AnalogWrite creates a PWM when value is 255

[jpcornil-git]

On legacy arduino platforms, when analogWrite value is 255, you get a DC/HIGH output (aliased to digitalWrite HIGH), see:

https://github.com/arduino/ArduinoCore-avr/blob/master/cores/arduino/wiring_analog.c#L110

While on UNO R4/renesas platform, 255 is converted to 255/256 and you get a PWM signal, see:
https://github.com/arduino/ArduinoCore-renesas/blob/main/cores/arduino/analog.cpp#L819

To match behavior, value should either be normalized to 255 ((1 << _writeResolution) - 1) or aliased to a digitalWrite as well

[jpcornil-git]

Workaround for this one is to use digitalWrite(pin, HIGH) (following a pinMode(pin, OUTPUT) to address #58) but it is not very clean

[metehoca]

This issue still appears to be unresolved, and it's causing the internal RGB LED on the Nano R4—which is connected as active low—not to turn off completely.

[jpcornil-git]

Issue is probably simple to solve if you replace L817 of cores/arduino/analog.cpp:

    ptr->pulse_perc((float)value * 100.0 / (1 << _writeResolution));

with (with 8b resolution, 100% is 255 not 256):

    ptr->pulse_perc((float)value * 100.0 / (1 << _writeResolution - 1));

32b float arithmetic looks OK as long as _writeResolution is less than 17 (from 17, IEEE 23b mantissa can't represent (1 << _writeResolution - 1)* 100.0f without truncation leading to 99.9999... iso 100.0%). I quickly looked at the call stack pulse_perc/pulseWidth_raw/.../ and it should propagate to R_GPT_DutyCycleSet without rounding errors and that one seems to accept a value equal to the period that generates a steady '1'

[leonardocavagnis]

@jpcornil-git Good catch! thanks for pointing it out!

@metehoca I opened this PR #483, please give a spin with NanoR4 and let me know.

[metehoca]

@leonardocavagnis works like a charm!