 25V C0G/NP0 SMD Multilayer Chip Ceramic CapacitorC271
C1808X102KHRACKEMET±10% 3kV X7R High Voltage SMD Multilayer Chip Ceramic CapacitorC246
CC1206JRNPOBBN100YAGEO PHYCOMP±5% 500V C0G/NP0 SMD Multilayer Chip Ceramic CapacitorC70
CP2105-F01-GMSILICON LABSSingle-Chip USB to Dual UART BridgeIC3
CRCW04021K40FKVISHAY±1% 0.063W ±100ppm/°C General Purpose Thick Film Chip ResistorR206
CRCW0402210KFKVISHAY±1% 0.063W ±100ppm/°C General Purpose Thick Film Chip ResistorR211
CRCW0402402RFKVISHAY±1% 0.063W ±100ppm/°C General Purpose Thick Film Chip ResistorR253
CRCW0402698RFKVISHAY±1% 0.063W ±100ppm/°C General Purpose Thick Film Chip ResistorR214
CX3225SB48000D0FPJC1KYOCERATolerance: ±10ppm, Stability: ±15ppm, 8pF, -40°C to +85°C, Surface Mount CrystalQ6
DAC8562SDSCTEXAS INSTRUMENTS16-Bit, Low-Power, Buffered, Voltage-Output DACs With 2.5-V, 4-PPM/°C Internal ReferenceIC10
EEFGX0D471RPANASONIC±20% 2V ESR 0R003 Polymer Aluminum Electrolytic CapacitorC273
HSB18-232310SAME SKYHeat Sink BGA Aluminum Alloy 3.7W @ 75°C Top MountHS1
LSHM-120-02.5-L-DV-A-SSAMTEC40 Contacts, Pitch 0.5mm, Razor Beam™ Vertical High Speed Hermaphroditic Strip, with Shield, Lead Style-02.5J6
MAX9374AEKAMAXIMDifferential LVPECL-to-LVDS TranslatorIC26
MMSZ4683ON SEMICONDUCTOR3V 500mW Zener Voltage RegulatorD17
MTLW-103-07-L-S-250SAMTEC3 Contacts, Pitch 2.54mm, Single Row Through Mount Straight Header (Post Lead = 6.35mm, Tail Lead 3.05mm, Insulator 1.52mm)J3
PA2607.181NLPULSE ELECTRONICS±15% 62A SMT Power InductorL9
PNM0402E5000BST1VISHAY±0.1% 0.05W ±25ppm/°C Surface Mount Precision Thin Film Non-Magnetic ResistorR11
PRTR5V0U2XNXP SEMICONDUCTORS5.5V Ultra Low Capacitance (1pF) Double Rail-to-Rail ESD Protection Diode (8kV)IC4
REF5030AIDTEXAS INSTRUMENTS3V, Low-Noise, Very Low Drift, Precision Voltage Reference, Standard GradeIC12
RN73C1J442RBTDFTYCO HOLSWORTHY±0.1% 0.063W ±10ppm Thin Film Chip ResistorR202
RN73C1J523RBTDFTYCO HOLSWORTHY±0.1% 0.063W ±10ppm Thin Film Chip ResistorR201
SG-8018CE 50.0000M-TJHPA3EPSON SEIKO1.6 to 3.6V, ±50ppm, 15pF, -40 to 105°C, CMOS Surface Programmable Crystal Oscillator (SG-8018CE Series)OSC1
SI5340B-D-GMSILICON LABSLow-Jitter, 4-Output, Any-Frequency, Any-Output Clock GeneratorIC18

About this design

Design
WREN XIE — OHWR0091 R2
Designed in
KiCad EDA
Board
14 copper layers, 162.5 × 100 mm, 803 components, 1125 nets
Licence
CERN-OHL-W-2.0
Source
https://edms.cern.ch/item/EDA-04811-V2-0/0 at commit EDMS EDA-048
Attribution
CERN EDMS item EDA-04811-V2-0 'WREN-XIe': document 3447981 v.2, retrieved 2026-10-23T04:40:43Z. Converted by E-Sharp with KiCad 10.0.6 (zones refilled); fit status and part identity from CERN's assembly BOM EDA-04811-V2-0_pcb-mat.xls.
Reviewed with
Probe by E# (E-Sharp AB)