Skip to the content
2 layers82 × 56 mmTitan 2026 rev a, the printed circuit board in 3D

Titan 2026 rev a

savob/titan_2026

TITAN Tandem Electronics System for 2026

Size
82 × 56 mm
Copper layers
2
Thickness
1.6 mm
Parts
87
Nets
138
Pads
439
Tracks, vias
693, 67
Pours
7
Routed
Yes
KiCad
KiCad 10.0 (file 20260206)
Created
8 Apr 2026
Updated
19 Sep 2026

Overview

Titan 2026 rev a is an open-source printed circuit board by savob, from the savob/titan_2026 repository on GitHub. It is a 2-layer board of 82 × 56 mm with 87 parts on 138 nets, routed with 693 tracks and 67 vias.

It is built around the STM32F105RCT6, AT2659S and ATGM336H-5N31, with the BME280, INA219AIDCN and TPS2116DRL among its other parts. Its parts include 24 resistors, 22 capacitors, 13 ICs, 12 connectors and 3 diodes. It belongs with the STM32 boards of this library.

Main components

12

README

hardware/wheel_2026_rev_a/README.md

TITAN Wheel 2026 Rev. A

This board houses a brake temperature sensor as well as a wheel encoder for monitoring speed and distance. It was designed to fit within the wheel bracket close to the brake disk without interfering with the rider's pedaling.

Changes from 2022

The main changes from the last version will be in improving physical reliability of the system

  • Changed everything to other than sensors to be surface mount on the top of the PCB to eliminate the possibility of shorting against mounting surface
  • Changed operational amplifier to a TLV2372IDR instead of an LM358 for better performance, namely rail-to-rail output
  • Using pads to connect the data cables, headers can be soldered in later if desired
  • Removed connector to daisy-chain wheels since that is now handled on the HAT
  • Added notes to operate at 3.3 V not 5 V - although the system can handle it

Looking on data from 2022, there was some noise on the encoder speed signal when plotted alongside the GPS derived speed. This could potentially benefit from filtering the signals, although I would really chalk it up to my firmware not properly capturing each spoke and not using a hardware timer.

Tuning the Encoder's Comparator Operation

The onboard comparator circuit for the wheel encoder needs to have the two potentiometers to be properly set for reliable operation. One is responsible for the baseline threshold the encoder is compared to to know if the spoke is present of not, labeled THRES. The other potentiometer adjusts the level of hysteresis preventing spurious switching by moving the active threshold so the signal can't repeatedly cross it, labeled HYST.

To begin, the board must be installed into the wheel bracket and powered, ideally via the TITAN HAT. An oscilloscope is strongly recommended to probe the test pads while doing this to complete it accurately but, but not necessary. If possible, perform this with the wheel enclosed in its fairing - or out at least out of direct sunlight as it may influence the IR sensor.

Without an Oscilloscope

  1. Connect the target wheel board to an RPi HAT board and supply power to the HAT board to power the sensor board - battery or USB power is fine.
  2. Gently turn both potentiometers entirely counterclockwise to reset them.
  3. Rotate the wheel so that the IR sensor is entirely clear of any brake disk spokes.
  4. Check that the LED is lit up, if not there may be an issue with the board or too much ambient IR which will need different conditions for testing.
  5. Rotate the wheel so a spoke is directly in front of the IR sensor.
  6. Turn THRES clockwise to increase it until the LED turns off, remember this as the "low" mark.
  7. Spin the wheel and make sure that the LED is turning off with each passing of a spoke, and then back on again. Gradually increase THRES by turning it clockwise until the LED is no longer turning on again when the spoke is gone - this is the "high" mark.
  8. Set THRES to be halfway between the "low" and "high" mark. Spinning the wheel should have the LED flash again with each spoke pass.
  9. Keep spinning the wheel and increasing HYST by turning it clockwise.
  10. Once the LED is no longer changing state with the wheel's rotation, this it the "high" mark.
  11. Set HYST to be about halfway from the starting/zero point to the "high" mark.
  12. Spin the wheel to ensure that the LED is flashing again.

This should roughly center the threshold correctly and provide a decent level of hysteresis. It is impossible for the human eye to perceive the true effect of hysteresis with the LED alone, so we're basically just guessing this.

With an Oscilloscope

The precise way to do this, no major guesstimations here!

  1. Connect the target wheel board to an RPi HAT board and supply power to the HAT board to power the sensor board - battery or USB power is fine.
  2. Gently turn both potentiometers entirely counterclockwise to "reset" them.
  3. Rotate the wheel so that the IR sensor is entirely clear of any brake disk spokes.
  4. Check that the LED is lit up, if not there may be an issue with the board or too much ambient IR which will need different conditions for testing.
  5. Read the voltage on the test point labeled RAW, this is our ambient level.
  6. Rotate the wheel so that there is a spoke directly in front of the IR sensor.
  7. Record the voltage on the test point labeled RAW, this is our presence level.
  8. Gradually turn THRES potentiometer until the voltage on the test point labeled THR is halfway between the ambient and presence levels. If the spoke is still in front of the IR sensor at this point, the LED should turn off.
  9. Spin the wheel to observe the LED turning on and off as the spoke passes.
  10. Keep the wheel spinning and probe OUT.
  11. Gradually increase HYST by turning the potentiometer clockwise until the signal on OUT shows a single pulse for each spoke pass rather than a series of pulses during the transition. Then turn HYST a sliver more - just not too much so that the OUT stops switching
    • If you're also able to probe the other two signals you should be able to see that THR is changing between two levels and that RAW should be able to cross THR only once per spoke passing.
    • RAW should continue to a level past THR after crossing it, so that even if THR didn't move RAW would not be causing potential switching with any sensor noise.
    • If the hysteresis gets too large then it might make it hard to catch occasional pulses leading to slower perceived speed.
  12. Spin the wheel to ensure that the LED is flashing with each spoke passing.

Known IR Encoder Issue

There is a fundamental issue with the use of the IR sensor for the wheel encoder: the wheel encoder would fail to track the wheel speed once the brakes were engaged as it gets saturated due to IR radiation from the brake disks reaching temperatures in the multi-hundred degrees as they stop TITAN at the end of runs.

This issue was tolerable as we still had the GPS speed to fallback to, and the brakes would only be engaged when a run was ending so there was no longer a need to know TITAN's exact speed. Future revisions should consider a different sensor type for the encoder, but this revision kept it since it was field proven and doesn't require modifications to TITAN's wheels.

A possible hack that could be done using these boards as-is, would be to replace the IR sensor with a visible light discrete LED and photodiode or photoreceptor pair. This should avoid the issue where the brake disk temperature affects the sensor while retaining compatability with the wheels as they are.

A larger overhaul of the board could move to using a magnetic system by attaching magnets to the spokes. If wheel modification is to be avoided, then inductive or capacitive proximity sensing system for the spokes could be considered but the rework needed there would be significant and need collaboration with the mechanical team.

Parts

87
QtyValueFootprintReferences
16100nCapacitor_SMD:C_0402_1005MetricC1, C2, C3, C4, C5, C6, C7, C8, C9, C11, C13, C18, C20, C21, C22, C23
102kResistor_SMD:R_0402_1005MetricR6, R7, R8, R9, R15, R16, R17, R21, R23, R24
5120Resistor_SMD:R_0402_1005MetricR1, R2, R4, R12, R14
55k1Resistor_SMD:R_0402_1005MetricR5, R10, R11, R19, R20
4MountingHole_2.7mm_M2.5_ISO14580_Pad_TopBottomMountingHole:MountingHole_2.7mm_M2.5_ISO14580_Pad_TopBottomMH1, MH3, MH4, MH4
310uCapacitor_SMD:C_0603_1608MetricC12, C16, C17
328kResistor_SMD:R_0402_1005MetricR3, R18, R22
3SolderJumper_3_OpenJumper:SolderJumper-3_P2.0mm_Open_TrianglePad1.0x1.5mmJP1, JP2, JP5
222uCapacitor_SMD:C_0603_1608MetricC14, C15
2B3B-XHConnector_JST:JST_XH_B3B-XH-A_1x03_P2.50mm_VerticalJ4, J5
2B5B-XHConnector_JST:JST_XH_B5B-XH-A_1x05_P2.50mm_VerticalJ2, J3
2B7B-XHConnector_JST:JST_XH_B7B-XH-A_1x07_P2.50mm_VerticalJ7, J8
2TPS2116DRLPackage_TO_SOT_SMD:SOT-583-8U9, U10
2WHITELED_SMD:LED_0805_2012MetricD2, D3
110mResistor_SMD:R_0805_2012MetricR13
13u3Inductor_SMD:L_Changjiang_FNR6045SL1
1470pCapacitor_SMD:C_0402_1005MetricC19
16n8Inductor_SMD:L_0402_1005MetricL2
18MCrystal:Crystal_SMD_2520-4Pin_2.5x2.0mmY1
1AT2659SPackage_TO_SOT_SMD:SOT-23-6U13
1ATGM336H-5N31titan:ATGM336H-5NU11
1BME280Package_LGA:Bosch_LGA-8_2.5x2.5mm_P0.65mm_ClockwisePinNumberingU1
1BWSMA-KWE-Z001Connector_Coaxial:SMA_BAT_Wireless_BWSMA-KWE-Z001J12
1DS1023-2*20SF11Connector_PinSocket_2.54mm:PinSocket_2x20_P2.54mm_VerticalJ1
1HB-PH9-12727PB2GOBConnector_PinHeader_1.27mm:PinHeader_2x07_P1.27mm_Vertical_SMDJ6
1INA219AIDCNPackage_TO_SOT_SMD:SOT-23-8U6
1ISO1642DWRPackage_SO:SOIC-16W_7.5x10.3mm_P1.27mmU4
1NRF24L01_Breakouttitan:nRF24L01_PA_LNA_BreakoutU12
1REDLED_SMD:LED_0805_2012MetricD1
1SCD41-D-R2Sensor:Sensirion_SCD4x-1EP_10.1x10.1mm_P1.25mm_EP4.8x4.8mmU2
1STM32F105RCT6Package_QFP:LQFP-64_10x10mm_P0.5mmU3
1SW_SPSTtitan:SW_SPST_SKPMSW1
1SolderJumper_2_OpenJumper:SolderJumper-2_P1.3mm_Open_Pad1.0x1.5mmJP4
1SolderJumper_3_OpenJumper:SolderJumper-3_P2.0mm_Open_TrianglePad1.0x1.5mm_NumberLabelsJP3
1TLV73333PDBVPackage_TO_SOT_SMD:SOT-23-5U7
1TPS563201Package_TO_SOT_SMD:SOT-23-6U8
1USB4105-GF-A-120Connector_USB:USB_C_Receptacle_GCT_USB4105-xx-A_16P_TopMnt_HorizontalJ9
1USBLC6-2SC6Package_TO_SOT_SMD:SOT-23-6U5
1XT30PW-F20.G.YConnector_AMASS:AMASS_XT30PW-F_1x02_P2.50mm_HorizontalJ10
1XT30PW-M30.G.YConnector_AMASS:AMASS_XT30PW-M_1x02_P2.50mm_HorizontalJ11

More from savob/titan_2026

1
  • 2 layers48 × 20 mmWheel 2026 rev a, the printed circuit board in 3D

    Wheel 2026 rev a

    savob/titan_2026

    TITAN Tandem Electronics System for 2026

    17 parts

Similar boards

  • 4 layers78 × 48 mmMotherboard rev 3, the printed circuit board in 3D

    Motherboard rev 3

    hpvdt/falcon_motherboard

    Firmware and hardware designs for the Falcon motherboard

    69 parts
  • 4 layers20.5 × 23 mmUblox breakout, the printed circuit board in 3D

    Ublox breakout

    TobleMiner/ublox-M8030-KT-breakout

    Breakout for the bare UBX-M8030-KT chip. Supports pin compatible chips from M6, M7 and M8 generations. JLCSMT assembly supported

    32 parts1
  • 2 layers100 × 80 mmLedboard, the printed circuit board in 3D

    Ledboard

    seisyuu-hantatsushi/pcb

    Printed Circuit Boards

    90 partsMIT
  • 2 layers71.5 × 63.8 mmMiniUSBFF, the printed circuit board in 3D

    MiniUSBFF

    Board-Folk/flash-floppy-drives

    Multiple hardware projects that use the Flash Floppy firmware by Kier Fraser for various systems

    58 parts3
  • 4 layers99.8 × 74.9 mmTo220 powerstage, the printed circuit board in 3D

    To220 powerstage

    rene-dev/stmbl

    AC Servo Driver for STM32F4

    134 parts990GPL-3.0
  • 4 layers60 × 70 mmControl board PCB, the printed circuit board in 3D

    Control board PCB

    arthur-wirjo/reaction-wheel-inverted-pendulum

    A custom-engineered Reaction Wheel Inverted Pendulum designed as a physical testbed for advanced modern control theory.

    77 partsMIT
  • 4 layers80 × 156 mmSupercap simple, the printed circuit board in 3D

    Supercap simple

    Twisted-Fields/acorn-robot-electronics

    PCB hardware designs in KICAD for Acorn, the precision farming rover from Twisted Fields.

    56 parts81
  • 2 layers49 × 18 mmIRF9640SPBF, the printed circuit board in 3D

    IRF9640SPBF

    kaminaris/GigaESC

    12 parts47

Questions

How big is the Titan 2026 rev a PCB?

It measures 82 × 56 mm and has 2 copper layers, 1.6 mm thick.

What parts does the Titan 2026 rev a use?

87 parts, built around the STM32F105RCT6, AT2659S, ATGM336H-5N31 and BME280. 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 Titan 2026 rev a KiCad files?

From the savob/titan_2026 repository on GitHub, at the commit this page was made from: the board is hardware/titan_2026_rev_a/titan_2026_rev_a.kicad_pcb and its schematic hardware/titan_2026_rev_a/interboard.kicad_sch. PCBKit opens the board in the browser, with nothing to install.

Can I use the Titan 2026 rev a design in my own project?

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