pulseio.PulseIn does not work on FeatherS3

[jschachter]

CircuitPython version

Adafruit CircuitPython 8.0.4 on 2023-03-15; FeatherS3 with ESP32S3

Code/REPL

import board, pulseio

p = pulseio.PulseIn(board.D12) # physical pin 10
while True:
    if len(p) > 0:
        print(p[0])
        p.clear()

Behavior

No output.

Description

Connected a RC receiver outputting PWM (3.3 volt output, I checked w/ scope).

The same code does output a series of numbers on a Raspi Pico.

Also discussed here: https://forums.adafruit.com/viewtopic.php?p=966813

Additional information

No response

[DavePutz]

I don't have a Feather S3, but using CircuitPython 8.0.5 on an Espressif ESP32-S3-DevKitC-1-N8R8 pulsein worked correctly.

[RetiredWizard]

I ran the test code using CP 8.0.5 on an ESP32-S3-DevKitC-1-N8R2 and an Adafruit Feather ESP32 S3 4/2 but didn't get any output. When I ran it on a Pico, it did display output as expected.

One possibly related issue I noticed while testing is that after running the test code, if I press ctrl-C to break out of the loop, the board crashes into safe mode if I press ctrl-D. This only seems to happen if I have sent a PWM signal to the receiving pin. If I just start the test program and then break out of it without ever attempting to send a PWM Ctrl-D works fine.

[jschachter]

@DavePutz How are you generating the input signal? I will see if I can reproduce.

@RetiredWizard I noticed intermittent restarts on the ESP even if it wasn't detecting anything. Not sure if this is the same cause.

[DavePutz]

@jschachter - I am running this script on a Pico using pulseout to generate the signals.

import pwmio
import pulseio
import board
from time import sleep
import array

pulse = pulseio.PulseOut(board.GP27, frequency=40000, duty_cycle=2**16 - 1)
data = array.array("H",[373, 215, 285, 327, 479])
while True:
    pulse.send(data)
    sleep(1)

I did also note that breaking out of the receive loop with CTRL-C sent the ESP32S3 into safe mode. I
believe that is a separate issue, so I'll take a look at what that may be causing that.

[RetiredWizard]

I was simply using pwmio.PWMOut(PinNumber,5000,5000) to generate the test signal (which does work when received by a Pi Pico and not the S3 boards) but when I used the above script with pulseio.PulseOut the receiving program worked on the S3 boards as well.

I'm not sure the S3 results are correct though. The S3 board printed a constant value of 330, while the Pico printed sequences that were somewhat similar to the test signal (65535, 404, 229, 290, 331, 482 was an example of one of the sequences).

[HelenFoster]

Seems continuous signals are ignored on purpose on espressif ports (I'm not sure what the reason is).

circuitpython/ports/espressif/common-hal/pulseio/PulseIn.c

Line 125 in e8711ee

config.rx_config.idle_threshold = 30000; // 30*3=90ms idle required to register a sequence

[RetiredWizard]

I assume this is because an 8.0.5 PR hasn't been moved to the main branch yet, but when I use 8.1.0beta.1 on the S3 boards the receiving program doesn't display any output, switching back to 8.0.5 results in the 330's being displayed again.

[dhalbert]

I assume this is because an 8.0.5 PR hasn't been moved to the main branch yet, but when I use 8.1.0beta.1 on the S3 boards the receiving program doesn't display any output, switching back to 8.0.5 results in the 330's being displayed again.

The main branch is (supposed to be) up to date with fixes from the 8.0.x branch. I think you are seeing a regression due to newer changes that are only on main.

[RetiredWizard]

Okay, I'll keep poking at it 😁

[RetiredWizard]

I think I found the regression in this 1063ec0 merge to main. Specifically, If I undo the following change:

and add:

void port_background_tick(void) {
}

to the current main branch, the S3 board running the test program above displays values of 330 when the test transmit program is run and connected.

I'm still concerned that the 330 values aren't correct and the S3 should be displaying a sequence similar to the one that the Pico displays.

[jschachter]

Using @DavePutz's code on a pico, and the following code on the FeatherS3 (running 8.0.4), I get the following response, which matches the last part of the signal.

[330, 483]
[330, 483]
[330, 483]
[330, 483]
[330, 483]
[330, 483]

import board, pulseio

p = pulseio.PulseIn(board.D12)
print(p.paused)
while True:
    if len(p) >0:
        x = []
        for i in range(len(p)):
            x.append(p[i])
        print(x)
        p.clear()

[jschachter]

Emulating a servo pwm output, on the pico and reading the input with the code above yields no output.

# pico
import pwmio
import board
from time import sleep

servo_center = int(1.5 / 20.0 * (2**16))
pwm1 = pwmio.PWMOut(board.GP22, duty_cycle=servo_center, frequency=50)

while True:
    sleep(0.1)
    print('hi')

Code on the ESP32S3:

# esp32 
import board, pulseio
p = pulseio.PulseIn(board.D12)
print(p.paused)
while True:
    if len(p) >0:
        x = []
        for i in range(len(p)):
            x.append(p[i])
        print(x)
        p.clear()

[DavePutz]

I'm not sure why, but this script shows all the incoming pulse values, not just the last 2:

import board, pulseio
p = pulseio.PulseIn(board.IO12,maxlen=600, idle_state=True)
print(p.paused)
while True:
    if len(p) >0:
        x = []
        for i in range(len(p)):
            x.append(p[i])
        print(x)
        p.clear()

The only differences are the addition of maxlen and idle_state on the PulseIn init line.

[jschachter]

@DavePutz idle_state is unneeded, and maxlen defaults to 2, thus getting the two measurements.

[jschachter]

I think pulseio.PulseIn cannot detect single pulses.

on pico, as above, except data = array.array("H",[1500, 1500, 1500])

on esp32, as above, [1506, 1503, 1503] repeating.

[DavePutz]

Looks like the pwm not showing results is due to this line in Espressif common-hal/pulseio/PulseIn.c:
config.rx_config.idle_threshold = 30000; // 30*3=90ms idle required to register a sequence
Using a PWM frequency higher than 10 will result in an idle less than 90 ms; and so the pulsein read will never complete. The PWM example on the pico will result in pulses being seen on the ESP32S3 if frequency is 10 or less. I am not sure of the reason the pulsein idle_threshold was set to this seemingly arbitrary value; perhaps someone else could answer that question.

[jschachter]

@DavePutz fwiw default is 12000 in https://github.com/espressif/esp-idf/blob/c9646ff0beffc86d2c6d1bfbad34da16e328e0e3/components/driver/include/driver/rmt.h#L126

i'm not sure that's the issue, though. two pulses with a similar frequency were picked up. was using 50Hz as the frequency above.

[jschachter]

Re-closing due to accidental re-open

[dhalbert]

@jschachter did you delete your remark that it wasn't working?

[jschachter]

Yes, I should not have replied.

…

On Mon, Apr 10, 2023 at 8:39 AM Dan Halbert ***@***.***> wrote: @jschachter <https://github.com/jschachter> did you delete your remark that it wasn't working? — Reply to this email directly, view it on GitHub <#7817 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAEXFUUEG7MEQC2LXCUIYTXAQSUHANCNFSM6AAAAAAWMRAGJQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>