Inconsistent wake_on_ext1 and deepsleep()

[davefes]

Doing a CTRL-D (SOFT_RESET) then going to deepsleep, when taking
wake_on_ext1 high did not always trigger it out of deepsleep.

I needed to do a EN or my own power_on reset circuit then it would
come out of deepsleep ... reliably.

However, after running the full script below 20-30 times I then removed
the poweron_reset stuff and it came out of deepsleep every time.

There is something happening that I do not understand. As far as I
can tell nothing changed between the two test scenarios.

Pin33 drives a NPN transistor which can pull the EN line low.

Test script:

import umachine
import utime
from umachine import Pin, deepsleep
import esp32


wake1 = Pin(35, Pin.IN)  # DS3231 wake 
wake2 = Pin(26, Pin.IN, Pin.PULL_DOWN) #  wake on rxDone

esp32.wake_on_ext1([wake1, wake2], level = esp32.WAKEUP_ANY_HIGH)

#poweron_reset = Pin(33, Pin.OUT)
#poweron_reset.off()


#if umachine.reset_cause() == umachine.SOFT_RESET:
#    print('waiting 5 seconds so that I can do a CTRL-C')
#    utime.sleep(5)

#    print('do a poweron_reset')
#    utime.sleep_ms(10)

#    poweron_reset.on()

#    print('going to deepsleep')
#    utime.sleep_ms(10)

#    deepsleep()

print('waiting 5 seconds before going to deepsleep')
utime.sleep(5)

print('going to deepsleep')
utime.sleep_ms(10)

deepsleep()

[robert-hh]

About this sequence of statements:

poweron_reset.on()
print('going to deepsleep')
utime.sleep_ms(10)
deepsleep()

Assuming you have the base of the NPN transistor at Pin 33 (I hope with a resistor), the emitter at GND and the collector at EN, the operation seems questionable. As soon as poweron_reset.on() is executed and Pin 33 is set to high, EN goes low and Pin 33 is going low immediately as well, resp is switched to input mode, switching the transistor off again. So the EN low pulse is very short and I do not know if the ESP32 is fully reset. If you see the output of the print statement, it's not.

[davefes]

Yes, with a 1K series resistor. OK, on the possibly too short pulse time. I have used a 10uF on the base to ground to pulse-stretch before, but this time I didn't. https://github.com/orgs/micropython/discussions/9345

The main "issue" that I thought it fixed was going to deepsleep after a CTRL-D seemed to bad things to the wake_on_ext1.

I'll do some more testing. Thanks for the response.

[davefes]

@robert-hh
Built up another test board and this one seems to behave properly. The ESP32 that seems to misbehave is quite well buried in a project so will try something like:

if umachine.reset_cause() != umachine.DEEPSLEEP_RESET:
    print('waiting 2 seconds so that I can do a CTRL-C')
    utime.sleep(2)

    umachine.reset()

I think to do a proper external poweron_reset circuit would require a one-shot that pin33 could trigger. That could then hold the EN pin low for as long as it takes to do a proper board reset.