Hi @jhfoo2

well, in my case, without the 1 second delay the UI label wasn't fully updated on the display. Or in other words, ESP32 already went into light sleep while in the background the display was still finishing to update. (In your case it might be ok without the delay as the print() takes some time.)

Edit:
I think the reason for not getting the touched X, Y is that internally the wake-up code already read the touch IC registers and by doing so cleared the values. But I could be wrong about that as it is simply a guess.

I updated my example M5PaperS3_Light_Sleep_Test_UIFlow2.4.3 in the UIFlow2 Project Zone to read touch coordinates after wake-up from touch. It's a bit tricky as M5.update() needs to be called repeatedly until the number of touch points drops back to 0.

Thanks
Felix

@foofera I don't know if you ever progressed this issue, but I suggest you try adding a 100us delay between API calls to the PPS library.

I suggest this because I've found the example code unreliable with a core2, with periodic lockups showing Nack errors on the core2. Very sensitive to host app timing. I think the firmware sometimes goes into error handling and the restart logic is too slow for quick subsequent calls to succeed.

With stock code, I find the example app hangs after a few seconds to a few minutes. I've tweaked the library to check the gap between the end of one call and the start next, and if under 100us it adds delayMicroseconds() to extend gap to 100us. It now runs for days without issue.

I think the real fix is to also improve the library error handling with retries, but the delay fix works for my use case.

@pabou try using uiflow to generate the code, then peek copy from there.

Hello @Wambo6000

maybe you are hearing what's described on the product page here: "Abnormal Device Noise".

Thanks
Felix

I took a closer look and compared M5StampS3A vs M5StampS3 BAT and I must say that it is not possible to do any piggyback for pin mapping at the M5StampS3A level since ESP32-S3FN8 vs ESP32-S3-PICO-1-N8R8 have differences in the number of GPIOs (6 pins error), see the difference table below:

f8a8e80c-1dca-472a-ab48-dac7e6656c42-image.png

So it only remains for M5Stack team to upgrade M5StampS3A for 2/8 PSRAM, i.e. have M5StampS3B version.

In addition, the following screen snippet shows two possible suggestions how the 2x9@2.54mm pins can be assigned.

efd89c09-1cce-4886-8c79-39b9ed5a4d05-image.png

Hi. Thanks for posting the code. Can you confirm if this is a once-and-done write to the BMI? I code in micropython and wonder if I can run your code and then flash back to micropython and it'd still work.

the following is a screen snippet of this board:

alt text

@shaulliv
I also tried changing the following settings:

frame_buffer_count: 2 frame_buffer_location: dram

and

priority: 1000

lowered resolution to 320X240.
and

External_Clock frequency: 8MHz

All without success.