RATOPUS STAGE PFC BUCK ACDC
OHWR0049 · revision R1
RATOPUS STAGE PFC BUCK ACDC (OHWR0049 R1) is an open-hardware circuit board from the CERN Open Hardware Repository, designed in Altium Designer: a 4-layer board of 160 × 100 mm with 101 components and 55 nets, published under CERN-OHL-W-2.0.
review 96% walked · 10 findings: 5 major · 5 minor (13 items)
Open in the interactive viewerRead the full review

Design review
Probe recorded 10 findings: 5 major · 5 minor (13 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.
Mains bridge D1: BOM orders GBU6B (100 V) while the schematic and the BOM value say GBU6J (600 V)
The input bridge rectifier D1 of this 230 VAC stage is ordered as GBU6B, a 100 V part, while its BOM value and the schematic both say GBU6J (600 V). On a 230 VAC line the bridge sees about 325 V peak (more with line tolerance and surges clamped only at the 250 VAC varistor RV1); a GBU6B fitted from the MPN column would break down at first power-up.
Item Value Source D1 BOM row MPN GBU6B (FAIRCHILD), value GBU6J, 'Bridge Rectifier' BOM Schematic D1 GBU6J schematic sheet 1 GBU6B V_RRM / max RMS input 100 V / 70 V onsemi GBU6A-GBU6M data sheet p2 GBU6J V_RRM / max RMS input 600 V / 420 V same table Bridge input line / neut from J3 via F1, L4 (common-mode choke), RV1 250 VAC netlist 230 VAC peak about 325 V (about 360 V at +10 %) computed The BOM's MPN column should read GBU6J (or a higher grade).
- DecapViaFanoutmajorOpen
C21 on 'Vcc': its rail pad reaches its via over 2.00 mm of track, adding about 0.63 nH to its loop, past Probe's 0.5 nH starting value; put the vias at the pads
C21 on 'Vcc': its rail pad reaches its via over 2.00 mm of track, adding about 0.63 nH to its loop, past Probe's 0.5 nH starting value; put the vias at the pads. C21.2 (rail) runs 2.00 mm along its copper to its nearest via, 1.00 mm past the 1 mm beside the pad: 0.63 nH; C21.1 (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): 0.63 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 1.23 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 0.38 mm depth, ASSUMED). 0.5 nH is Probe's starting value, about the whole mounted inductance of a well-mounted 0402.
At C21
- PartLifecyclemajorOpen
PartLifecycle: 2 part numbers×2
2 Obsolete · on 3 placements
At C1, L1
Each finding, with its own headline
C1DE2F3KY103MA3BM02F on 2 placements (C1, C2): the distributor catalogue reports "Obsolete" for Murata Electronics's DE2F3KY103MA3BM02F
DE2F3KY103MA3BM02F on 2 placements (C1, C2): the distributor catalogue reports "Obsolete" for Murata Electronics's DE2F3KY103MA3BM02F — a sourcing/longevity risk to confirm against the MANUFACTURER before acting. That is the maker this board's BOM names (MURATA). 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/DE2F3KY103MA3BM02F/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 (2024-03-31, 959 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.
L1B82144B2473J000 on L1: the distributor catalogue reports "Obsolete" for TDK's B82144B2473J000
B82144B2473J000 on L1: the distributor catalogue reports "Obsolete" for TDK's B82144B2473J000 — a sourcing/longevity risk to confirm against the MANUFACTURER before acting. That is the maker this board's BOM names (TDK). 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/B82144B2473J000/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 384 left. No approved alternate is recorded for this position.
- ReferencePlaneDiscontinuitymajorOpen
ReferencePlaneDiscontinuity: 3 boards×3
Checked by the review, stands · Claude walker W1 (Cern session) — Checked: splits and voids read from the copper stand; they sit on the slow pwm and bias nets of a PFC controller, and no datasheet or app note on file sets a limit. Left for the owner.
At along@PLANE_1:GND_A|Vdc, split@PLANE_1:GND_A|Vcc, split@PLANE_1:Vcc|Vdc
Each finding, with its own headline
along@PLANE_1:GND_A|Vdc'pwm' runs ALONG the split between the GND_A and Vdc planes on PLANE_1 at 1 place, the longest 5.09 mm; the impedance rises from 28 ohm to 36 ohm there at worst, 12.6 % of each edge reflected (fastest: pwm at 100 ps 10-90 %, ASSUMED)
On PLANE_1 the GND_A and Vdc planes part, and 'pwm' 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 28 ohm; along the slot 36 ohm, reflecting up to 12.6 % 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 36 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.
- 'pwm' on TOP_LAYER at (93, 83.47), 8.48 mm, running ALONG the split for 5.09 mm, not across it, 28 ohm over solid plane, 36 ohm along it: 12.6 % 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/50/reference-planes?group=along%40PLANE_1%3AGND_A%7CVdc serves each with its own layer_image region; all of them lie in x 92..95.75, y 79.78..90.8 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.
split@PLANE_1:GND_A|Vcc'pwm' crosses the split between the GND_A and Vcc planes on PLANE_1 at 1 place, with no plane on the track's other side (fastest: pwm at 100 ps 10-90 %, ASSUMED)
On PLANE_1 the GND_A and Vcc planes part, and 'pwm' crosses from one to the other at 1 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 1 crossing(s) the track's other neighbour carries a plane, which the return can use there; 1 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.
- 'pwm' on TOP_LAYER at (93.68, 72.07), 0.79 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/50/reference-planes?group=split%40PLANE_1%3AGND_A%7CVcc serves each with its own layer_image region; all of them lie in x 92.5..95.14, y 70.89..73.25 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@PLANE_1:Vcc|Vdc'NetC11_1' crosses the split between the Vcc and Vdc planes on PLANE_1 at 1 place, with no plane on the track's other side (fastest: NetC11_1 at 100 ps 10-90 %, ASSUMED)
On PLANE_1 the Vcc and Vdc planes part, and 'NetC11_1' crosses from one to the other at 1 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 1 crossing(s) the track's other neighbour carries a plane, which the return can use there; 1 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.
- 'NetC11_1' on TOP_LAYER at (104.45, 80.95), 6.25 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/50/reference-planes?group=split%40PLANE_1%3AVcc%7CVdc serves each with its own layer_image region; all of them lie in x 103.45..108.25, y 77.25..83.5 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.
- FiducialMissingmajorOpen
The bottom side carries 21 SMT component(s) and no fiducial was found on it
The bottom side carries 21 SMT component(s) but no fiducial was found on it — a placement machine has no optical reference to align to. Probe found no pad labelled .pad_usage=g_fiducial/l_fiducial (basis pad_usage); no single-pad, netless, undrilled component whose 0.5-3 mm round pad has a concentric solder-mask opening at least 1.5x its diameter with no other copper inside it (basis geometry); and no component whose refdes starts FID or whose part or package name contains FIDUCIAL (basis footprint-name). The other side has 3, found by geometry (3). If your library names fiducials something Probe does not recognise, mark this benign — disposition_finding(rule='FiducialMissing', entity='bottom', status='benign', reason=…) — the name match only ever silences this check, it never asserts anything. If the side really has no fiducial, say so with status='confirmed': that answers the finding and keeps it gating release.
At bottom
IC1 (TL2843B) start-up supply reaches only about 8.5 V at worst-case tolerance, below the 9.0 V maximum start threshold
At worst-case tolerances the PFC buck controller IC1's start-up supply reaches about 8.5 V, below the TL2843B's 9.0 V maximum start threshold, so a unit at that corner may sit in UVLO and never start. Typical margin is about 1 V. The same start-up circuit on OHWR0048 RATOPUS POWER is OBS-87.
The start-up path
Before IC1 switches, its VCC comes only from the bootstrap source follower T3; the auxiliary-winding bias (D8, T6, D10, D9) works only after start-up (schematic note 'after startup, Operates when Vdc>40V'). The 'Vcc' jumper header selects Vcc_int (or an external supply on J4).
``
mermaid flowchart LR Z["D6 BZV55-B13 on T3 gate (NetC16_2)"] --> T3["T3 IPA80R280P7 source follower"] --> Vcc1 --> D5["D5 PMLL4148L"] --> Vcc_int --> JMP["jumper 'Vcc'"] --> Vcc["IC1 VCC pin 7"]``Vcc = Vz(D6) - Vgs(T3) - Vf(D5)
Tolerance stack
Term Typical Worst case Source Vz, D6 BZV55-B13 13.0 V 12.7 V min BZV55 data sheet p5 Vgs(th), T3 3.0 V 3.5 V max IPA80R280P7 data sheet p4 Vf, D5 about 0.6 V 0.72 V max at 5 mA PMLL4148L data sheet p3 Vcc at start about 9.4 V about 8.5 V TLx843B start threshold 8.4 V 9.0 V max TL284xB data sheet p7 A higher zener voltage, or removing the D5 drop from the start-up path, restores margin.
Mains varistor RV1 (S14K250) is rated 250 V RMS continuous, below the 253 V RMS top of a 230 V +10 % line
The mains varistor RV1 (S14K250) is rated 250 V RMS continuous, below the 253 V RMS top of the +/-10 % band on the board's 230 V RMS input. Run above that rating it leaks, heats, ages faster and can fail short, blowing F1.
Where it is
RV1 (B72214S0251K101 / S14K250) sits across line and neut after F1 and the common-mode choke L4, ahead of bridge D1 (schematic sheet 1).
The numbers
Quantity Value Source Stated input 230 V RMS nominal (RaToPUS 230 V AC input on J3) RaToPUS design (psu-rad-acdc-230v) Mains tolerance +/-10 % IEC 60038 / EN 50160 Top of the band 230 V x 1.1 = 253 V RMS derived RV1 maximum continuous operating voltage 250 V RMS (320 V DC) TDK SIOV datasheet, S14K250 row, p.9 Why it matters
A varistor run above its maximum continuous voltage leaks more, heats and ages faster, and in the limit fails short, which here would blow F1.
Fix: the next step up in the same family (S14K275, 275 V RMS) is the usual choice for 230 V lines. If the input is known to stay within 250 V RMS, record that and this can be dispositioned.
The same varistor on OHWR0048 RATOPUS POWER is OBS-79.
No fiducials on the bottom side, which carries 21 SMT parts
OHWR0049 R1 RATOPUS STAGE PFC BUCK ACDC has three fiducials on the top (FTG1-FTG3) and none on the bottom, although the bottom carries 21 SMT placements (for example D8, D9, D10, T6). A bottom-side reflow pass has no optical reference unless the panel provides one. The same gap on OHWR0048 RATOPUS POWER is OBS-83.
Side SMT placements Fiducials Source Top 47 (+3 fiducials) FTG1 (154.2, 97.8), FTG2 (5.8, 97.8), FTG3 (154.0, 2.3) ODB++ placement Bottom 21 none ODB++ placement - ReferencePlaneDiscontinuityminorOpen
2 nets change reference between GND_C and Vcc at 4 vias, none with a reference that continues across (fastest: comp at 100 ps 10-90 %, ASSUMED)
At 4 via(s), on 2 net(s), the signal leaves a layer referenced by one of GND_C and Vcc 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 served. Every site is served, so the finding is the lower tier (info). 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 GND_C and Vcc 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. None is partial: at every one, no reference continues across the via.
- 'rtct' at (100.5, 60), TOP_LAYER -> BOTTOM_LAYER: Vcc on PLANE_1 to GND_C on PLANE_2; pure, no reference continues across; stitching served, 15.1 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_C on PLANE_2 and Vcc on PLANE_1 are adjacent, 0.127 mm apart: their own capacitance carries the return. Served.
- 'rtct' at (105, 73.25), TOP_LAYER -> BOTTOM_LAYER: Vcc on PLANE_1 to GND_C on PLANE_2; pure, no reference continues across; stitching served, 15.1 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_C on PLANE_2 and Vcc on PLANE_1 are adjacent, 0.127 mm apart: their own capacitance carries the return. Served.
- 'comp' at (104.5, 68), TOP_LAYER -> BOTTOM_LAYER: Vcc on PLANE_1 to GND_C on PLANE_2; pure, no reference continues across; stitching served, 6.5 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. GND_C on PLANE_2 and Vcc on PLANE_1 are adjacent, 0.127 mm apart: their own capacitance carries the return. Served.
- 'comp' at (110, 65.5), TOP_LAYER -> BOTTOM_LAYER: Vcc on PLANE_1 to GND_C on PLANE_2; pure, no reference continues across; stitching served, 6.5 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. GND_C on PLANE_2 and Vcc on PLANE_1 are adjacent, 0.127 mm apart: their own capacitance carries the return. Served.
Every site is a row of this finding's evidence, keyed by net, layers and position. GET /api/odb/boards/50/reference-planes?group=refchange%3AGND_C%7CVcc serves each with its own layer_image region; all of them lie in x 98.5..112, y 58..75.25 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.
At refchange:GND_C|Vcc
- ConnectorDomainMixedminorOpen
J5 takes 2 different supplies off the board on one connector: Vcc (10.0 V), ref (5.0 V)
J5 carries 2 different supplies out of the board on one connector: Vcc (10.0 V) on 3, ref (5.0 V) on 4. Declare the intent. Two rails on one connector is ordinary in a docking or power-distribution interface and a hazard in a signal one, and the difference is not visible from the netlist — but note that adjacent contacts at different voltages are exactly what a mirrored, reversed or off-by-one cable shorts together, so the cost of getting the mating wrong is a rail collision rather than a dead signal. If this is a standard connector, declare its family (connector_family_declare) and file the standard's contact map: supplies on the contacts it assigns to power are then not reported.
At J5
Main parts
| Part number | Maker | Description | Where |
|---|---|---|---|
| 1755736 (MSTBVA 2,5/2-G-5,08) | PHOENIX CONTACT | Connector 2 Male | J1 |
| 1755749 (MSTBVA 2,5/3-G-5,08) | PHOENIX CONTACT | Connector 3 Male | J3 |
| 51939-667 | AMPHENOL | Thru-Hole, Lead 3.43mm, PowerBlade Right Angled Power Connector 3ACP 2P 24S 4P Contacts | J2 |
| B72214S0251K101 or S14K250 | EPCOS / TDK | Varistor | RV1 |
| BZV55-B12 | NEXPERIA | Zener Diode | D10 |
| BZV55-B13 | NEXPERIA | Zener Diode | D6 |
| GBU6B | FAIRCHILD SEMICONDUCTOR | Bridge Rectifier 1+ 2AC 3AC 4- | D1 |
| HTSW-101-23-L-D | SAMTEC | Connector 2 Male | J4 |
| MTLW-104-07-L-D-250 | SAMTEC | Connector 8 Male | J5 |
| SF-3812TL100T-2 | BOURNS | Fuse | F1 |
| SMBJ12A | FAIRCHILD SEMICONDUCTOR | Zener Diode | D7 |
| TL2843BDR-8 | TEXAS INSTRUMENTS | High-Performance Current-Mode PWM Controllers, -40°C to 85°C | IC1 |
| ES1D | FAIRCHILD SEMICONDUCTOR | Diode | D2, D8 |
| FMMT597 | DIODES INCORPORATED | PNP Transistor | T2, T4 |
| IPA80R280P7 | VISHAY SILICONIX | N-CHANNEL MOSFET WITH DIODE | T1, T3 |
| PMLL4148L | NXP SEMICONDUCTORS | Diode | D5, D9 |
| STPSC10H065DI | ST MICROELECTRONICS | Schottky Diode | D3, D4 |
| BC817-25 | NXP SEMICONDUCTORS | NPN Transistor | T5, T6, T7 |
| B32923C3105M000 | EPCOS | Non-polarised capacitor | C4 |
| B82144B2473J000 | TDK | Inductor With Core | L1 |
| B82721K2122N020 | TDK | Inductor Common Mode | L4 |
| BFC233920224 | VISHAY BCCOMPONENTS | Non-polarised capacitor | C3 |
| CC0805_100NF_50V_10%_X7R | GENERIC | Non-polarised capacitor | C15 |
| CC0805_10NF_50V_10%_X7R | GENERIC | Non-polarised capacitor | C20 |
| CC0805_22NF_50V_10%_X7R | GENERIC | Non-polarised capacitor | C14 |
| KP1830122631 | VISHAY ROEDERSTEIN | Non-polarised capacitor | C11 |
| R0805_0R_JUMPER | GENERIC | Resistor | R17 |
| R0805_100R_1%_0.125W_100PPM | GENERIC | Resistor - 1% | R16 |
| R0805_150K_1%_0.125W_100PPM | GENERIC | Resistor - 1% | R30 |
| R0805_15K_1%_0.125W_100PPM | GENERIC | Resistor - 1% | R34 |
| R0805_1M_1%_0.125W_100PPM | GENERIC | Resistor - 1% | R23 |
| R0805_220R_1%_0.125W_100PPM | GENERIC | Resistor - 1% | R36 |
| R0805_300R_1%_0.125W_100PPM | GENERIC | Resistor - 1% | R37 |
| R0805_33K_1%_0.125W_100PPM | GENERIC | Resistor - 1% | R32 |
| R0805_4K7_1%_0.125W_100PPM | GENERIC | Resistor - 1% | R35 |
| R0805_5K1_1%_0.125W_100PPM | GENERIC | Resistor - 1% | R26 |
| R0805_7K15_0.1%_0.1W_10PPM | GENERIC | Resistor - 0.1% | R33 |
| R0805_8K2_1%_0.125W_100PPM | GENERIC | Resistor - 1% | R31 |
| R1206_8R2_1%_0.25W_200PPM | GENERIC | Resistor - 1% | R1 |
| 7447070 | WURTH ELEKTRONIK | Inductor | L2, L3 |
About this design
- Design
- RATOPUS STAGE PFC BUCK ACDC — OHWR0049 R1
- Designed in
- Altium Designer
- Board
- 4 copper layers, 160 × 100 mm, 101 components, 55 nets
- Licence
- CERN-OHL-W-2.0
- Source
- https://gitlab.com/ohwr/project/psu-rad-acdc-230v-12v5v-110w at commit
aa0f8b805fdf - Attribution
- CERN Open Hardware Repository, project ohwr/project/psu-rad-acdc-230v-12v5v-110w (commit aa0f8b805fdf947458e09c4c64a58de31cbf2940, 2025-04-10). Licence: CERN-OHL-W-2.0. Source: https://gitlab.com/ohwr/project/psu-rad-acdc-230v-12v5v-110w. 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)