[Question] GT9895 touch panel appears to be less sensitive than HI8561 one on T-Display P4

[lyusupov]

Hello Lewis !

I've released SoftRF for LilyGO T-Display P4 yesterday https://github.com/lyusupov/SoftRF/releases/tag/1.9

Description of use case is there: https://github.com/lyusupov/SoftRF/wiki/Concorde-Edition

I've noticed that the touch panel sense function operates differently on TFT and AMOLED versions of the T-Display.
In most cases, a swipe gesture with HI8561 works as expected while the GT9895 one sometimes does not give positive effect necessary.

I use a common approach with both sensors equally:
https://github.com/lyusupov/SoftRF/blob/master/software/firmware/source/SoftRF/src/driver/DSI.cpp#L283-L341

I've attached a "merged" firmware binary for the P4.

SoftRF-P4_merged.zip

Please, try it out on the T-Display AMOLED and TFT when able.

I will appreciate if you can give me an advice on how to configure GT9895 driver to be on par with performance of the HI8561 one.

Best regards,
Linar.

P.S. Lewis, I have to mention that the T-Display needs ESP_Hosted firmware in C6 co-processor for the SoftRF firmware to operate properly.
https://github.com/Xinyuan-LilyGO/T-Display-P4/tree/main/firmware/%5BT-Display-P4%5D%5Besp32c6_slave_esp_hosted_mcu_network_adapter%5D

[lewisxhe]

I got, I'll test it when I have time.

[lewisxhe]

I've added a sample test; try it and see if it works. This test works fine for me.

[lyusupov]

; try it and see if it works.

I've tested both

1. your standalone test
2. your method been integrated into my sketch lyusupov/SoftRF@cb43aa1

Standalone test works fine.

However, after integration into the sketch, GT9895 driver still does not respond on many of actual gestures.

This is an example of 2 successful gestures.

$GNGGA,075628.000,5536.10292,N,03750.32844,E,1,14,0.8,469.8,M,14.0,M,,*42
$GNRMC,075628.000,A,5536.10292,N,03750.32844,E,77.58,208.01,130526,,,A,V*34
===== Gesture Result =====
id      : 0start   : (322, 312)end     : (323, 416)delta   : dx=1 dy=104time    : 39 mssamples : 3gesture : DOWNswipe cooldown: 600 ms
$PFLAU,0,1,2,1,0,,0,,,*4C
$GNGGA,075629.000,5536.08379,N,03750.31072,E,1,14,0.8,470.0,M,14.0,M,,*40
$GNRMC,075629.000,A,5536.08379,N,03750.31072,E,78.39,207.25,130526,,,A,V*37
$PFLAU,0,1,2,1,0,,0,,,*4C
$GNGGA,075630.000,5536.06459,N,03750.29324,E,1,14,0.8,470.1,M,14.0,M,,*4B
$GNRMC,075630.000,A,5536.06459,N,03750.29324,E,77.67,207.22,130526,,,A,V*3E
$PFLAU,0,1,2,1,0,,0,,,*4C
$GNGGA,075631.000,5536.04548,N,03750.27596,E,1,14,0.8,470.0,M,14.0,M,,*49
$GNRMC,075631.000,A,5536.04548,N,03750.27596,E,77.47,207.01,130526,,,A,V*3E
$PFLAU,0,1,2,1,0,,0,,,*4C
===== Gesture Result =====
id      : 0start   : (451, 1072)end     : (456, 1152)delta   : dx=5 dy=80time    : 34 mssamples : 2gesture : DOWNswipe cooldown: 600 ms
$GNGGA,075632.000,5536.02641,N,03750.25881,E,1,14,0.8,469.8,M,14.0,M,,*4F
$GNRMC,075632.000,A,5536.02641,N,03750.25881,E,77.10,206.91,130526,,,A,V*32
$PFLAU,0,1,2,1,0,,0,,,*4C
$GNGGA,075633.000,5536.00722,N,03750.24181,E,1,14,0.8,469.7,M,14.0,M,,*4F

With many gestures been missed - your method (V2) does not give an advantage over my previous one (V1).

[lyusupov]

@lewisxhe

However, after integration into the sketch, GT9895 driver still does not respond on many of actual gestures.

I suppose that some of TOUCH / DETOUCH events from GT9895 I2C sensor gets lost when Arduino sketch loopl() cycle is busy while doing a side load / work.

I hope to create a "minimum" example ( MCVE ) ESP32-P4 sketch this month so you could reproduce the issue at your location.

[lewisxhe]

Sorry, I wasn't able to follow up on this issue in time. If the test sketch is invalid, it's highly likely that the number of polling iterations isn't enough to sample the number of touch points, which is why the gesture isn't detected.

Try reducing the number of functions with excessive latency in the loop, or move the touch scanning to a separate task.