Issues with iot_i2s_lcd_write_cmd in 8-bit LCD mode (AEGHB-293)

[csobrinho]

INSTRUCTIONS

Environment

• Development Kit: Hitec WiFi Kit 32
• Module or chip used: ESP32
• IDF version (run git describe --tags to find it):
  v4.1-dev-1935-g647cb628a
• Build System: CMake
• Compiler version (run xtensa-esp32-elf-gcc --version to find it):
  xtensa-esp32-elf-gcc (crosstool-NG esp-2019r2) 8.2.0
• Operating System: macOS
• Power Supply: USB

Problem Description

Having some issues with the iot_i2s_lcd_write_cmd component in 8 bit mode. Also the clock I2S[i2s_num]->clkm_conf.clkm_div_num = 2; doesn't seem to work reliably with 8 bit mode. Had to use 4.

Expected Behavior

iot_i2s_lcd_write_cmd (that uses the FIFO) should work when a DMA is also underway. My impression is that in 8 bit mode, the I2S[i2s_num]->state.tx_idle doesn't get updated quickly and the second write of the LOW byte cancels the previous one. I believe that adding a small us delay fixed it but my memory is not very sure here.

Actual Behavior

A few issues that I noticed and validated with the MSO1074z logic analyzer:

• iot_i2s_lcd_write_cmd doesn't seem to work properly. I can see the RS line going down and up but no data is written. I confirmed this with the LA. Replacing it with the i2s_lcd8_write_data and toggling the RS line manually fixed this issue (https://github.com/csobrinho/hydroponics/blob/11b9b3b9bd3c9ae62058ade886d847e1b19deb0f/main/driver/i2s_lcd8.c#L102-L104). This problem was independent on the clock and was the same for clkm_div_num = {2|4|8}

• I2S[i2s_num]->clkm_conf.clkm_div_num = 2 should be 3 or 4 for 8 bit mode. From what I could understand from the RTM, from the source code comments and from the forum, for 8 bits the clock is divided by two and must be 3 or 4. With 3 I got some issues during reboot where in some cases the byte was transmitted and in other cases it wasn't. I also remember reading some info regarding 8bit mode and I2S1 being better than I2S0 but I can't find my reference.

This board has the CPUs set at 240 MHz so not sure if that was the reason but it was strange. If I defined a bus with of 16 I saw the data changing but I missing half of the data.

Steps to reproduce

1. call iot_i2s_lcd_write_cmd.
2. call iot_i2s_lcd_write_data. Only the rs toggle and the iot_i2s_lcd_write_data was properly transmitted.

From https://github.com/csobrinho/hydroponics/blob/master/main/display/lcd.c#L18-L32:

static i2s_lcd8_dev_t dev = {
        .base = {
                .i2s_lcd_conf = {
                        .data_width = 8,
                        .data_io_num = {
                                GPIO_NUM_23, GPIO_NUM_18, GPIO_NUM_14, GPIO_NUM_27,
                                GPIO_NUM_26, GPIO_NUM_21, GPIO_NUM_22, GPIO_NUM_19,
                        },
                        .ws_io_num = GPIO_NUM_17,
                        .rs_io_num = GPIO_NUM_12,
                },
                .i2s_port = I2S_NUM_1,
        },
        .rst_io_num = OLED_RESET,
};

Code to reproduce this issue

I have made a wrapper around the iot_i2s_lcd component here: https://github.com/csobrinho/hydroponics/blob/master/main/driver/i2s_lcd8.c

Debug Logs

N/A but can provide upon request.

Other items if possible

There are other issues (#19) and (#26) with LCD in 8 bit mode but I believe they are not related to this particular issue.

I'm using a RM68090 display and it accepts 8 bit commands sent as 0b00000000{cmd} bits so no need to change the function to only send one byte.

• sdkconfig file (attach the sdkconfig file from your project folder)
  https://github.com/csobrinho/hydroponics/blob/master/sdkconfig

• elf file in the build folder (note this may contain all the code details and symbols of your project.)

• coredump (This provides stacks of tasks.)

[InfiniteYuan]

This component has not been adapted for other IDF versions.

I hope you understand that it will be adapted to other versions in the near future.

[nopnop2002]

@csobrinho

I don't know your TFT driver, but give it a try.

It's work using esp-idf version:v4.3-dev-2586

https://github.com/nopnop2002/esp-idf-parallel-tft

[csobrinho]

Thanks nopnop2002. I still have my I2S LCD driver working but I'm moving to a MHS40 screen with an 80 MHz SPI interface so I'll soon deprecate that driver. I did manage to get really fast speeds though.

[leeebo]

From ESP-IDF v5.0, we suggest using the new general LCD driver https://github.com/espressif/esp-idf/tree/master/examples/peripherals/lcd