Some problem with i2s_lcd_write_data() about esp-iot-solution HOT 8 CLOSED

I think we should use the uint16_t array to test the i2s of the 16-bit data width. For example:

uint16_t fb[10] = {0xabab, 0xabab, 0x55af, 0x3d6f, 0x4478, 0xabab, 0xabab, 0x55af, 0x3d6f, 0x4478};

from esp-iot-solution.

I have found an error from hardware. GPIO0 should not be used as LCD_D2_PIN from ESP32-PICO-D4 for some unknown reason(s). Comparing the byte mismatch it is always the GPIO 0 pulled to 0. So I have changed to GPIO9 as LCD_D2_PIN. #define LCD_D2_PIN (9) // 0->9. And I have tested with uint16_t arrays with result below:

Src code:

DRAM_ATTR static const uint16_t fb[10] = {0xabcd, 0xcdba, 0xab88, 0xac23, 0x5544, 0xaf54, 0x3d2a, 0x6f56, 0x4423, 0x78fa};

    for(;;){
    	i2s_lcd_write_data(d2805_handle, (const uint16_t *)&fb, sizeof(fb), portTICK_PERIOD_MS, false);
    	vTaskDelay(500 / portTICK_PERIOD_MS);
    }

Waveform now matches perfectly but there is an error when compiling because it is a const char *src in prototype of i2s_lcd_write_data(). The error as: **expected 'const char *' but argument is of type 'const uint16_t * {aka const short unsigned int *}'**

Another test with uint8_t array with data width changed to 8 does not sound correct:

void app_main()
{
    dummy_init();

    i2s_lcd_config_t d2805_conf={
    		8,	//data_width = 8
			{
			        LCD_D0_PIN, LCD_D1_PIN, LCD_D2_PIN, LCD_D3_PIN,
			        LCD_D4_PIN, LCD_D5_PIN, LCD_D6_PIN, LCD_D7_PIN,
			        //LCD_D8_PIN, LCD_D9_PIN, LCD_D10_PIN,LCD_D11_PIN,
			        //LCD_D12_PIN, LCD_D13_PIN, LCD_D14_PIN, LCD_D15_PIN
			},
			LCD_WR_PIN,
			LCD_RS_PIN
    };

    i2s_lcd_handle_t d2805_handle = i2s_lcd_create(I2S_PORT_NUM, &d2805_conf);
    DRAM_ATTR static const char fb[10] = {0xab, 0xcd, 0xab, 0x23, 0x44, 0xaf, 0x3d, 0x56, 0x42, 0xfa};

    for(;;){
    	i2s_lcd_write_data(d2805_handle, (const char *)&fb, sizeof(fb), portTICK_PERIOD_MS, false);
    	vTaskDelay(500 / portTICK_PERIOD_MS);
    }
}

Screen shot from Logic analyzer:

According to the probe result the byte sequence sent from LCD_D0_PIN[7:0] are 0xab 0xab 0x44 0x3d 0x42

Can't we send with 8-bit 8080?

from esp-iot-solution.

I think the reason why 8-bit didn't work is now clear. Needed to put #define CONFIG_BIT_MODE_8BIT right at the beginning of i2s_lcd.c to let the system know we are using 8-bit. So I believe the default is 16-bit with definition set somewhere...
#define CONFIG_BIT_MODE_8BIT is put right next to #include "esp_log.h"

Clean and make flash will give the right waveform for 8-bit:

Thank you

from esp-iot-solution.

Some more findings. It seems there is a problem with iot_i2s_lcd_write_data(i2s_lcd_handle_t i2s_lcd_handle, uint16_t data) when data width is set to 8BIT. In i2s_lcd_com.c the code is :

#ifdef CONFIG_BIT_MODE_8BIT

void iot_i2s_lcd_write_data(i2s_lcd_handle_t i2s_lcd_handle, uint16_t data)
{
    i2s_lcd_write_data(i2s_lcd_handle, (char *)&data, 2, 100, true);
}

When 8 bit is defined, data type should be uint8_t, and the 3rd argument should be 1 instead of 2?

Another issue is that, RS pin is never strobed when iot_i2s_lcd_write_cmd(i2s_lcd_handle_t i2s_lcd_handle, uint16_t cmd) is called.

from esp-iot-solution.

The problem with RS pin never strobed has been solved. Mistake has been made with calling i2s_lcd_create() without setting the direction of RS pin which is treated as a gpio. Instead of i2s_lcd_create the function iot_i2s_lcd_pin_cfg() should be called which include both i2s_lcd_create() & gpio_set_direction() for RS pin.

The root of another problem with iot_i2s_lcd_write_cmd(i2s_lcd_handle_t i2s_lcd_handle, uint16_t cmd) for CONFIG_BIT_MODE_8BIT seems to be a 8bit left shift (cmd>>8) as follows:

void iot_i2s_lcd_write_cmd(i2s_lcd_handle_t i2s_lcd_handle, uint16_t cmd)
{
    i2s_lcd_t *i2s_lcd = (i2s_lcd_t *)i2s_lcd_handle;
    i2s_port_t i2s_num = i2s_lcd->i2s_port;
    GPIO.out_w1tc = (1 << i2s_lcd->i2s_lcd_conf.rs_io_num);
    REG_WRITE(I2S_FIFO_ADD[i2s_num], (cmd >> 8));
    I2S[i2s_num]->conf.tx_start = 1;

This function assumed a uint16_t cmd and user should allocate the high byte for the actual command. Example, if 0x2A is wanted one need to send like iot_i2s_lcd_write_cmd(i2s_lcd_handle, 0x2A00), but it turns out that a dummy 0x00 is also sent in the tail after 0x2A. Below is a screen shot from analyzer:

from esp-iot-solution.

Hi
After much effort in understanding how i2s_8080 work, it seems 8 bit parallel bus is working OK with modification to two functions in i2s_lcd_com.c.

(1)

#ifdef CONFIG_BIT_MODE_8BIT

void iot_i2s_lcd_write_data(i2s_lcd_handle_t i2s_lcd_handle, uint16_t data)
{
    //i2s_lcd_write_data(i2s_lcd_handle, (char *)&data, 2, 100, true); //original
    i2s_lcd_write_data(i2s_lcd_handle, (char *)&data, 1, 100, false); //new
}

(2)

void iot_i2s_lcd_write_cmd(i2s_lcd_handle_t i2s_lcd_handle, uint16_t cmd)
{
    i2s_lcd_t *i2s_lcd = (i2s_lcd_t *)i2s_lcd_handle;
    i2s_port_t i2s_num = i2s_lcd->i2s_port;
    GPIO.out_w1tc = (1 << i2s_lcd->i2s_lcd_conf.rs_io_num);

    REG_WRITE(I2S_FIFO_ADD[i2s_num], (uint8_t)cmd); //new
    I2S[i2s_num]->conf.tx_start = 1;
    while (!(I2S[i2s_num]->state.tx_idle)) {
        // vTaskDelay(20);
        ;
    }
    I2S[i2s_num]->conf.tx_start = 0;
    I2S[i2s_num]->conf.tx_reset = 1;
    I2S[i2s_num]->conf.tx_reset = 0;
    I2S[i2s_num]->conf.tx_fifo_reset = 1;
    I2S[i2s_num]->conf.tx_fifo_reset = 0;

    GPIO.out_w1ts = (1 << i2s_lcd->i2s_lcd_conf.rs_io_num);
}

Repeat sending command & data in 8-bit 8080 with snippet :

void app_main()
{
    i2s_lcd_config_t d2805_conf={
    		8,	//data_width = 8
			{
			        LCD_D0_PIN, LCD_D1_PIN, LCD_D2_PIN, LCD_D3_PIN,
			        LCD_D4_PIN, LCD_D5_PIN, LCD_D6_PIN, LCD_D7_PIN,
			},
			LCD_WR_PIN,
			LCD_DC_PIN
    };

    i2s_lcd_handle_t d2805_handle = iot_i2s_lcd_pin_cfg(I2S_PORT_NUM, &d2805_conf);
    for(;;){
    	iot_i2s_lcd_write_cmd(d2805_handle, 0x2A);
    	iot_i2s_lcd_write_data(d2805_handle, 0xef);
    }
}

Dump screen from logic analyzer is shown here.

from esp-iot-solution.

Please see #19

from esp-iot-solution.

The code in 8-BIT 8080 mode is working. Picture shown below for my LCD.

With a logic analyzer WR# pulse frequency is measured to be ~ 13MHz. I can see that the WR# pulse rate is actually BCK clock rate. Its frequency is calculated by Fbck = Fpll /(N + (b/a))/M.

Inside esp_err_t i2s_set_parallel_mode(i2s_port_t i2s_num),

I2S[i2s_num]->clkm_conf.clkm_div_b = 1; 
I2S[i2s_num]->clkm_conf.clkm_div_a = 0;

I don't understand it when clkm_conf.clkm_div_a is set '0', b/a would be undefined. Why a 13MHz is obtained?

So I tried

I2S[i2s_num]->clkm_conf.clkm_div_b = 0; 
I2S[i2s_num]->clkm_conf.clkm_div_a = 1;
``` hoping to get a max. bck rate = 160/2/2 = 40MHz with N=M=2 but the screen no longer works. 

from esp-iot-solution.