Skip to the content
2 layers104.1 × 104.1 mmHaptic console wheel module stripboard, the printed circuit board in 3D

Haptic console wheel module stripboard

speedypleath/wheel-module-kicad

PCB for wheel module

Size
104.1 × 104.1 mm
Copper layers
2
Thickness
1.6 mm
Parts
21
Nets
39
Pads
71
Tracks, vias
92, 1,331
Routed
Yes
KiCad
KiCad 10.0 (file 20260206)
Created
27 Aug 2026
Updated
4 Sep 2026

Overview

Haptic console wheel module stripboard is an open-source printed circuit board by speedypleath, from the speedypleath/wheel-module-kicad repository on GitHub. It is a 2-layer board of 104.1 × 104.1 mm with 21 parts on 39 nets, routed with 92 tracks and 1,331 vias.

Its parts include 6 resistors, 2 capacitors, 2 diodes, 2 connectors and 1 IC. It belongs with the RP2040 boards of this library.

README

README.md

Haptic Console - Wheel Module (M3) Encoder Sensor Board

KiCad project for the electronics of the Wheel Module (M3) of Haptic Console v1.0, part of the ITPMA dissertation Tactile I/O: Exploring physicality in MIDI controllers.

The mechanical side of this module (200 mm DIN aluminium handwheel, PETG shaft, 6204-2RS bearings, GT2 1:1 belt drive) is documented in the Obsidian vault under Notes/Haptic Console - Handwheel Belt Drive Module Plan.md. This repository covers only the sensor board that reads the optical encoder and presents it to the console.

What the board does

The handwheel drives an E38S6G5-600B-G24N 600 P/R optical encoder through a 1:1 GT2 belt. A Raspberry Pi Pico reads the A/B quadrature signals locally, converts them into a clean position packet, and exposes that packet to the Teensy 4.1 master over I2C. The module behaves as a self-contained tactile sense organ rather than a peer processor, matching the connector standard's design philosophy.

The Wheel Module encoder board, hand-wired on 2.54 mm protoboard

The module as actually built. Component side, left to right: J2 (black 5-pin encoder header) and the pull-up/series resistors bottom left, C2 (yellow disc) beside them, U1 centre with C1's electrolytic can at its USB end, J1 (white XH2.54 to the hub) top right, and the D1/D2 status LEDs bottom right.

The encoder interface, and why the pull-ups are there

The encoder is an E38S6G5-600B-G24N: photoelectric incremental, 600 P/R, rated DC 5-24V, with NPN open-collector outputs. The 600B suffix denotes AB two-phase — there is no Z index channel. Absolute zero is established in firmware, not from hardware index.

The open-collector output is the important electrical detail. An open-collector output can only pull its line low. It never drives a line high. The high level is therefore set entirely by whatever rail the line is pulled up to, and is completely independent of the encoder's own supply voltage.

That gives the safety property this board relies on:

  • R1, R2 (4.7k) pull ENC_A_RAW and ENC_B_RAW up to +3V3.
  • The Pico therefore sees a 0V to 3.3V swing regardless of whether the encoder is run from 5V or 12V.
  • R4, R5 (330R) sit in series into the Pico GPIO as fault-current limiters, protecting the pin if a line is ever accidentally shorted to the encoder supply.

4.7k is comfortable here. With ~2 m of encoder cable at roughly 120 pF/m plus pin and stray capacitance (~300 pF total), the RC rise time is about 3 us 10-90%. At a realistic handwheel speed of 300 rev/min each channel runs at 3 kHz, a 333 us period, so the rise occupies under 1% of it. Pull-up current is 0.7 mA per line, far inside the output stage's sink capability.

Without the pull-ups the inputs would float and the quadrature decode would be noise. The pull-ups are not optional here.

Encoder power is taken from the console's 5V rail, which is inside the encoder's rated 5-24V window. If the encoder proves sluggish or noisy at 5V, the documented alternative is a small 5V to 12V boost feeding J2 pin 1 only. Because the pull-ups reference 3.3V, that change requires no modification to the logic side.

Decision, 2026-08-28: the 12V path is deliberately not built. The planned Eurorack translation module needs 24V regardless, so if a higher-voltage rail is ever added it should be designed once for both modules rather than bolted onto M3 alone. Treat 5V as the committed supply until that point. Note also that if the encoder signals ever look sluggish, supply voltage is the less likely cause: edge shape here is set by the pull-up RC (4.7k against ~300 pF, ~3 us) which does not change with encoder supply at all. Rounded edges point at the pull-ups (fit 2.2k or 1k); missed counts that worsen with rotation speed point at the supply.

Console interface

J1 is the standard XH2.54 6-pin inter-module connector (standard v1.1):

PinSignalBoard use
1GNDCommon ground
23.3VNot connected (see below)
35VPowers the Pico via VSYS, and the encoder
4SIG1 / SDAI2C data, Pico GP4
5SIG2 / SCLI2C clock, Pico GP5
6SIG3 / IRQData-ready, Pico GP6

Pin 2 is deliberately left unconnected. The connector standard warns against feeding both VSYS and 3V3 as competing power paths into a Pico. The board takes 5V into VSYS and uses the Pico's own regulated 3V3 output as the logic reference for the pull-ups.

I2C bus pull-ups live on the master/hub board, not here, per the same standard.

Encoder connector

J2 is a 5-pin 2.54mm header matching the encoder's four conductors plus its shield:

PinEncoder wireBoard use
1RedV+, from the console 5V rail
2Black0V / GND
3WhiteChannel A, to R1/R4
4GreenChannel B, to R2/R5
5Bare / foilCable shield, to GND

Pin 5 carried ENC_Z in the first revision. The encoder has no Z channel, so it is now the shield return, terminated to GND at the board end only — single-point, to avoid a ground loop down the 2 m cable run.

Pin map

Pico pinNetFunction
GP2 (pin 4)LNKLink/comms status LED
GP4 (pin 6)SDAI2C target data
GP5 (pin 7)SCLI2C target clock
GP6 (pin 9)IRQData-ready out to master
GP16 (pin 21)ENC_AEncoder channel A
GP17 (pin 22)ENC_BEncoder channel B
VSYS (pin 39)+5VModule power in
3V3 (pin 36)+3V3Logic reference out

At 600 P/R with 4x quadrature decoding, one full handwheel turn is 2400 counts.

Why C1 is there, and why it is an electrolytic

C1 is a bulk reservoir, not a decoupling capacitor, and the two jobs are different. C2 (100 nF on 3V3) is decoupling: it kills the nanosecond-scale current spikes from GPIO switching, and wants to be small and close to the pin. C1 sits on the 5 V rail at the connector entry and does the opposite job — it is a local charge store at the far end of the hub cable.

The 5 V arrives over a metre or two of XH2.54 flying lead, perhaps 1 uH/m of inductance plus some milliohms. Every change in this board's load current has to be supplied through that inductance: the encoder's output transistors switching at 2 x 3 kHz, the Pico's RT6150 buck-boost drawing pulsed input current, and above all hot-plugging J2 with the encoder attached. Without local bulk the rail at VSYS dips and rings instead of holding up. Steady-state draw is modest — encoder 40-100 mA, Pico 25-30 mA, LEDs ~3 mA — so it is the transients that matter.

100 nF will not do this job: it is 100x less charge. A 100 mA step lasting 300 ns gives dV = I*dt/C ~= 300 mV on 100 nF against ~3 mV on 10 uF, and it moves the cable-inductance resonance from ~50 kHz up to ~500 kHz where it is far less damped.

Electrolytic is the right part, not a compromise — bulk capacitance is what aluminium electrolytics are for. The first revision specified a 10 uF ceramic only because Device:CP cached as a zero-pin stub during project creation and switching to the non-polarised Device:C sidestepped the question. (The real cause was simpler: KiCad 10 has no Device:CP at all; the symbol is Device:C_Polarized.)

C1 is now polarised, and nothing in ERC or DRC will catch it if you fit it backwards. Pad 1 is + and goes to +5V; the lead beside the printed stripe is - and goes to GND. On a 5 V rail a reversed electrolytic will vent. Both build guides call this out at the step where it matters.

Bill of materials

RefValueFootprint
U1Raspberry Pi PicoModule:RaspberryPi_Pico_Common_THT
J1XH2.54 6-pin console connectorConnector_JST:JST_XH_B6B-XH-A_1x06_P2.50mm_Vertical
J25-pin encoder headerConnector_PinHeader_2.54mm:PinHeader_1x05_P2.54mm_Vertical
R1, R24.7k pull-up to 3.3VResistor_THT:R_Axial_DIN0207_L6.3mm_D2.5mm_P10.16mm_Horizontal
R4, R5330R series protectionResistor_THT:R_Axial_DIN0207_L6.3mm_D2.5mm_P10.16mm_Horizontal
R7, R81k LED current limitResistor_THT:R_Axial_DIN0207_L6.3mm_D2.5mm_P10.16mm_Horizontal
C110uF 25V bulk on 5V, polarisedwheel-module:CP_Radial_D4.0mm_P2.50mm
C2100nF decoupling on 3.3VCapacitor_THT:C_Disc_D5.0mm_W2.5mm_P5.00mm
D1PWR green 3mmLED_THT:LED_D3.0mm
D2LNK red 3mmLED_THT:LED_D3.0mm

D1 and D2 match the PWR/LNK status cluster in the front panel drawing (Notes/Haptic Console - Wheel Module Front Panel.md).

Layouts and build guides

One schematic, three board files. They share the netlist and nothing else — pick the one that matches how you are building.

BuildBoard fileBuild guide
Stripboard (Veroboard)project/haptic-console-wheel-module-stripboard.kicad_pcbdocs/stripboard-wiring.md, or the interactive docs/stripboard-wiring.html with a live board map
Plain perfboardproject/haptic-console-wheel-module-perfboard.kicad_pcbdocs/perfboard-wiring.md
Fabricated PCBproject/haptic-console-wheel-module.kicad_pcbnone needed — Gerber set

Schematic

Schematic

Full page as renders/schematic-v2.pdf. ERC report: docs/erc-report.txt; exported netlist: docs/netlist.net.

Stripboard (Veroboard)

40 x 40 holes, 104.14 mm square — the full uncut board. The circuit lives in cols 1-28 / rows 1-28; the strips run the board's whole width regardless, which is why the audit matters more than the placement.

Component sideSolder side
Stripboard, component sideStripboard, solder side

The solder-side view is the one to build from: white X marks are the 23 track cuts, and the coloured rods are the eight link wires, drawn to the control unit's colour convention (red power, blue GND, yellow I2C/IRQ, green /LNK). Cuts and links are listed step-by-step in docs/stripboard-wiring.md.

Plain perfboard

28 x 28 holes, 71.12 mm square, every connection a discrete wire.

Perfboard, component side

Wire-by-wire build order in docs/perfboard-wiring.md.

Fabricated PCB

68 x 58 mm, 2-layer, all through-hole.

Front (F.Cu)Back (B.Cu)
Fabricated PCB, frontFabricated PCB, back

The back is mostly the filled GND pour that all 14 GND pads tie into; only /LNK and the +3V3 branch to R7 are routed there.

Status

  • Schematic: complete, ERC 0 errors / 0 warnings, netlist verified by parsing the exported docs/netlist.net — 13 components, 15 nets, every node checked against the pin map above. A clean ERC alone does not prove connectivity; always read the netlist.
  • Encoder revision: ENC_Z, R3 and R6 removed after the datasheet confirmed the 600B part is AB two-phase. J2 pin 5 repurposed as shield. GP18 is now free.
  • Perfboard layout: complete, 28x28 holes on 2.54mm pitch (71.12 x 71.12 mm cut), 0 DRC errors / 9 cosmetic warnings. Every pad verified on-grid with no shared holes. The 32 unconnected pads DRC reports are expected - the board is hand-wired and carries no copper traces.
  • Stripboard (Veroboard) layout: complete, 40x40 holes on 2.54mm pitch (104.14 mm square, the full uncut board), 0 DRC errors / 0 unconnected pads. The circuit occupies cols 1-28 / rows 1-28; the rest is spare board, but the strips still run its full width. Continuous copper strips, 23 track cuts, 11 link connections. Two cuts are load-bearing: row 9 keeps AGND off the GND bus, and column 5 keeps J1 pin 2 (hub 3.3V) unconnected. Rows 4/14/19 are left uncut on purpose - they carry GND on both sides of the Pico and form the GND bus for free. The ~240 DRC warnings are expected: perfboard holes modelled as vias dangle by construction. Generated from scripts/layout.py by the shared ~/KiCad/kicad-stripboard engine - rebuild it rather than editing the board. The eight physical link wires are modelled as colour-coded 3D jumper wires on the solder side, keeping the crowded component side clear.
  • Manufactured PCB layout: complete, 2-layer, 68 x 58 mm, all through-hole. 0 DRC errors, 0 shorts, 0 unconnected items — every net is connected. 38 routed segments plus a filled GND pour on B.Cu (3270 mm² of copper), which all 14 GND pads tie into directly, so GND needs no traces. Two runs sit on the back layer because they would otherwise cross the J1 signal fan on the front: /LNK (U1.4 → R8.1) and the +3V3 branch feeding R7, which goes round the bottom at y=56.5 rather than threading the 0.94 mm gaps between the Pico's pad rows. The front +3V3 spine is offset to x=48.5 rather than running down the x=46.5 pad column, because R5.1 (/ENC_B) sits mid-column at 46.5,43.18.
  • Cosmetic: reference and value text is auto-placed and overlaps net labels in a few places. Harmless electrically, but worth a tidy-up pass in the GUI before the schematic goes into the dissertation.

Verification

bash
export KICAD_SYMBOL_DIR=/Applications/KiCad/KiCad.app/Contents/SharedSupport/symbols
export KICAD_FOOTPRINT_DIR=/Applications/KiCad/KiCad.app/Contents/SharedSupport/footprints
kicad-cli sch erc --format report project/haptic-console-wheel-module.kicad_sch -o docs/erc-report.txt
kicad-cli sch export netlist --format kicadsexpr project/haptic-console-wheel-module.kicad_sch -o docs/netlist.net

Setting both environment variables matters. Without them kicad-cli cannot resolve the stock libraries and reports spurious violations.

Related

  • docs/perfboard-wiring.md - hand-wiring guide for the plain-perfboard build
  • docs/stripboard-wiring.md - cut/link build guide for the Veroboard build, plus the interactive docs/stripboard-wiring.html
  • scripts/layout.py - the stripboard layout as data; source of truth for that board
  • ../kicad-stripboard - the shared engine that builds and audits strip board layouts
  • ../control-unit-kicad - the Control Unit (M6) board, already fab-ready
  • Obsidian: Notes/Haptic Console - Connector Standard.md
  • Obsidian: concepts/haptic-console-wheel-module.md

Parts

21
QtyValueFootprintReferences
3GNDW6, W7, W8
2+5VW1, W3
21kR_Axial_DIN0207_L6.3mm_D2.5mm_P2.54mm_VerticalR7, R8
2330RR_Axial_DIN0207_L6.3mm_D2.5mm_P5.08mm_VerticalR4, R5
1+3V3W4
1/IRQW2
1/LNKW5
1100nFC_Disc_D5.0mm_W2.5mm_P5.00mmC2
110uFCP_Radial_D4.0mm_P2.50mmC1
14.7kR_Axial_DIN0207_L6.3mm_D2.5mm_P5.08mm_VerticalR2
14.7kR_Axial_DIN0207_L6.3mm_D2.5mm_P2.54mm_VerticalR1
1E38S6G5-600B-G24N ABConnector_PinHeader_2.54mm:PinHeader_1x05_P2.54mm_VerticalJ2
1LNK RedLED_THT:LED_D3.0mmD2
1PWR GreenLED_THT:LED_D3.0mmD1
1RaspberryPi_PicoModule:RaspberryPi_Pico_Common_THTU1
1XH2.54-6P ConsoleConnector_JST:JST_XH_B6B-XH-A_1x06_P2.50mm_VerticalJ1

More from speedypleath/wheel-module-kicad

1

Similar boards

  • 2 layers28.3 × 39.8 mmLevelShifter, the printed circuit board in 3D

    LevelShifter

    gusmanb/logicanalyzer

    24 channel, 100Msps logic analyzer hardware and software

    21 parts5.1kGPL-3.0
  • 2 layers171.2 × 72.9 mmAmiga gfxcard before pour, the printed circuit board in 3D

    Amiga gfxcard before pour

    mntmn/amiga2000-gfxcard

    MNT VA2000, an Open Source Amiga 2/3/4000 Graphics Card (Zorro II/III), written in Verilog

    21 parts1k
  • 2 layers93.9 × 46 mmRectifier, the printed circuit board in 3D

    Rectifier

    rene-dev/stmbl

    AC Servo Driver for STM32F4

    21 parts990GPL-3.0
  • 2 layers45.5 × 42.9 mmEsphome han reader, the printed circuit board in 3D

    Esphome han reader

    hackclub/OnBoard

    💡 Join 1,000 teenagers and make your first PCB with a $100 grant!

    21 parts915
  • 2 layers31.8 × 14.5 mmEsp prog, the printed circuit board in 3D

    Esp prog

    drandyhaas/KiCadRoutingTools

    A set of Python/Rust tools to aid routing in KiCad

    21 parts744MIT
  • 2 layers23.5 × 15.8 mmQuiverAttachPCB, the printed circuit board in 3D

    QuiverAttachPCB

    Arrow-air/project-quiver

    Project Quiver documentation, designs, and project artifacts.

    21 parts617CERN-OHL-S-2.0
  • 2 layers69.9 × 41.3 mmHeaterMeterLCD, the printed circuit board in 3D

    HeaterMeterLCD

    CapnBry/HeaterMeter

    HeaterMeter and LinkMeter Arduino BBQ Controller

    21 parts557MIT
  • 2 layers161.3 × 111.8 mmDoj, the printed circuit board in 3D

    Doj

    kelu124/echomods

    Open source ultrasound processing modules and building blocks

    21 parts463

Questions

How big is the Haptic console wheel module stripboard PCB?

It measures 104.1 × 104.1 mm and has 2 copper layers, 1.6 mm thick.

What parts does the Haptic console wheel module stripboard use?

21 parts. 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 Haptic console wheel module stripboard KiCad files?

From the speedypleath/wheel-module-kicad repository on GitHub, at the commit this page was made from: the board is project/haptic-console-wheel-module-stripboard.kicad_pcb and its schematic project/haptic-console-wheel-module.kicad_sch. PCBKit opens the board in the browser, with nothing to install.

Can I use the Haptic console wheel module stripboard design in my own project?

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