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

FMC ADC 500M14B4CHA

OHWR0056 · revision R1

FMC ADC 500M14B4CHA (OHWR0056 R1) is an open-hardware circuit board from the CERN Open Hardware Repository, designed in Altium Designer: a 8-layer board of 84 × 69 mm with 402 components and 255 nets, published under CERN-OHL-S-2.0.

review 88% walked · 12 findings: 1 blocker · 4 major · 7 minor (21 items)

Open in the interactive viewerRead the full review

FMC ADC 500M14B4CHA (OHWR0056 R1) circuit board, top view: copper, silkscreen and components
Top view, rendered by Probe from the design's own fabrication data.

Design review

Probe recorded 12 findings: 1 blocker · 4 major · 7 minor (21 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. observationOBS-206blockerOpen

    LT3094 U10 exposed pad (IN) is on GND, shorting the -2.64 V VNEG2V64 rail to ground

    LT3094 U10's exposed pad (pin 13) is IN, but it is on GND, so soldering it shorts the -2.64 V VNEG2V64 rail to ground: the ADP5071 inverter goes into current limit and -1V02 never comes up.

    Where it is

    U10 pinNetSource
    1, 2 (IN)VNEG2V64netlist: component U10
    3 (EN/UV)VNEG2V64netlist: component U10
    5 (PGFB)VNEG2V64netlist: component U10
    13, exposed padGNDnetlist: component U10
    • U10 is an LT3094IDD, a -1.02 V LDO.
    • VNEG2V64 is the -2.64 V inverting output of ADP5071 U8 (sheet 5 note).
    • The datasheet: "EXPOSED PAD (PIN 13) IS IN, MUST BE SOLDERED TO PCB" (LT3094 datasheet, p.3); "IN (Pins 1, 2, Exposed Pad Pin 13): Input" (p.15).

    ``mermaid flowchart LR A["U8 ADP5071<br/>inverting output"] -- "VNEG2V64 -2.64 V" --> B["U10 LT3094<br/>IN pins 1, 2"] B -- "exposed pad pin 13 is IN,<br/>soldered to GND" --> D["GND<br/>short"] B -- "-1V02" --> E["VNLMHA..D<br/>LMH5401 negative supplies"] ``

    Why it matters

    The soldered pad connects VNEG2V64 directly to GND. The negative rail is shorted, the ADP5071 inverter goes into current limit, and -1V02 (VNLMHA..D, the LMH5401 negative supplies) never comes up.

  2. ReferencePlaneDiscontinuitymajorOpen

    'XSHIELD' crosses one void in the GND plane on GND at 1 place, the longest 1.34 mm (fastest: XSHIELD at 100 ps 10-90 %, ASSUMED)

    Along 'XSHIELD' on TOP_LAYER, the GND plane on GND has 1.34 mm under the track with no copper, at (47.7, -23.46), running to the end of the track. 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 0.89 mm). What cleared it: the hole at x 43.84..49.19, y -23.61..-23.31. 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.

    • 'XSHIELD' on TOP_LAYER at (47.7, -23.46), 1.34 mm, running ALONG the bare copper for 0.89 mm, not across it over no copper, to the end of the track; other side: none: the signal layer is outermost

    Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/57/reference-planes?group=void%3AGND%40GND%3Ahole%2043.84%2C-23.61%7C49.19%2C-23.31 serves each with its own layer_image region; all of them lie in x 43.62..49.39, y -28.93..-22.31 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.

    At void:GND@GND:hole 43.84,-23.61|49.19,-23.31

  3. DecapOrdermajorOpen

    DecapOrder: 5 findings (4 pins, 1 component)×5

    Checked by the review, stands · Claude walker W3 (Cern session) — Checked: the distances are read from placement as stated. This is a layout-practice heuristic (Probe's starting thresholds); no datasheet on file fixes the order of these values at the 1-5 mm scale, where via and plane-spreading inductance usually dominate. Left for the owner.

    At U11, U3A.5, U3B.5, U3C.5, U3D.5

    Each finding, with its own headline
    • U11U11 supply pins B2, C2 on 'VADJ_PIN': the smallest capacitor CP0 sits 4.6 mm away, behind CP1 at 1.1 mm

      U11 supply pins B2, C2 on 'VADJ_PIN': the smallest capacitor CP0 sits 4.6 mm away, behind CP1 at 1.1 mm. U11.B2: the smallest value, CP0 (100 nF), is 4.6 mm from the pin and CP1 (1 uF) 1.1 mm: the high-frequency part sits behind more inductance than the bulk part, which needs it least. U11.C2: the smallest value, CP0 (100 nF), is 4.1 mm from the pin and CP1 (1 uF) 1.3 mm: the high-frequency part sits behind more inductance than the bulk part, which needs it least. Swap them, or bring the smallest value to the pin: it serves the fastest current and is the one the loop inductance defeats. Distances are pad centre to pin centre in the board plane. The 5 mm radius, the 10x value step and the 1 mm / 1.5x tie tolerance are Probe's starting values.

    • U3A.5U3A supply pin 5 on 'VPLMHA': the smallest capacitor C49A sits 3.8 mm away, behind C42A at 1.2 mm

      U3A supply pin 5 on 'VPLMHA': the smallest capacitor C49A sits 3.8 mm away, behind C42A at 1.2 mm. U3A.5: the smallest value, C49A (10 nF), is 3.8 mm from the pin and C42A (100 nF) 1.2 mm: the high-frequency part sits behind more inductance than the bulk part, which needs it least. Swap them, or bring the smallest value to the pin: it serves the fastest current and is the one the loop inductance defeats. Distances are pad centre to pin centre in the board plane. The 5 mm radius, the 10x value step and the 1 mm / 1.5x tie tolerance are Probe's starting values.

    • U3B.5U3B supply pin 5 on 'VPLMHB': the smallest capacitor C49B sits 3.8 mm away, behind C42B at 1.2 mm

      U3B supply pin 5 on 'VPLMHB': the smallest capacitor C49B sits 3.8 mm away, behind C42B at 1.2 mm. U3B.5: the smallest value, C49B (10 nF), is 3.8 mm from the pin and C42B (100 nF) 1.2 mm: the high-frequency part sits behind more inductance than the bulk part, which needs it least. Swap them, or bring the smallest value to the pin: it serves the fastest current and is the one the loop inductance defeats. Distances are pad centre to pin centre in the board plane. The 5 mm radius, the 10x value step and the 1 mm / 1.5x tie tolerance are Probe's starting values.

    • U3C.5U3C supply pin 5 on 'VPLMHC': the smallest capacitor C49C sits 3.8 mm away, behind C42C at 1.2 mm

      U3C supply pin 5 on 'VPLMHC': the smallest capacitor C49C sits 3.8 mm away, behind C42C at 1.2 mm. U3C.5: the smallest value, C49C (10 nF), is 3.8 mm from the pin and C42C (100 nF) 1.2 mm: the high-frequency part sits behind more inductance than the bulk part, which needs it least. Swap them, or bring the smallest value to the pin: it serves the fastest current and is the one the loop inductance defeats. Distances are pad centre to pin centre in the board plane. The 5 mm radius, the 10x value step and the 1 mm / 1.5x tie tolerance are Probe's starting values.

    • U3D.5U3D supply pin 5 on 'VPLMHD': the smallest capacitor C49D sits 3.8 mm away, behind C42D at 1.2 mm

      U3D supply pin 5 on 'VPLMHD': the smallest capacitor C49D sits 3.8 mm away, behind C42D at 1.2 mm. U3D.5: the smallest value, C49D (10 nF), is 3.8 mm from the pin and C42D (100 nF) 1.2 mm: the high-frequency part sits behind more inductance than the bulk part, which needs it least. Swap them, or bring the smallest value to the pin: it serves the fastest current and is the one the loop inductance defeats. Distances are pad centre to pin centre in the board plane. The 5 mm radius, the 10x value step and the 1 mm / 1.5x tie tolerance are Probe's starting values.

  4. DiffPairMismatchmajorOpen

    DiffPairMismatch: 2 nets×2

    Checked by the review, stands · Claude walker (cern session) — LA07 and LA09 are used single-ended, not as differential pairs. LA07_P goes to U15 (2N7001T level converter) pin 4 and thermistor RT2, and LA09_P to U13 pin 4 and RT1; LA07_N and LA09_N reach only the FMC P1 (H14, D15). The asymmetry is real and deliberate, with no matched-routing consequence. Stands as stated.

    At LA07_P, LA09_P

    Each finding, with its own headline
    • LA07_P'LA07_P' / 'LA07_N' (pair inferred from names) is pin-count asymmetric (3 vs 1)

      Differential pair 'LA07_P' / 'LA07_N', INFERRED FROM THE NET NAMES (nothing in the design declares it a pair), is pin-count asymmetric (3 vs 1) - one side reaches a pin the other does not, so the two halves cannot be routed as a matched pair. NOTE: this check does not verify WHICH contact either net reaches; a correctly-symmetric pair can still be swapped at a connector.

    • LA09_P'LA09_P' / 'LA09_N' (pair inferred from names) is pin-count asymmetric (3 vs 1)

      Differential pair 'LA09_P' / 'LA09_N', INFERRED FROM THE NET NAMES (nothing in the design declares it a pair), is pin-count asymmetric (3 vs 1) - one side reaches a pin the other does not, so the two halves cannot be routed as a matched pair. NOTE: this check does not verify WHICH contact either net reaches; a correctly-symmetric pair can still be swapped at a connector.

  5. ReferencePlaneDiscontinuitymajorOpen

    ReferencePlaneDiscontinuity: 2 boards×2

    Checked by the review, stands · Claude walker W3 (Cern session) — Checked: the reference break is read from the copper as stated. Whether it matters depends on the signal's edge rate and the receiver's margin, which no datasheet or app note on file fixes for this net; the impedance step and reflection figures are Probe's model, and the thresholds are its starting values. Layout practice, not a datasheet requirement. Left for the owner.

    At along@GND:GND|XSHIELD, split@GND:GND|XSHIELD

    Each finding, with its own headline
    • along@GND:GND|XSHIELD'CLKIN1N' runs ALONG the split between the GND and XSHIELD planes on GND at 1 place, the longest 0.36 mm; the impedance rises from 49 ohm to 58 ohm there at worst, 8.7 % of each edge reflected (fastest: CLKIN1N at 1 ns 10-90 %, suggested by its name)

      On GND the GND and XSHIELD planes part, and 'CLKIN1N' 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 49 ohm; along the slot 58 ohm, reflecting up to 8.7 % 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 76 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.

      • 'CLKIN1N' on TOP_LAYER at (44.26, -23.42), 0.72 mm, running ALONG the split for 0.36 mm, not across it, 49 ohm over solid plane, 58 ohm along it: 8.7 % of each edge reflected; other side: none: the signal layer is outermost

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/57/reference-planes?group=along%40GND%3AGND%7CXSHIELD serves each with its own layer_image region; all of them lie in x 42.84..45.5, y -24.48..-22.31 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@GND:GND|XSHIELD3 nets cross the split between the GND and XSHIELD planes on GND at 3 places, with no plane on the track's other side (fastest: NetU5_5 at 100 ps 10-90 %, ASSUMED)

      On GND the GND and XSHIELD 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 0 of the 3 crossing(s) the track's other neighbour carries a plane, which the return can use there; 3 are on an outer layer, with no other neighbour. No crossing has a plane on the track's other side, so the whole return meets the break.

      • 'CLKIN2P' on TOP_LAYER at (49.11, -23.48), 0.32 mm; other side: none: the signal layer is outermost
      • 'NetU5_5' on TOP_LAYER at (46.36, -23.46), 0.3 mm; other side: none: the signal layer is outermost
      • 'NetU5_6' on TOP_LAYER at (46.88, -23.45), 0.3 mm; other side: none: the signal layer is outermost

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/57/reference-planes?group=split%40GND%3AGND%7CXSHIELD serves each with its own layer_image region; all of them lie in x 45.36..50.19, y -24.61..-22.31 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.

  6. observationOBS-207minorOpen

    1V8A is set to 1.9 V, above the Si5394 VDD maximum and at the ADS54J54 AVDDC and PLLVDD maximum

    U2 (TPS7A4700) sets 1V8A to 1.9 V: 10 mV above the Si5394 U5 core VDD recommended maximum, and exactly at the ADS54J54 AVDDC and PLLVDD maximum, with no room for regulator tolerance.

    The numbers

    QuantityValueSource
    U2 0P4V (pin 10)GNDnetlist: component U2
    U2 0P1V (pin 12)GND through fitted R1 0Rnetlist: component U2; BOM
    1V8A_INT and 1V8A1.4 + 0.4 + 0.1 = 1.9 Vderived
    ADS54J54 AVDD18typ 1.9 V, suitsADS54J54 datasheet
    ADS54J54 AVDDC1.7 / 1.8 / 1.9 V maxADS54J54 datasheet, p.5
    ADS54J54 PLLVDD, via filter C1101.7 / 1.8 / 1.9 V maxADS54J54 datasheet, p.5
    Si5394 U5 core VDD, via filter C661.71 / 1.80 / 1.89 V maxSi5394 datasheet, p.27

    ``mermaid flowchart LR A["U2 TPS7A4700<br/>1V8A = 1.9 V"] --> B["U7 AVDD18<br/>typ 1.9 V"] A --> C["U7 AVDDC<br/>max 1.9 V"] A -- "C110" --> D["U7 PLLVDD<br/>max 1.9 V"] A -- "C66" --> E["U5 Si5394 VDD<br/>max 1.89 V"] ``

    Why it matters

    The Si5394 core runs 10 mV above its recommended maximum at nominal. AVDDC and PLLVDD sit at their maximum with no room for the regulator tolerance.

  7. observationOBS-208minorOpen

    ADS54J54 SPI enable SDENb has no pull-up and floats until the carrier drives it

    ADS54J54 U7 SPI enable SDENb (pin 4, net LA06_N) has no pull-up and no internal pull, so it floats until the carrier FPGA drives it.

    Where it is

    • LA06_N goes straight to FMC P1.C11, with no resistor on the board.
    • The ADS54J54 pin table lists no internal pull on SDENb, while it does list one on ENABLE and SRESETb (ADS54J54 datasheet, p.4).

    Why it matters

    • Until the carrier FPGA drives LA06_N, the ADC serial interface enable floats.
    • SCLK/SDATA activity or noise during carrier configuration can reach the ADC registers.
    • The Si5394 select on the same bus is safe, because its CSb has an internal ~20 kOhm pull-up (Si5394 datasheet, p.47).
  8. observationOBS-209minorOpen

    U1 TPS7A4700 dissipates about 1 W; its junction margin falls under 20 degC above about 70 degC ambient

    U1 (TPS7A4700, VIM5V to 3V3A) dissipates about 1 W and reaches Tj 119 C at 85 C ambient against a 125 C rating; U2 reaches 108 C. The board states no maximum ambient.

    U1: VIM5V 4.98 V to 3V3A 3.3 V

    QuantityValueSource
    LoadsADS54J54 AVDD33 500 mA + Si5394 VDDA 115 mA typADS54J54 datasheet, p.7; netlist: component U1
    Currentabout 0.615 Aderived
    Dissipation0.615 A × 1.68 V = 1.03 Wderived
    Theta-JA32.5 C/WTPS7A4700 datasheet, p.5
    Rise34 Cderived
    Tj at 85 C ambient119 C, against the 125 C ratingderived

    U2: 3P3V to 1.9 V

    QuantityValueSource
    LoadsAVDD18 320 mA + AVDDC 18 mA + PLLVDD 42 mA + Si5394 VDD 125 mA typ = 0.5 Anetlist: component U2
    Dissipation0.71 Wderived
    Rise23 Cderived
    Tj at 85 C ambient108 Cderived

    Why it matters

    The junction margin of U1 falls under 20 C above about 70 C ambient, and the board states no maximum ambient.

  9. observationOBS-210minorOpen

    No probe access on the analog supply rails

    No probe access on the analog supply rails: the board has no test points, so the generated rails a bring-up checks first can only be reached on capacitor pads.

    • The only probe access is an unfitted 9-hole row labelled 3P3V, GND, VADJ, GND and the SPI signals.
    • Rails without access: VIM5V, VPOS5V, VNEG2V64, 4V22, -1V02, 3V3A, 1V8A.
  10. observationOBS-211minorOpen

    Silkscreen designators are about 0.65 mm high, under the 0.8 mm printable floor

    Silkscreen designators are about 0.63-0.66 mm high, under the roughly 0.8 mm height that screen printing resolves reliably.

    LayerGlyph outlinesOf which 0.63-0.66 mm tall
    TOP_OVERLAY638538
    BOTTOM_OVERLAY822762

    About 1,300 of the 1,460 glyph outlines. Examples: R35, C122 and U10 on the top, and the channel groups on the bottom.

  11. observationOBS-212minorOpen

    LMZM23601 U01 MODE/SYNC floats until the carrier drives it; the datasheet says do not float this pin

    LMZM23601 U01 MODE/SYNC (pin 2, net DCDCSYNC2) floats until the carrier drives it, and forever if the carrier never does; the datasheet says "Do not float this pin."

    • DCDCSYNC2 goes only to FMC P1.H38, with no resistor on the board.
    • U01 starts as soon as the carrier applies 12P0V (EN is tied to VIN), before any carrier gateware drives the pin.
    • The datasheet requires the pin to be tied low (auto PFM), high (forced PWM) or clocked (LMZM23601 datasheet, p.4).
    • The ADP5071 SYNC input on the same board (DCDCSYNC, U8) has a 4.7 kOhm pull-down (R34), so the LMZM23601 was probably meant to have one too.
  12. ReferencePlaneDiscontinuityminorOpen

    ReferencePlaneDiscontinuity: 4 boards×4

    Checked by the review, stands · Claude walker W3 (Cern session) — Checked: the reference break is read from the copper as stated. Whether it matters depends on the signal's edge rate and the receiver's margin, which no datasheet or app note on file fixes for this net; the impedance step and reflection figures are Probe's model, and the thresholds are its starting values. Layout practice, not a datasheet requirement. Left for the owner.

    At refchange:1V8A|GND, refchange:4V22|GND, split@PWR:1V8A|TGND, split@PWR:4V22|TGND

    Each finding, with its own headline
    • refchange:1V8A|GND3 nets change reference between 1V8A and GND at 3 vias, each with another reference that continues across (fastest: LA04_N at 100 ps 10-90 %, ASSUMED)

      At 3 via(s), on 3 net(s), the signal leaves a layer referenced by one of 1V8A and GND 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 1V8A and GND 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.

      • 'LA05_N' at (54.59, -20.72), PWRSIG -> BOTTOM_LAYER: GND on GND, 1V8A on PWR to GND on GND_2; partial, GND continues through a via of GND 3.74 mm away; stitching none, 57.9 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. 1V8A and GND 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.
      • 'LA04_N' at (54.59, -21.52), PWRSIG -> BOTTOM_LAYER: GND on GND, 1V8A on PWR to GND on GND_2; partial, GND continues through a via of GND 3.51 mm away; stitching none, 53.0 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. 1V8A and GND 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.
      • 'LA04_P' at (54.59, -22.41), PWRSIG -> BOTTOM_LAYER: GND on GND, 1V8A on PWR to GND on GND_2; partial, GND continues through a via of GND 2.71 mm away; stitching none, 47.7 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. 1V8A and GND 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/57/reference-planes?group=refchange%3A1V8A%7CGND serves each with its own layer_image region; all of them lie in x 52.59..56.59, y -24.41..-18.72 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:4V22|GND'NetC58A_1' changes reference between 4V22 and GND at 4 vias, each with another reference that continues across (fastest: NetC58A_1 at 100 ps 10-90 %, ASSUMED)

      At 4 via(s), on 1 net(s), the signal leaves a layer referenced by one of 4V22 and GND 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: 4 with none. 4 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 4V22 and GND 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.

      • 'NetC58A_1' at (24.48, -11.66), PWRSIG -> BOTTOM_LAYER: GND on GND, 4V22 on PWR to GND on GND_2; partial, GND continues through a via of GND 2.31 mm away; stitching none, 10.5 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. 4V22 and GND 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.
      • 'NetC58A_1' at (24.48, -18.96), PWRSIG -> BOTTOM_LAYER: GND on GND, 4V22 on PWR to GND on GND_2; partial, GND continues through a via of GND 2.3 mm away; stitching none, 10.5 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. 4V22 and GND 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.
      • 'NetC58A_1' at (24.48, 3.44), PWRSIG -> BOTTOM_LAYER: GND on GND, 4V22 on PWR to GND on GND_2; partial, GND continues through a via of GND 2.25 mm away; stitching none, 10.5 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. 4V22 and GND 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.
      • 'NetC58A_1' at (24.48, -3.86), PWRSIG -> BOTTOM_LAYER: GND on GND, 4V22 on PWR to GND on GND_2; partial, GND continues through a via of GND 2.14 mm away; stitching none, 10.5 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. 4V22 and GND 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/57/reference-planes?group=refchange%3A4V22%7CGND serves each with its own layer_image region; all of them lie in x 22.48..26.48, y -20.96..5.44 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@PWR:1V8A|TGND5 nets cross the split between the 1V8A and TGND planes on PWR at 5 places, all with a plane on the track's other side (fastest: LA04_N at 100 ps 10-90 %, ASSUMED)

      On PWR the 1V8A and TGND planes part, and 5 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 45% 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).

      • 'CS' on PWRSIG at (54.61, -23.46), 0.42 mm; other side: GND on GND, 0.2 mm from the track against 0.24 mm to PWR: 45% of the return on PWR
      • 'LA04_N' on PWRSIG at (55.25, -22.18), 0.42 mm; other side: GND on GND, 0.2 mm from the track against 0.24 mm to PWR: 45% of the return on PWR
      • 'LA04_P' on PWRSIG at (55.25, -23.06), 0.42 mm; other side: GND on GND, 0.2 mm from the track against 0.24 mm to PWR: 45% of the return on PWR
      • 'LA05_N' on PWRSIG at (55.25, -21.38), 0.42 mm; other side: GND on GND, 0.2 mm from the track against 0.24 mm to PWR: 45% of the return on PWR
      • 'NetC58A_1' on PWRSIG at (38.74, -3.45), 0.3 mm; other side: GND on GND, 0.2 mm from the track against 0.24 mm to PWR: 45% of the return on PWR

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/57/reference-planes?group=split%40PWR%3A1V8A%7CTGND serves each with its own layer_image region; all of them lie in x 37.59..56.4, y -24.61..-2.45 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.

    • split@PWR:4V22|TGND'NetC58A_1' crosses the split between the 4V22 and TGND planes on PWR at 2 places, all with a plane on the track's other side (fastest: NetC58A_1 at 100 ps 10-90 %, ASSUMED)

      On PWR the 4V22 and TGND planes part, and 'NetC58A_1' 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 45% 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).

      • 'NetC58A_1' on PWRSIG at (35.89, -16.86), 0.3 mm; other side: GND on GND, 0.2 mm from the track against 0.24 mm to PWR: 45% of the return on PWR
      • 'NetC58A_1' on PWRSIG at (35.89, -1.66), 0.3 mm; other side: GND on GND, 0.2 mm from the track against 0.24 mm to PWR: 45% of the return on PWR

      Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/57/reference-planes?group=split%40PWR%3A4V22%7CTGND serves each with its own layer_image region; all of them lie in x 34.74..37.04, y -17.86..-0.66 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
24LC64T-I/MNYMicrochip64Kbit, 400kHz, 2.5V, I2C Serial EEPROM, 8-Pin TDFN (NiPdAu finish), Industrial Temperature, Tape and ReelU0
ADP5071Analog Devices2 A/1.2 A DC-to-DC Switching Regulator with Independent Positive and Negative OutputsU8
ADS54J54Texas InstrumentsADS54J54 Quad Channel 14-Bit 500 MSPS ADCU7
LMZM23601Texas InstrumentsLMZM23601, 36-V, 1-A Step-Down DC/DC Power Module in 3.8-mm x 3-mm PackageU01
LT3045IDDLinear Technology20V, 500mA, Ultralow Noise, Ultrahigh PSRR Linear RegulatorU9
LT3094IDDLinear Technology-20V, 500mA, Ultralow Noise, Ultrahigh PSRR Negative Linear RegulatorU10
MC-HPC-10SamtecMC-HPC-10, FMC High-pin count connector, lead free, 400 I/Os, male, 10 mm mated stack heightP1
Si53340Silicon LabsLow-Jitter LVDS Fanout Clock Buffer, 2:1 selectable MUX, 4 OutputsU6
SI5394Silicon LabsClock Generator 0.008MHz to 750MHz-IN 1028MHz-OUT 44-Pin QFN EP TrayU5
TLV4031R1Texas InstrumentsTLV40x1Small-Size, Low-Power Comparator with Precision ThresholdU12
TPS22970YZPRTexas InstrumentsTPS22970, 3.6-V, 4-A, 4.7-mO On-Resistance Load SwitchU11
2N7001TDCKTexas Instruments2N7001T, Single-Bit, Dual-Supply Buffered Voltage Signal ConverterU13, U15
DFLS240L-7Diodes IncLow VF Schottky Barrier Rectifier, 2 A, 40 V, -55 to 125 degC, 2-Pin SMD, RoHS, Tape and ReelD2, D3
DS90LV011ATMF/NOPBTexas InstrumentsSingle High Speed Differential Driver, 5-pin SOT-23, Pb-FreeU14, U16
TPS7A4700RGWTTexas InstrumentsRF LDO Voltage Regulator, 1 A, 1.4 to 20.5 V, -40 to 125 degC, 20-pin VQFN (RGW), Green (RoHS & no Sb/Br), Tape and ReelU1, U2
LMH5401Texas InstrumentsLMH5401, 8-GHz, Low-Noise, Low-Power, Fully-Differential AmplifierU4A, U4B, U4C, U4D
MMBD352WT1GOn SemiconductorDual Schottky Barrier Diodes, 3-Pin SOT-323, Pb-Free, Tape and ReelD1A, D1B, D1C, D1D
OPA376AIDCKTTexas InstrumentsPrecision, Low Noise, Low Iq Operational Amplifier, 2.2 to 5.5 V, -40 to 125 degC, 5-pin SOT23 (DCK5), Green (RoHS & no Sb/Br)U3A, U3B, U3C, U3D
CONSMA002-L-GLinxSMA Female Right Angle PCB Mount Connector - ExtendedJ1, J2, J3, J4, J5
150060GS75000Wuerth ElektronikSMD mono-color Chip LED, WL-SMCW, GreenLED1, LED2, LED3, LED4, LED5, LED6
MMBT3904LT1GOn SemiconductorGeneral Purpose Transistor, NPN Silicon, 3-Pin SOT-23, Pb-Free, Tape and ReelQ1, Q2, Q3, Q4, Q5, Q6
18K7Vishay18K7 0.063W 1% 0402 (1005 Metric) SMDRI1
C0603C473K4RACTUKemetCap Ceramic 0.047uF 16V X7R 10% SMD 0603C114, C123
GJM1555C1H4R3WB01DMurataCAP 0402, 4.3pF 50 V, C0G, -0.05pFC60A, C60B, C60C, C60D
3K32Vishay3K32 0.063W 1% 0402 (1005 Metric) SMDRP1, RT1, RT2, RTA, RTB
CRCW0402100RFKEDVishay DaleRes Thick Film 0402 100 Ohm 1% 1/16W ±100ppm/°CRL1, RL2, RL3, RL4, RL5, RL6
GJM1555C1H2R0WB01DMurataCAP 0402, 2.0pF 50 V, C0G, -0.05pFC59A, C59B, C59C, C59D, C61A, C61B, C61C, C61D
RC0402FR-0749R9LYageoRes 0402 49.9 Ohm 1% 1/16W ±100ppmR13, R14, R15, R16, R19, R21, R22, R23
RN73C1E100RBTDFTE Connectivity / AMPResistor 0402 100R 0.1% 10PPMR17, R18, R20, R24, R9A, R9B, R9C, R9D
CRCW040240R2FKEDVishay DaleRes Thick Film 0402 40.2 Ohm 1% 1/16W ±100ppm/°CR10A, R10B, R10C, R10D, R7A, R7B and 6 more
GRM188R71A105KA61DKemet0603 1 uF 10 V ±10% Tolerance X7R Multilayer Ceramic Chip CapacitorC116, C119, C122, C126, C21, C27 and 26 more

About this design

Design
FMC ADC 500M14B4CHA — OHWR0056 R1
Designed in
Altium Designer
Board
8 copper layers, 84 × 69 mm, 402 components, 255 nets
Licence
CERN-OHL-S-2.0
Source
https://gitlab.com/ohwr/project/fmc-adc-500m14b4cha at commit b9b019991576
Attribution
CERN Open Hardware Repository, project ohwr/project/fmc-adc-500m14b4cha (commit b9b0199915764114457813d61f853163706f2918, 2025-04-10). Licence: CERN-OHL-S-2.0. Source: https://gitlab.com/ohwr/project/fmc-adc-500m14b4cha. This is a snapshot of the copy published on the OHWR GitLab pages (R1), not CERN's released record. CERN's source of truth is EDMS (https://edms.cern.ch), and the current release may be newer.
Reviewed with
Probe by E# (E-Sharp AB)