[Rotaryio] filtered rotary encoder always has a position of 0 when connected to a RP2350b

[timothyBoye]

CircuitPython version and board name

Adafruit CircuitPython 9.2.4-7-g1c0b37e504 on 2025-02-04; Pimoroni Pico Plus 2 W with rp2350b

Code/REPL

import board
import rotaryio
import time

encoder = rotaryio.IncrementalEncoder(board.GP15, board.GP14)

while True:
    time.sleep(0.1)
    print(encoder.position)


# Code to test output of encoder using digitalio
#
# import board
# import digitalio
# import time

# encoderpin1 = digitalio.DigitalInOut(board.GP14)
# encoderpin2 = digitalio.DigitalInOut(board.GP15)
# encoderpin1.direction = digitalio.Direction.INPUT
# encoderpin2.direction = digitalio.Direction.INPUT

# while True:
#     time.sleep(0.1)
#     print(encoderpin1.value, " - ", encoderpin2.value)

Behavior

Rotaryio position of a rotary encoder attached to an external filter circuit then to GPIO pins 14 and 15 on a Pimoroni Pico Plus 2W (RP2350b) does not change when turned.

Description

I have an EC12E type rotary encoder connected to a Pimoroni Pico Plus 2 W (GPIO14 and 15). The encoder uses the recommended external filter circuit from the datasheet (see image) rather than connect directly to the board.

If I try to use rotaryio the position never updates, it's constantly stuck on a position value of 0. However, if I read the encoder pins as digitalio inputs I can see both pins read True by default and when I spin the encoder I can see it flipping between True and False as you would expect.

This seems to be the same behaviour as #9695 but this was reported fixed in PR #9682, a further RP bug was reported #10024 which references these but this was also reported fixed by PR #10025.

Additional information

I attempted to use the same board with a spare encoder and no filtering circuit and that did seem to work with rotaryio and so I thought that the circuit might be at fault.

I have a handheld multimetre which has some basic oscilloscope functionality. So I hooked it up to the GPIO and when using rotaryio I can see the state as high but the low signal does not seem to be returning to GND it hovers at what looks like around halfway between the high signal and GND (if my multimetre has a way to see the actual voltage number on the oscilloscope graph I don't know how). I thought there might be a slight short so I checked the pins and cleaned the board, no change.

I then thought to check those same pins connected to the same filtered encoder circuit but using digitalio instead of rotaryio, using the oscilloscope to measure that, the circuit is going to GND properly as the encoder is turned. This implies to me that the circuit does not have a short or anything like that and that is has to be something about how rotaryio is implemented that is different to just reading the pins using digitalio.

I can only deduce that either the way this circuit is designed is incompatible with rotaryio or something about how roraryio works has a bug for RP2350b when using an external filtering circuit like this.

[todbot]

I just verified this on Pico 2 RP2350. The issue is with the series resistors. Having the external pull-up resistors and the filtering caps are fine. Add the series resistors and rotaryio stops registering pulses. Use 1k resistors instead of 10k and it works. (Seems almost like this is somehow related to the RP2350 E9 errata)

On Pico RP2040, rotaryio works in all cases. Adding the series resistors or any of the other external components in that filter circuit does not affect functionality.

@timothyBoye, as a workaround, omit the series resistors or use 1k. Or omit the entire filter circuit: most designs do not use them; see, I think, any Adafruit product that use a rotary encoder. (e.g. 1, 2)

[timothyBoye]

@todbot thanks Tod! Your work around should be easy to do on the PCB just need to desolder those resistors and replace with a wire. Will try tomorrow.

[timothyBoye]

I removed those resistors and bridged the connection where they were, but kept the the caps and pull up resistors. Testing the rotaryio code again all looks good, it is registering the encoder turns now and updating the position in both directions correctly. Thanks again @todbot for the support.

Should I close this issue as there is a workaround, or leave it open for a CircuitPython fix?

[dhalbert]

Should I close this issue as there is a workaround, or leave it open for a CircuitPython fix?

I think, yes, this is probably an E9 problem. There's no fix except to document not to use a >8.2k resistor to ground.

Just to be clear, are you still using 5V pullups? 3.3v would be fine and safer for the RP2350B, depite the "5v tolerant" verbiage. See https://forums.raspberrypi.com/viewtopic.php?t=375118 for background.

[todbot]

Yes upon further reflection, this is another variation of RP2350 E9 problem. The series resistor turns into a pull-down when the encoder pulse happens.

[timothyBoye]

The 5v I did without thinking as I was just blindly following the data sheet forgetting the rp2350 is a 3.3V device primarily. In my attempt to debug the encoder not working I thought that the 5V might have been the problem so I cut the 5V line before the pull-ups and instead tied it to the 3.3V line (this was before I posted this issue). So it was originally 5V -> 10k ohm -> encoder pin -> 10k ohm -> 0.1uf cap -> pico gpio, but i changed it before i posted the issue to 3.3V -> 10k ohm -> encoder pin -> 10k ohm -> 0.1uf cap -> pico gpio to see if that would fix it and neither had worked.

As you suggested I removed the series resistors but left the pullup resistors which worked. So it's now 3.3V -> 10k ohm -> encoder pin -> 0.1uf cap -> pico gpio. Hopefully that makes sense.