Fan controller txu0204-pico

Overview - Fan Controller TXU0204 Pico

⚠️⚠️⚠️ WARNING ⚠️⚠️⚠️

i am just making this for a hobby. i have no idea what i am doing. as you can see from my profile picture, i am obviously a cat, which means my electrical and maths skills in general are questionable at best.

⚠️⚠️⚠️ WARNING ⚠️⚠️⚠️

i hope github completely mangles my indentation.

using the gratuitously over-qualified raspberry pi pico series, this expansion board is meant to control up to 4 fan connections that had better not exceed 1.5a. they should be 12vdc as well.

pwm output from pico is shifted from 3.3vdc to 5vdc because if i used a 12vdc noctua fan, that's what noctua expects. intel recommended supporting 3.3vdc logic, but that's just a recommendation in what seems to be a de-facto standard, not an industry requirement.

the current immediate usage scenario is controlling an air filtration box using a total of 6 arctic p12 pro pst ln fans. these fans are daisy-chainable, with the power and pwm chained to every fan in the series, but (obviously) only the first fan's tachometer signal connected.

however, i would like this design to support more than 2 channels of fan pwm and fan tachometer, so that it can be used on other projects.

CREDITS FOR FONTS

WebPlus_ToshibaSat_8x14 and WebPlus_ToshibaSat_9x16 by VileR

Schematic Notes

expected fan needs gleaned from various documentation

power 12vdc: 12vdc (less than 12.6vdc) power to fan

• current draw per fan probably no higher than 0.7a

power 5vdc:

• rp2040 and rp2350 draw < 0.2A (the datasheets' highest charts were like 37mW for the rp2350, and 94.1mA for the pico 2 board from vbus. these datasheets kinda suck)
  • for cyw43439, probably around 0.5A
• txu0204 power draw

pwm signal:

• 5vdc pullup, max 5.25vdc
• should be open-collector
• max current 5mA, but device should be capable of 8mA minimum
• 21 - 28kHz is the range for arctic's accepted quality control; 25kHz preferred

tach signal:

• usually pulled up using 12vdc
• noctua says max of 13v, but fan manufacturers probably only expect 5vdc can be pulled up to 5vdc or 3.3vdc
• probably open-collector
• max current 5mA, but device should be capable of 8mA minimum
• 2 pulses per revolution

Layout Considerations

might be 2 layer should really be 4 layer sig / gnd / pwr / sig

• oshpark

general routing

• every trace will "try" to return to ground using shortest possible path under? the trace
• try to avoid splitting ground planes, and use vias to help keep ground planes contiguous
• every time trace uses a via to switch layers, add via to ground as close as possible to give the voltage a return shorter path

signal routing:

• if having to cross another signal on another layer, try to do so perpendicularly
• try to keep the traces as spread out as possible
• larger traces have better noise rejection

Component Notes

other signal translation component possibilities: SN74LVC2G17 TXU0204 - integrated pull-down resistors TXU0202 - integrated pull-down resistors

TXU0204

"4-Bit Fixed Direction Voltage-Level Translator with Schmitt-Trigger Inputs and 3-State Outputs"

• vcc determines output voltage level
• using 2 to convert pico-output 3v3 pwm signals to 5v fan-input
• using 2 to convert fan-output 5v tachometer signals to 3v3 pico-input
• integrated 5Mohm pull-down resistors

Power Distribution

syntax of subsections:

### (heading size 3) Xvdc rail

device / component on the rail

* note about that device / component
* another note on the device / component
    * subnote on another note
    * another subnote on another note
* yet another note on that device / component

general rail stuff

regarding rail traces / zones in general

• keep rail traces and zones off to the side as much as possible
• attempt to give them clear ground returns with gratuitous vias
• the zones for the positive inputs / outputs are mostly for heat sinking; the small size of the traces are plenty for these low voltages and amperages

12vdc rail

chanzon 2abn036f 120vac 1a to 12vdc 3a wall wart:

• claims to be UL listed and shit but i highly doubt it is
• powers pico and this expansion board
• powers 4 x P12 Pro PST @ up to 0.3a each
• powers 1 x OKI-78SR-5/1.5-W36-C @ 0.69a
• desired circuit protections: transient, reverse polarity, and overcurrent protection; maybe also overvoltage and undervoltage space constraints
  • tvs diode: SMAJ16CA
  • efuse choice 1: TPS259482A space constraints
    • protects against overvoltage, overcurrent, shorts, reverse current, undervoltage
  • efuse choice 2: TPS259472A space constraints
    • protects against overvoltage, overcurrent, shorts, reverse current, reverse polarity, undervoltage
  • reverse polarity protection
    • p-mosfet
  • input tvs diodes on main power and +12vdc fan pins
    • bidirectional to protect against surges from GND
  • current limit
  • input limit <=2A 12vdc to 5vdc psu
  • idk for fans
• may need filtering but i don't own a real oscilloscope soooooooooo

OKI-78SR-5/1.5-W36-C (12vdc input):

• based on MP2467
• recommends 2a fuse
• inrush transient: 0.16 A2-Sec.
• might need an electrolytic capacitor for up to 3ft of "20awg" wire from the wall wart to prevent instability or whatever
• no mention of needing input filtering; that's very bold of you, murata
• the only filtering mentioned appears to be what is physically on the module's pcb: 2x 100uF (ceramic capacitors)
• desired circuit protections: counter-emf schottky diode
  • schottky diode SGL1-20 for protection against counter-emf when power is shut off if fuse blows / power is lost
  • output of 12vdc to 5vdc psu has built-in limiter
• with supposed ripple of 30mV peak-to-peak, doubt additional filtering is neeeded

5vdc rail

OKI-78SR-5/1.5-W36-C (5vdc output):

• output of 5vdc at up to 1.5a
• mostly selected because it's in my drawer from years ago and has like <30mV ripple, which appears to be completely obscene even at like $6
• the okami series that this dcdc module is from has a cool wolf logo
• no mention of adding additional filtering; output filtering on the module pcb (assuming that's what "Cvbus" is) is 1x 1000uF (ceramic capacitor)

pico supplying via vsys

• raspberry pi pico datasheets rec using p-mosfet DMG2305UX or a schottky diode
  • ORing circuit for 5vdc vbus < - > vsys
    • diode - feel like Vf drop too high, might hinder fans' pwm input
    • p-mosfet - i ain't wanna do another one of these
    • ORing dual ideal diode
    • lm66200 is like $0.60 each max, cool, simple, and small, vs p-fet at like $0.45 each, 40 mΩ on-resistance

3.3vdc rail

pico 5vdc to 3.3vdc converter (same for pico, w, 2, 2w): RT6150

• note that the datasheet won't fully cover the converter's output capabilities b/c the converter is the sum of the conversion circuit
• super noisy in power save mode (115mV pk-pk vs 31mV); power efficiency is much worse, though
• maybe should use linear regulator
• if using analog, rec use external shunt ref like lm4040 (in case using potentiometer for fan speed setting)
  • note: someone said for a human-operated potentiometer, the precision probably doesn't matter
• recommended: stay under 300mA

Signal Integrity

this should probably be a 4-layer board, i don't think i can avoid splitting the ground plane.

References

pcb production

oshpark: oshpark - 2 layer prototype service oshpark - after dark (also 2 layer) service oshpark - 4 layer prototype service

digikey: pcb builder - faq pcb builder - technical specs

efuses

texas instruments "basics of efuses"

general 12vdc pwm fan control (pc fans)

intel corporation "4-wire pulse width modulation (pwm) controlled fans" reference to pulling up to 12v power to fans on tachometer pin

edn article "4 wire pc fan" yoinked motherboard schematic picture from edn article pulls up to 12v power to fans on tachometer pin

texas instruments "fan controller overview" "The tachometer output from the fan is typically an open-collector (or open-drain) output that requires a pull-up resistor. It's customary to connect this pull-up to the same voltage as the fan power." "you can still pull up to a lower voltage though" (slight paraphrasing)

arctic p12 pro pst (ln)

pwm curve for p12 pro pst series p12 pro pst docs page p12 pro pst pwm guidelines p12 pro pst ln docs page p12 pst pro ln pwm guidelines

noctua fan jumpscare

noctua "noctua pwm specifications white paper" Vtachometer for 12V fans: 5V recommended (13V max.)" at <5mA for 12v fans

raspberry pi pico boards

pico power supply note: this applies to pico, pico w, pico 2, pico 2w

layout filched from KiNotes