SeeedStudio SenseCAP Indicator - SPI question

[u4mzu4]

Hello, and first of all thanks for this nice library!
I have a new gadet from SeeedStudio, a SenseCAP Indicator. It's an ESP32-S3 based 4" display, looks like a simple ESP32-S3-RGB.
However!
For some mysterious reason the SPI_CS and the LCD_RST pins were outsourced to a PCA9535 I2C expander, you can find a block diagram HERE.

I've seen that CS can be GFX_NOT_DEFINED (set fix LOW), so my setting are the following:

#define GFX_BL 45
Arduino_DataBus *bus = new Arduino_SWSPI(
    GFX_NOT_DEFINED /* DC */, GFX_NOT_DEFINED /* CS */,
    41 /* SCK */, 48 /* MOSI */, GFX_NOT_DEFINED /* MISO */);

// option 1:
// Uncomment for 4" rect display
Arduino_ESP32RGBPanel *rgbpanel = new Arduino_ESP32RGBPanel(
    18 /* DE */, 17 /* VSYNC */, 16 /* HSYNC */, 21 /* PCLK */,
    4 /* R0 */, 3 /* R1 */, 2 /* R2 */, 1 /* R3 */, 0 /* R4 */,
    10 /* G0 */, 9 /* G1 */, 8 /* G2 */, 7 /* G3 */, 6 /* G4 */, 5 /* G5 */,
    15 /* B0 */, 14 /* B1 */, 13 /* B2 */, 12 /* B3 */, 11 /* B4 */,
    1 /* hsync_polarity */, 10 /* hsync_front_porch */, 8 /* hsync_pulse_width */, 50 /* hsync_back_porch */,
    1 /* vsync_polarity */, 10 /* vsync_front_porch */, 8 /* vsync_pulse_width */, 20 /* vsync_back_porch */);
Arduino_RGB_Display *gfx = new Arduino_RGB_Display(
    480 /* width */, 480 /* height */, rgbpanel, 0 /* rotation */, true /* auto_flush */,
    bus, GFX_NOT_DEFINED /* RST */, st7701_type1_init_operations, sizeof(st7701_type1_init_operations));

Compiled PDQgraphicstest and running, the Serial output is the following:

Benchmark micro-secs
Screen fill 85303
Text 28389
Pixels 477361
Lines 598158
Horiz/Vert Lines 81183
Rectangles (filled) 496544
Rectangles (outline) 67436
Triangles (filled) 209196
Triangles (outline) 45312
Circles (filled) 48793
Circles (outline) 57702
Arcs (filled) 48146
Arcs (outline) 151570
Rounded rects (filled) 535074
Rounded rects (outline) 73095
Done!

Blacklight is OK, but I can not see anything on the display. Looks like the init was wrong, is there any solution to bitbang the CS (eg. though some macro), or did I screw something else up?

[u4mzu4]

Ooh-ahh, it's working!
I wrote (messed up) my own Arduino_DataBus, which is a mixture of Arduino_SWSPI and Arduino_XL9535SWSPI (for CS). Ugly source code, full of inelegant solutions, but it's working.
@moononournation: if you need "official" support for this device, we can clean/merge the source code, anyway I can go further with my own workaround.

[u4mzu4]

Hello!
I just used the driver https://github.com/semcneil/PCA95x5, but the "builtin" XL9535 would be better I think.
Here is my working demo (just rename to .cpp):
Arduino_SWSPI.txt

The SPI speed is not critical because only the init happens there.
If you want to rotate the display you can use the X-direction control command below as well.

[mc32768]

Thanks so much! It turns out I had done almost exactly the same thing as you, even using the same PCA95x5 library. The mistake was that I forgot to specify the pins in Wire.begin(39, 40, 400000) so that is why the PCA9535 wasn't working for me. It was really helpful to be able to check out your code and I appreciate your willingness to share it.

One minor difference I noticed was that the Seeed ESP-IDF sample code (components\bsp\src\boards\sensecap_indicator_board.c) seems to leave the PCA9535 pins in their default state (inputs) and then it specifically sets only 4,5,7,8 and 10 as outputs. Whereas in the demo code linked above, it is setting all pins to outputs (ioex.direction(PCA95x5::Direction::OUT_ALL)). Seeed's block diagram does show a one-way arrow coming from the PCA9535, implying all pins are outputs, but when I looked at the touch panel datasheet it says the INT pin for the touch panel (IO0_6 on PCA9535) is an output of the touch panel, so if it's also set as an output of the PCA9535, that could cause a conflict. It might be safer to set only specific known pins as outputs and leave the rest as inputs.

I am glad to finally see the Sensecap Indicator working - the device is new, so it's hard to find documentation or functional sample code at the moment, and the design seems to have quite a few quirks. Thanks again for taking the time to help me with this!

[u4mzu4]

@mc32768
I'm glad that you was able to use my trashy workaround. That would be great if we can clenaup and add as official solution. Anyway, regardinig your issue with the directions: if you check the block diagram carefully, you can find the GPIO45 as interrupt of the touch panel, which is really nice (instead of polling in every n millis). However you can find this GPIO as above
#define GFX_BL 45
so, this is a clear output for backlight (even in the Seeed code base), and never used (unforunately) as input.
So summary: there is no input for PCA9535 unfortunately.

[mc32768]

I wonder if there may be a mistake in that block diagram, since they have ESP32 GPIO45 shown in two places. It is shown as PWM backlight control (output of ESP32) and as PCA9535 interrupt (input of ESP32, output of PCA9535). It does not seem logical that those two functions would coexist on the same pin. I agree with you that it appears to be used for backlight.

In my earlier message, the input signal I was referring to is between the PCA9535 and the touch panel controller (TP_INT on IO0_6 of the PCA9535). I think that one should be configured as an input to the PCA9535 since it is definitely an output of the touch panel controller. It would be bad if both devices (PCA9535 and touch panel controller) were driving it concurrently so that's why I was thinking it should be configured as an input on PCA9535, even if it's never used in code anywhere else.

I haven't actually gotten to the point of making the touch panel work yet so I might know more if I can proceed to the point where I need to use it.

[u4mzu4]

You're right, IO0_6 is an input on the PCA9535, I've used as input as well when I played with the touch driver, and the elegant configuration is the input. My pain is that it can't be used for anything.

[u4mzu4]

In the meantime I've just run into a new issue: the st7701_type1_init_operations is upside down on this device, and the SW rotation (2 /* rotation */) skips a few lines at the bottom of the display in case of big/full frame buffer.
I've checked the original SeeedStudio init sequence and found the line:
WRITE_C8_D8, 0xC7, 0x04,
which is a
SDIR (C7h/C700): X-direction Control
based on the ST7701S datasheet.
So maybe you can use this above as a 180° HW rotation solution.

[chalei]

WRITE_C8_D8, 0xC7, 0x04,

where should I put this in Arduino_RGB_Display.h file

static const uint8_t st7701_type1_init_operations[] = {
    BEGIN_WRITE,
    WRITE_COMMAND_8, 0xFF,
    WRITE_BYTES, 5, 0x77, 0x01, 0x00, 0x00, 0x10,

    WRITE_C8_D16, 0xC0, 0x3B, 0x00,
    WRITE_C8_D16, 0xC1, 0x0D, 0x02,
    WRITE_C8_D16, 0xC2, 0x31, 0x05,
    WRITE_C8_D8, 0xCD, 0x08,

    WRITE_COMMAND_8, 0xB0, // Positive Voltage Gamma Control
    WRITE_BYTES, 16,
    0x00, 0x11, 0x18, 0x0E,
    0x11, 0x06, 0x07, 0x08,
    0x07, 0x22, 0x04, 0x12,
    0x0F, 0xAA, 0x31, 0x18,

    WRITE_COMMAND_8, 0xB1, // Negative Voltage Gamma Control
    WRITE_BYTES, 16,
    0x00, 0x11, 0x19, 0x0E,
    0x12, 0x07, 0x08, 0x08,
    0x08, 0x22, 0x04, 0x11,
    0x11, 0xA9, 0x32, 0x18,

    // PAGE1
    WRITE_COMMAND_8, 0xFF,
    WRITE_BYTES, 5, 0x77, 0x01, 0x00, 0x00, 0x11,

    WRITE_C8_D8, 0xB0, 0x60, // Vop=4.7375v
    WRITE_C8_D8, 0xB1, 0x32, // VCOM=32
    WRITE_C8_D8, 0xB2, 0x07, // VGH=15v
    WRITE_C8_D8, 0xB3, 0x80,
    WRITE_C8_D8, 0xB5, 0x49, // VGL=-10.17v
    WRITE_C8_D8, 0xB7, 0x85,
    WRITE_C8_D8, 0xB8, 0x21, // AVDD=6.6 & AVCL=-4.6
    WRITE_C8_D8, 0xC1, 0x78,
    WRITE_C8_D8, 0xC2, 0x78,

    WRITE_COMMAND_8, 0xE0,
    WRITE_BYTES, 3, 0x00, 0x1B, 0x02,

    WRITE_COMMAND_8, 0xE1,
    WRITE_BYTES, 11,
    0x08, 0xA0, 0x00, 0x00,
    0x07, 0xA0, 0x00, 0x00,
    0x00, 0x44, 0x44,

    WRITE_COMMAND_8, 0xE2,
    WRITE_BYTES, 12,
    0x11, 0x11, 0x44, 0x44,
    0xED, 0xA0, 0x00, 0x00,
    0xEC, 0xA0, 0x00, 0x00,

    WRITE_COMMAND_8, 0xE3,
    WRITE_BYTES, 4, 0x00, 0x00, 0x11, 0x11,

    WRITE_C8_D16, 0xE4, 0x44, 0x44,

    WRITE_COMMAND_8, 0xE5,
    WRITE_BYTES, 16,
    0x0A, 0xE9, 0xD8, 0xA0,
    0x0C, 0xEB, 0xD8, 0xA0,
    0x0E, 0xED, 0xD8, 0xA0,
    0x10, 0xEF, 0xD8, 0xA0,

    WRITE_COMMAND_8, 0xE6,
    WRITE_BYTES, 4, 0x00, 0x00, 0x11, 0x11,

    WRITE_C8_D16, 0xE7, 0x44, 0x44,

    WRITE_COMMAND_8, 0xE8,
    WRITE_BYTES, 16,
    0x09, 0xE8, 0xD8, 0xA0,
    0x0B, 0xEA, 0xD8, 0xA0,
    0x0D, 0xEC, 0xD8, 0xA0,
    0x0F, 0xEE, 0xD8, 0xA0,

    WRITE_COMMAND_8, 0xEB,
    WRITE_BYTES, 7,
    0x02, 0x00, 0xE4, 0xE4,
    0x88, 0x00, 0x40,

    WRITE_C8_D16, 0xEC, 0x3C, 0x00,

    WRITE_COMMAND_8, 0xED,
    WRITE_BYTES, 16,
    0xAB, 0x89, 0x76, 0x54,
    0x02, 0xFF, 0xFF, 0xFF,
    0xFF, 0xFF, 0xFF, 0x20,
    0x45, 0x67, 0x98, 0xBA,

    //-----------VAP & VAN---------------
    WRITE_COMMAND_8, 0xFF,
    WRITE_BYTES, 5, 0x77, 0x01, 0x00, 0x00, 0x13,

    WRITE_C8_D8, 0xE5, 0xE4,

    WRITE_COMMAND_8, 0xFF,
    WRITE_BYTES, 5, 0x77, 0x01, 0x00, 0x00, 0x00,

    WRITE_COMMAND_8, 0x21,   // 0x20 normal, 0x21 IPS
    WRITE_C8_D8, 0x3A, 0x60, // 0x70 RGB888, 0x60 RGB666, 0x50 RGB565

    WRITE_COMMAND_8, 0x11, // Sleep Out
    END_WRITE,

    DELAY, 120,

    BEGIN_WRITE,
    WRITE_COMMAND_8, 0x29, // Display On
    END_WRITE};

is it in the beginning of the st7701 definition or the end?

[u4mzu4]

@chalei

This is the original Seed solution: