Watchdog kills OMG on ESP32-C3 within seconds

Seems that there is difference on the LED side isn’t it and that the 2 boards would need a different definition on this ?

No, for the Lolin C3 mini these are just dummy pins for now. I think the same is true for the Seeed board

Once FastLED works, it might be different. Maybe then the on-board NeoPixel of the Lolin C3 mini can be used. It is connected to PIN 7. What’s it like on the Seeed board, any built-in LEDs?

Glad you looked at it this carefully. You are right. There is a GPIO9 on the Lolin C3 mini as well. It is not exposed to a pin header, so I missed it at first. GPIO9 is directly connected to a tiny momentary switch. Does that make it a good trigger for OMG? I never used the trigger, so I am not familiar with it. Are there any switches on the Seeed board already?

The Lolin C3 mini also exposes GPIO0 and GPIO1. The Seeed board doesn’t? At the end of the day, that one is just a comment. I can re-word it to capture both boards once I fully understand the differences.

PDF with the pin out scheme of the Lolin C3 mini:

I guess it is perfect. You can try pressing it longer than 15 sec. Then you would have to reenter your WIFI and MQTT data.

Right, the Seeed does not. It exposes 2 - 10, 20 and 21.
By the way, I had leds, resistors and a push button from former model railway times. I hooked them up to the Seeed on a breadboard. See you.

Good, got it. I updated the pull request accordingly.

There is one more crucial difference between the two boards: The Lolin C3 mini needs the SERIAL_JTAG directives for serial output via USB. I assume the Seeed board doesn’t?

At the current moment I cannot test my changes due to this bug: Error: 'GAEN' is not a member of 'TheengsDecoder::BLE_ID_NUM' · Issue #1513 · 1technophile/OpenMQTTGateway · GitHub

The issue says it’s resolved, and my fork shows as synced. Any suggestion how I can get past #1513 in my build?

Edit: the comment above is now obsolete. pio run -t clean was insufficient. pio run -t clean does not seem to clean up everything. However, deleting the local directory and a new git clone of my fork worked. I have now gotten my changes to compile again.

It does, double tilde before and after

1 Like

Thanks, @digiH It works, not on the entire block, just single lines. I edited my comment accordingly.

@mrickma I have now tested the trigger button on GPIO9. This is the button on the board itself. It resets the ESP configuration after a few seconds of holding it pressed. Nice.

are not needed for serial output from the Seeed board. You are really careful.

Then I’m not sure how to handle the environment definition in this respect. Other than that I am happy with it.

@mrickma Can you confirm the following: if you flash with serial output enabled, do you get a delay of mqtt messages/a delay in your led test when connecting to a USB power supply only? For me it seems fine to use the “with_serial” version under either of the following two conditions:

  • I am connected to a computer and listen to the serial output via USB. No delay.
  • I am connected to a USB power supply, no one can be listening to the serial output. No delay.

If I connect to a computer without a serial listener then and only then do I seem to have delays.

Is it the same for you?

@argafal: I understand the difficulties you are in.

180 sec, that was why I joined this thread, see post 13. In a small test program I found that not reading serial output slows down everything else. So I discourage this situation.
If I understood your struggles right we would need three configurations:

  1. Lolin serial reading: JTAG directives added to minimum

  2. Seeed serial reading: minimum

  3. Lolin and Seeed at power supply: LOGLEVEL directives added to minimum

That means both Lolin and Seeed each need two configurations to be complete.
3. seems to be the best possibility to serve both boards in a single but incomplete configuration. The documentation should disclose the other possibilities.
Hopefully I understood the problems right. See you.

Does one of you mind doing a test on your C3 board with:

esp32_c3_s3_platform = espressif32@6.1.0

And see if you get good stability and work as expected.

I’m testing my other boards with it

1 Like

I have the Seed board up now with the new platform. It is in the power supply configuration and the first 30 min look fine. After running for a day I will report again.
I am going to use the espressif32@6.1.0 also for the other ESP32C3 board I have.

1 Like

Okay. I am now testing OMG with espressif32@6.1.0 on lolin c3 mini. I will report back!

1 Like

Using espressif@6.1.0: uptime is 65000. New record, that looks good so far.

See you tomorrow with another update.

1 Like

The Seeed Studio XIAO ESP32C3 has uptime 92166 (>25h) now. That is really good. I switch it off now. My other ESP32C3 board is beyond 24h steadily increasing uptime. Platform espressif@6.1.0 seems to be a good choice.

Thanks for the feedback, I’m getting good results on regular esp32 too

It keeps on running and it runs perfectly. Almost two full days now without a problem. My only change from the lolin_c3_mini environment was the new espressif@6.1.0 dependency. Very nice :slight_smile: @1technophile Will you make this update or do you need a pull request? I propose we can really say that the C3 series is supported now. :slight_smile:

1 Like

Thanks for the feedback, PR currently in test with our different boards:

Just my 2 cents on the matter of stability: I was thinking the issue was just with the ESP32-C3, but it seem that also the ‘vanilla’ ESP32 has the same issues… just much later. Today, one of the ESP32 i’m testing (1.4.0 codebase) just ‘rebooted’
Alas, I have no way of connecting serials at the moment so I don’t know why, just that it lasted nearly 4 days:

We have added a fix on the dev branch for one problem that was generating restarts, also updating the esp32 platform increase the stability with BT from my first tests.