Skip to the content
2 layers155.5 × 65.9 mmPicozorro one th, the printed circuit board in 3D

Picozorro one th

maximevince/PicoZorro2

Amiga Zorro-II Bus Cards using nothing but an RPi Pico to interact with the bus

Size
155.5 × 65.9 mm
Copper layers
2
Thickness
1.6 mm
Parts
48
Nets
125
Pads
307
Tracks, vias
546, 286
Pours
2
Routed
Yes
KiCad
KiCad 10.0 (file 20260206)
Created
6 Oct 2026
Updated
19 Oct 2026

Overview

Picozorro one th is an open-source printed circuit board by maximevince, from the maximevince/PicoZorro repository on GitHub. It is a 2-layer board of 155.5 × 65.9 mm with 48 parts on 125 nets, routed with 546 tracks and 286 vias.

It is built around the AMS1117-3.3, LD1117V33 and CL31A226KAHNNNE, with the CC0603KRX7R9BB104, CL21A106KAYNNNE and MMBT3904 among its other parts. Its parts include 16 capacitors, 10 connectors, 7 resistors, 2 ICs and 2 diodes.

Main components

9

README

hardware/picozorro-one-th/README.md

PicoZorro One TH

The PicoZorro One is the Zorro II network and USB card of the PicoZorro project. The One TH is its through-hole version: bought modules and a handful of through-hole parts on a bare two-layer PCB. Every through-hole part also has an SMD alternate on the same board, so the same PCB can be assembled by JLCPCB instead (see "Two builds" below).

  • RP2350B: a Waveshare Core2350B0 module (B0 variant, no PSRAM) on four 2x8 pin strips. Its PIO state machines speak the Zorro II bus directly.
  • Ethernet: a W5500 "Lite" module (USR-ES1, WIZnet WIZ850io pinout) on SPI1.
  • USB host: the RP2350's own USB port, to a type A receptacle or a PC-style 2x5 internal header.
  • Optional: microSD on SPI1 (a breakout, or a push-push socket on the Partial SMD build) and a GY-PCM5102 I2S DAC breakout.

3D render of the component side, Full TH build

Status

Rev A. The schematic and the routed board pass KiCad ERC and DRC with zero violations, DRC with schematic parity (make check-one-th at the top of the repository). Rev A boards of the Partial SMD build are built and run in an Amiga: network, USB host and the microSD block device. The DAC does not play yet, and the Full TH build has not been built. The open points are listed at the end of this file.

Two builds: Full TH and Partial SMD

One PCB, two builds. The Full TH build (KiCad's default variant) fits every part through-hole, soldered by hand. The Partial SMD build (KiCad variant Partial SMD) has JLCPCB place an SMD set, almost all JLC Basic parts, and optionally solder the USB-A socket J2; the builder then solders only the modules on their headers (and J2 or J3 if JLC did not).

Each through-hole part and its SMD alternate sit on the same nets and share the footprint area; the two are never both fitted. The SMD alternate of a part has the part's number times 100 plus 1, 2, ... as its reference (D101 is D1's, C101-C103 are C1's bank), because KiCad cannot give one symbol two footprints. The board hides those references; silk frames group each pair under one label, and the JLC files name them "D1 (SMD)".

3D render of the component side, Partial SMD build

Files

PathWhat
picozorro-one-th.kicad_pro, .kicad_sch, .kicad_pcb, .kicad_druKiCad 10 project with the variant Partial SMD; the edge-connector and microSD socket footprints come from ../lib
fab/gerbers/, fab/picozorro-one-th-gerbers.zipGerber and Excellon drill files, the same for both builds
fab/picozorro-one-th-parts.csvparts list of the Full TH build (the SMD alternates show DNP)
fab/jlc/picozorro-one-th-bom-jlc.csv, fab/jlc/picozorro-one-th-cpl-jlc.csvJLC BOM (the SMD set with LCSC numbers; the through-hole alternates as DNP) and placement file (SMD set and J2, rotations corrected for JLC)
fab/order-notes.txtthe options to pick in the order form, and what the builder solders in each build
3d/STEP models of the modules and the PTC that the board references; origins and licences in 3d/SOURCES.txt
images/schematic (PDF, PNG), copper and silkscreen views, 3D renders: pcb-3d-th.png (Full TH build), pcb-3d-jlc.png (Partial SMD build, same view), pcb-3d-iso.png (Full TH build, isometric)

Schematic

What the builder solders

Full TH build, everything by hand. The last column names the SMD alternate the Partial SMD build fits instead (LCSC part number in brackets; JLC Basic parts unless marked Extended).

RefPartValue / typePackageFitted (Full TH)Partial SMD build alternate
U1 (P1-P4)Waveshare Core2350B0 module (no PSRAM)-four 2x8 2.54 mm pin stripsyessame, by hand
M1W5500 "Lite" module (USR-ES1, WIZ850io pinout)-two 1x6 2.54 mm rowsyessame, by hand
M2microSD breakout, 3.3 V, no level shifter-one 1x6 2.54 mm rowoptionalJ6, SHOU HAN TF PUSH push-push socket (C393941, Extended), with R2-R5 and C5, C6
M3PCM5102A DAC breakout "GY-PCM5102"-1x6 row (signals) + 1x9 row (holding only)optionalsame, by hand (optional)
D1Schottky diode1N5817DO-41, 10.16 mmyesD101, SS14, SMA (C2480)
F1resettable PTC fuse, 30 V (or 16 V) family0.9 A hold, e.g. Littelfuse RUEF090radial, 5.08 mm leadsyesF101, 1.1 A hold, 1812, Brightking SMD1812B110TF/33 (C269134, Extended)
C1electrolytic capacitor, bus +5 V100 uFradial, D6.3 mm, 2.5 mm pitchyesC101-C103, 3 x 22 uF 25 V X5R 1206 (C12891)
C2electrolytic capacitor, USB port220 uFradial, D8 mm, 3.5 mm pitchyesC201-C205, 5 x 22 uF 25 V X5R 1206 (C12891)
U23.3 V regulator for 3V3_AUXLD1117V33TO-220, lying downyesU201, AMS1117-3.3, SOT-223, tab = VOUT (C6186)
C3U2 input capacitor10 uF 25 V electrolyticradial, D5 mm, 2 mm pitchyesC301, 10 uF 25 V X5R 0805 (C15850)
C4U2 output capacitor10 uFdisc, 5 mm pitchyesR6 1.0 ohm 0603 (C22936) in series with C401 22 uF 1206 (C12891)
Q1NPN transistor2N3904TO-92, E-B-CyesQ101, MMBT3904, SOT-23, 1 B / 2 E / 3 C (C20526)
R1resistor, GPIO45 to Q1 base2.2 kaxial, DIN0207, 10.16 mmyesR101, 2.2 k 0603 (C4190)
J2USB 2.0 type A receptacle, right angleMolex 67643 patternthrough-holeone of J2 / J3SHOU HAN AF 90 WJDG (C456018), soldered by JLC, optional at order time
J3PC internal USB header2x5, pin 9 absent2.54 mm headerone of J2 / J3same, by hand
J4SWD header: SWCLK / GND / SWDIO-1x3 2.54 mmno (developers)same
J5recovery header: RUN / GND / BOOTSEL-1x3 2.54 mmno (recovery)same

The Partial SMD build adds, with J6 only: R2-R5, 10 k 0603 pull-ups on DAT2, CMD (MOSI), DAT0 (MISO) and DAT1 (C25804); C5 100 nF 0603 (C14663) and C6 10 uF 0603 (C96446) at J6's VDD.

U1 must be the B0 variant in both builds: on the B1 / B2 modules GPIO47 is the PSRAM chip select, and this board needs GPIO47 for the I2S word clock. The microSD chip select (GPIO18) works on any variant.

The parts lists are fab/picozorro-one-th-parts.csv (Full TH) and fab/jlc/picozorro-one-th-bom-jlc.csv (Partial SMD).

F1 alternatives: Littelfuse RUEF090 or 30R090U, Bourns MF-R090-0-9, Littelfuse RUSBF090 (0.07-0.12 ohm new, 0.22 ohm at most after a trip; holds 0.57-0.61 A at 60 C). Not the 0.5 A hold parts: they hold 0.36 A at 50 C and drop up to 0.59 V. The footprint is KiCad's Fuse_Bourns_MF-RHT070 (5.1 mm lead spacing with the 1.2 mm stagger of a radial PTC's leads, 0.71 mm holes for 0.51 mm leads); the KiCad library has no footprint for the parts named above.

Q1 alternatives: the footprint is three holes in line, base in the middle, emitter - base - collector as pins 1-2-3.

PartPin order 1-2-3Fits
2N3904, PN2222A, 2N4401, S8050E-B-Cas drawn
BC547, BC548, BC337C-B-Erotated 180 degrees (silkscreen: "BC547 C-B-E: turn round")

Only the 2N3904 and the BC547 were checked against a data sheet; check the others before fitting.

U2 alternatives (pin order GND / VOUT / VIN seen from the front, tab = VOUT): LD1117V33, LM1117T-3.3, AZ1117T-3.3, LM1086CT-3.3, LT1086CT-3.3. They do not fit with VIN and VOUT swapped, middle pin GND, tab GND: LF33CV, LM3940IT-3.3, MCP1825S, MCP1826S. The silkscreen names the three holes. C3 is an electrolytic: mind its polarity.

Solder jumpers

JumperAs deliveredChange
JP1bridged 1-2 by a trace: Q1 collector to /INT2 (J1 pin 19)cut 1-2 and bridge 2-3 for /INT6 (J1 pin 22)
JP2open in both builds: U2 (Full TH) or U201 (Partial SMD) supplies 3V3_AUXfallback for a board without U2: bridge it, and the module's 3V3 (U1 P2.16) feeds 3V3_AUX. Never bridge it with U2 or U201 fitted

Pin map

The PicoZorro GPIO map (../PINMAP.md) applies to GPIO0-17, 19-43 and 45. Four pins differ on this board:

GPIOModule pinShared mapThis board
18P3.6/LDSmicroSD chip select
44P4.15/ARMI2S data (DIN)
46P2.4UART0 TXI2S bit clock (BCK)
47P2.3not connected / PSRAM chip selectI2S word clock (LCK), B0 module only

The whole map:

GPIOSignalGoes to
0-15D0-D15J1
16/ASJ1 74
17/UDSJ1 72
18SD_CSM2 pin 2 / J6 DAT3
19READJ1 68
20-26A1-A7J1
27XRDYJ1 18, direct
28-35A16-A23J1
36/CFGINJ1 12
37/CFGOUTJ1 11
38/BUSRSTJ1 94
39NIC_INTM1 INTn (the module's LED1 is on the same pin, lit while INTn is high)
40SPI1 MISOM1, M2 / J6 DAT0
41NIC_CSM1 SCSn
42SPI1 SCKM1, M2 / J6 CLK
43SPI1 MOSIM1, M2 / J6 CMD
44I2S_DINM3
45INT_DRVR1 / R101, Q1 / Q101 base
46I2S_BCKM3
47I2S_LCKM3

Bus signals use only GPIO0-38 (5 V tolerant); GPIO39-47 never see the bus. The bus nets go straight from the edge fingers to the module pins: no series resistors, no buffers. XRDY comes from GPIO27 directly (36 ns from /AS to XRDY measured at 240 MHz). The edge connector (J1) has +5 V on pins 5 and 6, all GND pins connected and pin 91 tied to GND.

There is no UART: debug is SWD with RTT on J4.

/LDS (J1 pin 70) is not connected. The bus PIO programs wait on /AS and branch on READ only; the strobes are two bits of the sampled word, used by the register model to tell word from byte access. With /UDS alone, word access, Autoconfig (its writes are /UDS-only bytes on D15-D8) and the rejection of an odd-byte write (/UDS high) work; an even-byte write cannot be told from a word write and is taken as a word write.

GPIO47 and the module variant. On the Core2350B B1 / B2 modules the PSRAM chip select sits directly on GPIO47 with a 10 k pull-up (R9); R11 (0 R) only links the net to header pin P2.3. Driving GPIO47 there selects the PSRAM during flash access. I2S therefore needs the B0 module, and the firmware must not drive GPIO47 unless a PSRAM probe has found none.

Firmware

This board's image is the default firmware build plus the feature board-one (make fw-uf2 FEATURES=board-one at the top of the repository). The defaults xrdy-direct (GPIO27 pulls XRDY) and int-nfet (GPIO45 push-pull, high asserts /INT) match the bus side; board-one drives GPIO18 high from boot as the microSD chip select, leaves GPIO44 / 46 / 47 to the DAC and has no UART log. A default build without it drives GPIO44 as /ARM and GPIO46 as the UART log. make fw-one builds a bring-up image with the USB CDC log and bench commands (W5500 version, microSD init, /INT toggle, a DAC test tone).

board-one brings the microSD block device (feature sd; registers in docs/REGISTERS.md, "microSD block device"):

  • SPI1 is shared. The SD card releases its data output one clock after its chip select rises, so the firmware clocks one byte with both chip selects high before selecting the W5500; each device gets its own SCK rate.
  • An SD card that has not been put into SPI mode still listens to the command line. The firmware puts the card into SPI mode before network traffic starts; a card inserted later sees W5500 traffic first.
  • There is no card-detect line: after a second without commands the firmware checks the card, or tries to initialise one, and reports a change on /INT.

The I2S driver is not written yet; I2S is enabled only after a PSRAM probe has found none.

Power

text
J1 +5V --+-- C1 --GND
         |
         +-- D1 -->|-- VIN ---- U1 P2.15 (VBUS: module LDO input)
         |                 +--- M3 VIN
         |
         +-- F1 ---+--- VBUS_PORT --- J2.1, J3.1, J3.2
         |         +--- C2 --GND
         |
         +-- C3 --- U2 --- C4 --- 3V3_AUX --- M1, M2 / J6

U1 P2.16 (3V3, module LDO output) --- JP2 (open; bridged only without U2) --- 3V3_AUX

Partial SMD build: D101, F101 and U201 (C301 at its input, R6 + C401 at its output) in place of D1, F1 and U2 (C3, C4); the bulk is two MLCC banks, C101-C103 on the bus +5 V for C1 and C201-C205 on VBUS_PORT for C2 (40.9 and 68.2 uF effective at 5 V). JP2 stays open in both builds.

  • D1 (Schottky) keeps bench power off the Amiga's +5 V rail: the module's FPC USB adapter feeds the same VBUS net.
  • F1 (PTC) is all there is between the bus +5 V and the USB port, as on PC motherboards (Intel 815E reference board; Intel, "Power Delivery Design Issues for Hi-Speed USB on Motherboards") and the fuse-only hobby host boards. No series diode (with a Schottky the port sat at about 4.45 V), no power switch IC.
  • The card is powered by the Amiga, or by the FPC adapter on the bench. The port never powers the card.
  • Not protected: a hub that feeds its own 5 V back into the port reaches the Amiga's +5 V rail through F1, also with the Amiga switched off. USB 2.0 section 7.2.1 forbids a device to source current on its upstream port.

Port voltage from a 5.0 V rail at 500 mA: 5.0 - 0.06 (F1 new, 0.11 after a trip) = 4.94 V (4.89 V). USB 2.0 Table 7-7 asks for 4.75 V at a high-power port, so the rail may sag to 4.86 V. Section 7.2.1.2.1 asks for resettable over-current protection below 5.0 A: F1 trips at 1.8 A (Intel's guide sizes a single-port fuse at 1.5 A trip or more). The fault is not reported to the host stack: there is no pin for it.

Bulk: C1 and C2

USB 2.0 section 7.2.4.1 asks for no less than 120 uF of low-ESR capacitance on the downstream VBUS "per hub", to keep the power of the other ports within 330 mV when a device is plugged in (load model 10 uF with 44 ohm). This card has one port: the only other load on that VBUS is the card's own +5 V entry (D1 to VIN, U2) and, behind it, the Amiga rail. So both stay: C1 at the entry, C2 at the port, F1 between them (on a microsecond scale the PTC is a plain resistor that isolates the two).

A hot-plug simulation (1 m cable, assumed rail and ESR values, not a measurement) puts the entry droop at 0.23 / 0.15 V (PTC 0.1 / 0.3 ohm) with the Partial SMD build's ceramic banks, the smallest of the arrangements tried at 0.3 ohm, matched only by a low-ESR polymer 220 uF at the entry (0.22 / 0.17 V). The Full TH build's two electrolytics give 0.43 / 0.29 V; a single ordinary electrolytic at the entry (0.15 ohm ESR) 0.61 / 0.45 V, worse than either split. The port itself dips 0.4 to 3 V for microseconds whatever is fitted: no arrangement holds it within 330 mV with a stick plugged straight in. Both builds lay the two groups out alike: the electrolytic on top, its MLCC row under it in the 15 mm band.

3.3 V

3V3_AUX (M1, M2 / J6) has its own regulator, U2, fitted in both builds; JP2 stays open and the module's LDO feeds only the module. The loads: M1 132 mA typical (W5500 data sheet section 5.4; the WIZ850io module 141 mA), M2 up to 100 mA, the maximum averaged over 1 s in SPI mode (SD Physical Layer Simplified Specification v3.01, Table 3-4 note 7; SanDisk's manual gives 65 mA read and 75 mA write at most). The simplified specification gives no peak or inrush figure. That is about 0.22 A typical, 0.24-0.27 A at those ceilings. M3 is not on this rail: its VIN comes from VIN and it has its own two regulators.

U2 takes the bus +5 V directly, not behind D1. Full TH build: LD1117V33. ST guarantees 3.235-3.365 V out for Vin 4.75-10 V at 0-800 mA (TJ 0-125 C, LD1117 data sheet DocID2572 Rev 38, Table 6), dropout 1.15 V at most at 500 mA: it regulates from 4.45 V, a 0.30 V margin at the bus's 4.75 V minimum. The Partial SMD build fits U201, an AMS1117-3.3 (SOT-223, the same pin order, tab = VOUT) under U2's footprint, its tab on the same 3V3_AUX copper area. Its only guaranteed dropout is 1.3 V at 800 mA, so it needs 4.60 V: a margin of about 0.15 V. R6 (1.0 ohm) in series with C401 (22 uF) at its output gives the ceramic the ESR of the tantalum the AMS1117 data sheet is written for. Dissipation in either build: (5.25 V - 3.3 V) x 0.3 A = 0.59 W at most.

  • C3, 10 uF at the input in both builds (Full TH: 25 V radial electrolytic; Partial SMD: C301, 0805 X5R): one value that satisfies the LD1117 (100 nF), the LM1117 and AZ1117 (10 uF) and the AMS1117 application circuit (10 uF). C4, 10 uF at the output (ST's figure; ST gives no ESR window).
  • With U2 fitted and the card on bench power only, M1 and M2 are unpowered while the module runs.

The module's regulator is the fallback (JP2 bridged on a board without U2), not the default: the ME6217C33M5G data sheet gives 800 mA as a typical IOUT(max) (and as the absolute maximum), no minimum, and notes it may not be reached because of package dissipation; it gives no thermal resistance. Its sibling ME6207 rates the SOT-23-5 package at 250 mW. The W5500 alone puts 0.25-0.41 W into it, with an SD card 0.45-0.66 W. Measured with a thermal camera, module plus W5500 on the module's 3V3, room 20 C: nothing on the module above 40 C (no SD card in that measurement).

There is no extra bulk capacitance on VIN or 3V3: it would slow the 3V3 rise against the Amiga's +5 V ramp.

/INT driver

Q1, an NPN transistor, drives the interrupt line open collector: emitter to GND, base through R1 (2.2 k) to GPIO45, collector through JP1 to /INT2 (default) or /INT6. GPIO45 high asserts the interrupt. GPIO45 resets with its pull-down on, so Q1 is off while the module is in reset; with the module absent the base is open and Q1 is off. GPIO45 is not 5 V tolerant, so it cannot pull the bus line itself.

Sizing: a B2000 and an A500+ pull both lines up with 1 k (5 mA at 5 V); the Technical Reference Manual's open-collector drive class is 0.5 V at 8 mA, so the driver must sink at least 8 mA. R1 gives (3.3 - 0.75) / 2.2 k = 1.2 mA of base current. The 2N3904 is specified at VCE(sat) 0.2 V maximum for 10 mA with 1 mA into the base, the BC547 at 0.25 V maximum (0.09 V typical) for 10 mA with 0.5 mA: both operating points are covered with margin.

A transistor and not a MOSFET: the through-hole MOSFETs hobby shops carry (2N7000, BS170) have gate thresholds up to 3 V and are not specified at 3.3 V drive; the one that is (TN0702) is a distributor-only part. Small NPN types are sold everywhere.

Simulated with ngspice (vendor models, 1 k to 10 k pull-ups, 4.75 to 5.25 V, 0 to 70 C, GPIO45 down to 3.0 V, R1 +5 %): /INT sits below 90 mV with a 2N3904 and below 40 mV with a BC547B, also at a forced 8 mA; GPIO45 sources 1.15 mA at most. Assertion takes under 125 ns into 1 nF. Release is the pull-up RC plus the transistor's storage time: 0.25 to 0.5 us for the 2N3904, 1.4 to 2.2 us for the BC547B, so 10 k with 500 pF reaches 2.0 V after 3.2 us (2N3904) or 4.9 us (BC547B). The 2N3904 is the default for that reason.

Modules

M1, W5500 Lite. Two 1x6 rows, 2.54 mm pitch, 20.32 mm apart; module 23.0 x 25.0 mm; pin 1 of each row at the jack-opening end, 6.4 mm from that edge; the jack overhangs that edge by 2.5-2.7 mm. Seen from the top with the jack opening away from the viewer (module silk J1 left, J2 right, jack HR961160C):

Left row (J1)Right row (J2)
1 GNDGND1 GNDGND
2 GNDGND2 3.3V3V3_AUX
3 MOSIGPIO433 3.3V3V3_AUX
4 SCLKGPIO424 NC-
5 SCSnGPIO415 RSTnnot connected
6 INTnGPIO396 MISOGPIO40

J1 pin 5 is SCSn as on the W5500 and the USR-ES1 schematic; WIZnet's WIZ850io page spells it SCNn.

RSTn stays open: the module has its own power-on reset (10 k, 100 nF, 1N4148) and the firmware resets the chip in software. The module has 10 k pull-ups on SCSn and INTn, so the W5500 is deselected while the RP2350 is in reset.

M2, microSD breakout. One 1x6 row: 3V3 (pin 1, to 3V3_AUX), CS (GPIO18), MOSI (GPIO43), CLK (GPIO42), MISO (GPIO40), GND. Board 18 x 18 mm (18.1 mm over its scored edges); the socket, 13.7 mm wide, overhangs the edge opposite the pin row by 3 mm and the card enters from there. No pull-up on the chip select: GPIO18 resets with its pull-down on (about 50-80 k), which selects the card while the RP2350 is in reset; nothing is clocked then, and the firmware drives the chip select high as soon as it has configured its GPIOs, before it uses SPI1. The breakout carries four 10 k pull-ups of its own, on lines not identified.

J6, microSD socket (Partial SMD build). A SHOU HAN TF PUSH push-push socket (PicoZorro:microSD_ShouHan_TF-PUSH, ../lib/README.md) under M2's outline, on M2's nets, the card entering over the top edge as with M2: CS = DAT3 (GPIO18), MOSI = CMD, CLK, MISO = DAT0, VDD = 3V3_AUX. 10 k pull-ups R2 (DAT2), R3 (CMD / MOSI), R4 (DAT0 / MISO), R5 (DAT1), none on CS; C5 100 nF and C6 10 uF at VDD. Card detect (pin 9) is not connected: no GPIO is free; the firmware polls the card instead. M2 and J6 are never both fitted.

M3, GY-PCM5102. The 6-pin row on its short edge: SCK to GND (selects the PCM5102A's internal PLL), BCK to GPIO46, DIN to GPIO44, LCK to GPIO47, GND, VIN to VIN. Pin 1 is SCK. The 9-pin row (FLT, DEMP, XSMT, FMT, A3V3, AGND, ROUT, AGND, LROUT from the SCK corner) gets nine unconnected holes: it only holds the board. Audio leaves through the breakout's 3.5 mm jack. The four strap bridges on its underside stay as usually delivered: FLT low, DEMP low, XSMT high, FMT low. BCK and LCK are on consecutive GPIOs so one PIO side-set drives both; PLL mode accepts BCK = 32 fs or 64 fs from 32 kHz to 384 kHz. Geometry, from a third-party footprint (sstojos, zynthian-miniature) cross-checked against board photos: board 31.6 x 17.0 mm; the 6-pin row 1.5 mm from the short edge; the 9-pin row 1.5 mm from the long edge next to the SCK pin, its first hole (FLT) 6.35 mm along the board from the 6-pin row and 0.635 mm outside the SCK hole.

USB host port, debug and recovery

  • The module's USB pins (D+ on P2.9, D- on P2.7, 27 R on the module) go to J2 and J3 in parallel; only one of the two is fitted.
  • J2: through-hole type A, right angle, the hole pattern shared by Molex 67643, Stewart SS-52100-001, Kycon KUSBX-AS1N, Connfly DS1095 and SHOU HAN AF 90 WJDG (the one JLC solders in the Partial SMD build). 1 VBUS_PORT, 2 D-, 3 D+, 4 GND, shell to GND.
  • J3: 2x5, pin 9 absent (key). 1 and 2 VBUS_PORT, 3 D-, 5 D+, 7 and 8 GND; 4, 6 (a second port's data) and 10 (over-current sense on a PC) not connected. A PC slot bracket's first socket works, its second has power only. A 1x4 cable fits the odd row.
  • J4 (SWD): SWCLK (P2.10), GND, SWDIO (P2.6), placed clear of every other connector.
  • J5: RUN (P2.11), GND, BOOTSEL (P2.5). RUN to GND resets the card; with BOOTSEL held to GND the RP2350 starts its USB bootloader. The module's FPC adapter has both as buttons.

Mechanical

  • Plain rectangle, no nose, bracketless, two layers, 1.6 mm FR-4. 155.5 x 65.92 mm including the 7.62 mm tongue; the body is 58.3 mm above the shoulder, the low-profile PC card envelope.
  • Edge connector: 2 x 50 gold fingers at 2.54 mm, orientation and geometry in ../EDGE_CONNECTOR.md. ENIG with a 0.5 x 45 degree bevel on the finger edge is recommended; HASL works. The finger area is one open solder-mask window by design. JLC's Economic PCBA offers no gold fingers: ENIG without the bevel there, or Standard PCBA for hard-gold bevelled fingers. The rev A boards so far were Economic PCBA with HASL and no bevel, and they work in the slot; ENIG with a bevel is likely the better choice for long-term use.
  • All parts go on the component side. The pin 1 / 2 end faces the rear panel on the Denise; the RJ45 and the USB-A socket go to that end. Tall parts sit high on the card, away from the fingers, to clear an accelerator; the 15 mm band above the fingers holds only lying axial parts and the solder jumpers.
  • Heights above the PCB: M1 about 17.8 mm (2.54 mm header plastic + 1.6 mm PCB + 13.61 mm jack; the header plastic cannot be left out, the W5500 sits on the module's underside); U1 on its header plastic (parts on its underside); U2 lying down.
  • The tall parts do not clear a neighbouring card. Free height to the next card's PCB is 13.6 mm on the Denise and 18.7 mm in an A2000, before that card's solder-side leads. M1 needs the adjacent slot on its side empty in every machine.
TargetBody heightLengthSlot pitch
Enterlogic A-NIC outline45.1155.5-
Denise, low-profile PC card envelope58.3 maxabout 16815.24
Denise, full-height PC card envelope100.6 maxabout 171 over the board15.24
A2000 and the Zorro II form factor114.5 max337.19 max20.32

On the Denise the card's shoulder rests 13.97 mm above the motherboard (EDAC 745-100-520-206, notched ends), and the card goes into the slot next to the CPU (CN4). Mini-ITX guarantees only 57 mm of free height over that slot, which is a 43.0 mm body; the One TH relies on the case for the rest.

First power-up

  1. Before soldering the module: confirm the edge connector's orientation against ../EDGE_CONNECTOR.md on the bare board (pins 2 and 4 are GND, pin 6 is +5 V; they sit at the rear/bracket end). Then, with the strips soldered and nothing else fitted, check for shorts between +5 V, 3V3 and GND.
  2. On the bench, not in the Amiga: build hello as a UF2 (make fw-uf2 BIN=hello at the top of the repository), hold BOOTSEL (J5, or the FPC adapter's button) while the module powers up from its FPC adapter, and copy firmware/hello.uf2 onto the USB drive that appears. Its USB CDC log shows the chip's stepping and MAC. D1 keeps this bench power off the bus pins.
  3. First time in the Amiga: bus-listen the same way (make fw-uf2 BIN=bus-listen, firmware/bus-listen.uf2). It drives nothing on the bus and prints the bus pins and the /AS statistics once a second on the USB CDC log, so a wrong finger shows up before anything is driven.
  4. Then the card's firmware: make fw-uf2 FEATURES=board-one, drop firmware/picozorro-install.uf2 the same way (or make fw-install with an SWD probe; docs/UPDATE.md), and pztest on the Amiga finds the card and exercises its scratch register.

Open points

  • U2 / U201 temperature at the full 3V3_AUX load (0.59 W at most, TO-220 or SOT-223 on the 3V3_AUX copper area). The ME6217 with M1 and an SD card on it applies only to a board without U2 (JP2 bridged): 40 C at most with M1 alone, and by its data sheet not a sound default.
  • The 3V3_AUX rise from U2 / U201 against the module's 3V3 and the Amiga's +5 V ramp. Measured only for the bare module: 18.6 us.
  • The hot-plug droop figures (Power, Bulk) are a simulation with assumed cable, rail and electrolytic ESR values, not a measurement.
  • Partial SMD build: U201's loop with R6 + C401 next to C6 and the W5500 module's own ceramics (not yet checked with a scope); the port droop with the MLCC banks at hot-plug; the 3V3_AUX rise from U201 against the module's 3V3.
  • J6 (TF PUSH): no maker footprint; the land pattern is LCSC's EasyEDA one, checked against the maker's drawing. The maker says nothing about tracks under the socket body.
  • Q101: an MMBT3904 in place of the simulated 2N3904 (a PMBT3904 model was among those simulated).
  • Q1 on the bus: the NPN driver is simulated only. The firmware's int-nfet drive was run against a 2N7002 MOSFET, not an NPN transistor.
  • GY-PCM5102: no maker document. Pin order confirmed from the silkscreen; the geometry is a third party's footprint that agrees with photo measurements to about 0.2 mm, not a measurement on a board.
  • F1: the footprint is that of another radial PTC family; any PTC close to the rating fits those holes.
  • microSD breakout: no maker document. Pin order, 3V3 end, socket overhang, board size and socket width are measured; the position of the row and of the socket along their edges (socket centred) is assumed; the lines of its four pull-ups are unknown.
  • Whether R9 / R11 are fitted on the B0 module (expected: yes, so GPIO47 has a 10 k pull-up and reaches P2.3).
  • B2000 pull-ups are from drawing 312726 rev 2 only; the Denise's own /INT pull-ups are not measured.
  • The free space next to CN4 on the Denise with an accelerator fitted (the TF536), at the height where the modules sit. Denise positions were measured on a board render; the PC envelope figures are PCI Express (CEM 3.0 Fig 9-26); A3000 / A4000 limits beyond the form factor are not checked; Commodore gives no component-height limits.
  • The release time of /INT against Exec clearing INTREQ after the server chain is not measured.
  • The DAC: a test tone stays silent over an idle buzz. Suspects: R11 on the module (GPIO47 to P2.3), the SCK / XSMT straps, the 32 fs PIO.

Sources

Parts

48
QtyValueFootprintReferences
922uCapacitor_SMD:C_1206_3216MetricC101, C102, C103, C201, C202, C203, C204, C205, C401
410kResistor_SMD:R_0603_1608MetricR2, R3, R4, R5
1100nCapacitor_SMD:C_0603_1608MetricC5
1100uCapacitor_THT:CP_Radial_D6.3mm_P2.50mmC1
110uCapacitor_THT:CP_Radial_D5.0mm_P2.00mmC3
110uCapacitor_SMD:C_0805_2012MetricC301
110uCapacitor_THT:C_Disc_D5.1mm_W3.2mm_P5.00mmC4
110uCapacitor_SMD:C_0603_1608MetricC6
11N5817Diode_THT:D_DO-41_SOD81_P10.16mm_HorizontalD1
11R0Resistor_SMD:R_0603_1608MetricR6
12.2kResistor_THT:R_Axial_DIN0207_L6.3mm_D2.5mm_P10.16mm_HorizontalR1
12.2kResistor_SMD:R_0603_1608MetricR101
1220uCapacitor_THT:CP_Radial_D8.0mm_P3.50mmC2
12N3904Package_TO_SOT_THT:TO-92_InlineQ1
13V3 linkJumper:SolderJumper-2_P1.3mm_Open_RoundedPad1.0x1.5mmJP2
1AMS1117-3.3Package_TO_SOT_SMD:SOT-223-3_TabPin2U201
1Core2350B P1Connector_PinHeader_2.54mm:PinHeader_2x08_P2.54mm_VerticalP1
1Core2350B P2Connector_PinHeader_2.54mm:PinHeader_2x08_P2.54mm_VerticalP2
1Core2350B P3Connector_PinHeader_2.54mm:PinHeader_2x08_P2.54mm_VerticalP3
1Core2350B P4Connector_PinHeader_2.54mm:PinHeader_2x08_P2.54mm_VerticalP4
1GY-PCM5102 6-pinConnector_PinHeader_2.54mm:PinHeader_1x06_P2.54mm_VerticalM3J1
1GY-PCM5102 9-pinConnector_PinHeader_2.54mm:PinHeader_1x09_P2.54mm_VerticalM3J2
1INT selJumper:SolderJumper-3_P1.3mm_Bridged12_RoundedPad1.0x1.5mm_NumberLabelsJP1
1LD1117V33Package_TO_SOT_THT:TO-220-3_Horizontal_TabDownU2
1MMBT3904Package_TO_SOT_SMD:SOT-23Q101
1PTC 0.9A holdFuse:Fuse_Bourns_MF-RHT070F1
1PTC 1.1A holdFuse:Fuse_1812_4532MetricF101
1RUN/BOOTConnector_PinHeader_2.54mm:PinHeader_1x03_P2.54mm_VerticalJ5
1SS14Diode_SMD:D_SMAD101
1SWDConnector_PinHeader_2.54mm:PinHeader_1x03_P2.54mm_VerticalJ4
1TF PUSHPicoZorro:microSD_ShouHan_TF-PUSHJ6
1USB intConnector_PinHeader_2.54mm:PinHeader_2x05_P2.54mm_VerticalJ3
1USB_AConnector_USB:USB_A_Molex_67643_HorizontalJ2
1W5500 Lite J1Connector_PinHeader_2.54mm:PinHeader_1x06_P2.54mm_VerticalM1J1
1W5500 Lite J2Connector_PinHeader_2.54mm:PinHeader_1x06_P2.54mm_VerticalM1J2
1Zorro_II_Edge_100PicoZorro:Zorro_II_Edge_100J1
1microSD breakoutConnector_PinHeader_2.54mm:PinHeader_1x06_P2.54mm_VerticalM2

Similar boards

  • 4 layers180 × 100 mmClass board V0.2, the printed circuit board in 3D

    Class board V0.2

    TIGP-Experimental-Methods/class-board-2026

    TIGP class instrument 2026: one board, five blocks - KiCad hardware, ESP32-S3 firmware, phone PWA, Python API, workbook

    342 parts
  • 4 layers110 × 90 mmFilMachine Mainboard, the printed circuit board in 3D

    FilMachine Mainboard

    yesnoj/FilMachine_Reworked

    FilMachine is an automated film processing machine designed for photographic film development. It handles the entire process: chemical baths, water rinses, temperature regulation, and motor-driven film agitation — all…

    205 parts
  • 2 layers92 × 69 mmBase controller, the printed circuit board in 3D

    Base controller

    dy-ma/halloween26

    A small LED/resistor practice board used to learn KiCad schematic capture, footprints, routing, silkscreen, and Gerber export. It is not the tractor controller and was not intended for manufacture.

    64 partsCERN-OHL-P-2.0
  • 2 layers40 × 111 mmLED Wled, the printed circuit board in 3D

    LED Wled

    edvard5/wEasyLed_V2

    96 parts
  • 2 layers100 × 100 mmHour meter board microfit can handroute, the printed circuit board in 3D

    Hour meter board microfit can handroute

    ilomon10/embedded-totalizer-hour-meter

    ESP32 engine hour meter & totalizer: firmware, SensorFrame protocol, Node host tools, KiCad PCB and a Deno device simulator

    152 parts
  • 2 layers130 × 164 mmCct main, the printed circuit board in 3D

    Cct main

    syncmeta/PendingHome

    全屋双色温灯光控制方案 基于 HA 和 ESPHome

    217 partsMIT
  • 4 layers88.1 × 59.4 mmChromapi motherboard v1, the printed circuit board in 3D

    Chromapi motherboard v1

    Mowibox/chromapi_motherboard

    Raspberry Pi control HAT dedicated to the quadruped robot Chromapi.

    139 parts2Apache-2.0
  • 4 layers50 × 69.5 mmRia, the printed circuit board in 3D

    Ria

    B44ken/ria

    ria is a small wheeled biped built for jumping and fast ground motion.

    107 parts

Questions

How big is the Picozorro one th PCB?

It measures 155.5 × 65.9 mm and has 2 copper layers, 1.6 mm thick.

What parts does the Picozorro one th use?

48 parts, built around the AMS1117-3.3, LD1117V33, CL31A226KAHNNNE and CC0603KRX7R9BB104. The full list of its parts is on this page, with links to each part's symbol and footprint where PCBKit's library has them.

Where can I download the Picozorro one th KiCad files?

From the maximevince/PicoZorro repository on GitHub, at the commit this page was made from: the board is hardware/picozorro-one-th/picozorro-one-th.kicad_pcb and its schematic hardware/picozorro-one-th/picozorro-one-th.kicad_sch. PCBKit opens the board in the browser, with nothing to install.

Can I use the Picozorro one th design in my own project?

Its repository states no licence, so its author keeps all rights: ask maximevince before reusing the design.