ESP32 S3/C3 Zero and AFSK_External_Radio.ino problem.

[f5djl]

The AFSK_External_Radio.ino example code works absolutely fine on Arduino UNO, Arduino MEGA , ESP32 DEV KIT 1 using decoder Direwolf to decode AX25 to check frame generated .

It compiles fine on ESP32 S3 Zero or ESP32 C3 Zero module but the AFSK frame although similar in duration is corrupted and it can not be decoded properly. ( All above using IDE 2.3.4)

Any suggestion will be much appreciated.

Very best regards
Jean Louis

[Crsarmv7l]

Hmm I have been having long term issues with SPI communications on the ESP32-S2 and ESP32-S3, and it sounds like you are generating and sending an AFSK packet right?

You might be having the same issue (although with a different radio)

Long discussion here: #1138

My troubleshooting is limited without a scope, but I believe noise is triggering spikes on CS. Ignore my ideas about decoupling caps.
Got some more hardware coming to test more but it will be a few weeks.

[jgromes]

@f5djl could you post the output? The only thing that could be going wrong here is PWM not being handled correctly by the ESP32.

[f5djl]

Hello Jan

You are right, problem lies most probably with PWM on C3/S3. I have attached two capture files of preamble. Tone waveform period 1200 looks wrong to me. (1200 period ) on S3/C3 device.

1/ ESP32 Wroom which works absolutely fine.
2/ ESP32 S3 Capture which Direwolf does not decode ( see 1200Hz tone).

Both are test compiled under IDE 2.3.4

Very best regards and thanks a lot for your help and a great library .

Jean-Louis

1

2

[jgromes]

It's hard to tell because the time information is missing from the screenshot, but it looks like the issue is switching between the 1200 and 2200 Hz tone. On ESP32 when using Arduino, tone function is emulated in RadioLib by using the EPS32 ledc API because there is no native support for tone in ESP32 Arduino core. There were some issues with this in the past, such as that changing the tone often would break it. At the moment, this is handled by only changing the tone when there is an actual change in the frequency, as you can see in the ArduinoHal::tone implementation .

Sorry to say but this will be nearly impossible to debug remotely. It really requires having acces to hardware with a scope hooked up to it and trying to figure out how to make that switch cleanly. It might involve things like stopping the ledc peripheral before changing the frequency.

I would set up a simple test ckech that just sets the 1200/2200 tones in a loop, each sent for e.g. 50 ms and trying to figure out a control for the ledc functions that works. But again, that is impossible without having the hardware on hand.

[f5djl]

Sorry i am not at home this weekend ..  i will try to do some more  timing test.Thanks for your help RegardsJLEnvoyé depuis mon appareil Galaxy -------- Message d'origine --------De : Jan Gromeš ***@***.***> Date : 22/02/2025 09:03 (GMT+01:00) À : jgromes/RadioLib ***@***.***> Cc : Jean-Louis F5DJL ***@***.***>, Mention ***@***.***> Objet : Re: [jgromes/RadioLib] ESP32 S3/C3 Zero and AFSK_External_Radio.ino problem. (Discussion #1431) It's hard to tell because the time information is missing from the screenshot, but it looks like the issue is switching between the 1200 and 2200 Hz tone. On ESP32 when using Arduino, tone function is emulated in RadioLib by using the EPS32 ledc API because there is no native support for tone in ESP32 Arduino core. There were some issues with this in the past, such as that changing the tone often would break it. At the moment, this is handled by only changing the tone when there is an actual change in the frequency, as you can see in the ArduinoHal::tone implementation . Sorry to say but this will be nearly impossible to debug remotely. It really requires having acces to hardware with a scope hooked up to it and trying to figure out how to make that switch cleanly. It might involve things like stopping the ledc peripheral before changing the frequency. I would set up a simple test ckech that just sets the 1200/2200 tones in a loop, each sent for e.g. 50 ms and trying to figure out a control for the ledc functions that works. But again, that is impossible without having the hardware on hand. —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you were mentioned.Message ID: ***@***.***>