Distorted image

[puffpastri]

I'm trying to run the program but I'm unable to get a clean image.

When running the vortex build I get this:

And when running the rotovox build I get this, but I'm targeting the vortex build specifically:

Both the panels work perfectly when running the demos from rpi-rgb-led-matrix. The adapter I'm using is the Adafruit Triple LED Matrix Bonnet for
Raspberry Pi with HUB75, and I've tried to match up the pins in gadget_vortex.h like so:

#ifndef _GADGET_H_
#define _GADGET_H_

#define SPIN_SYNC 1

#define RGB_0_B1 7
#define RGB_0_G1 27
#define RGB_0_R1 11
#define RGB_0_B2 10
#define RGB_0_G2 9
#define RGB_0_R2 8

#define RGB_1_B1 6
#define RGB_1_G1 5
#define RGB_1_R1 12
#define RGB_1_B2 20
#define RGB_1_G2 13
#define RGB_1_R2 19

#define ADDR__EN_MASK (1<<ADDR__EN)

#define ADDR_CLK 22
#define ADDR_DAT 23
#define ADDR__EN 24

#define RGB_BLANK 18
#define RGB_CLOCK 17
#define RGB_STROBE 4
#define RGB_BLANK_MASK (1<<RGB_BLANK)
#define RGB_CLOCK_MASK (1<<RGB_CLOCK)
#define RGB_STROBE_MASK (1<<RGB_STROBE)

#define RGB_0_MASK ((1<<RGB_0_R1)|(1<<RGB_0_G1)|(1<<RGB_0_B1)|(1<<RGB_0_R2)|(1<<RGB_0_G2)|(1<<RGB_0_B2))
#define RGB_1_MASK ((1<<RGB_1_R1)|(1<<RGB_1_G1)|(1<<RGB_1_B1)|(1<<RGB_1_R2)|(1<<RGB_1_G2)|(1<<RGB_1_B2))
#define RGB_BITS_MASK (RGB_0_MASK | RGB_1_MASK)

static const int matrix_init_out[] = {RGB_0_B1, RGB_0_G1, RGB_0_R1, RGB_0_B2, RGB_0_G2, RGB_0_R2, RGB_1_B1, RGB_1_G1, RGB_1_R1, RGB_1_B2, RGB_1_G2, RGB_1_R2, ADDR_CLK, ADDR_DAT, ADDR__EN, RGB_BLANK, RGB_CLOCK, RGB_STROBE};

#define PANEL_WIDTH  128
#define PANEL_HEIGHT 64
#define PANEL_COUNT 2
#define PANEL_MULTIPLEX 2
#define PANEL_FIELD_HEIGHT (PANEL_HEIGHT / PANEL_MULTIPLEX)

#define PANEL_0_ORDER(c) (c)
#define PANEL_1_ORDER(c) (c)

#define PANEL_0_ECCENTRICITY 13.5
#define PANEL_1_ECCENTRICITY 0.375

#define VOXELS_X 128
#define VOXELS_Y 128
#define VOXELS_Z 64

#define VOXEL_Z_STRIDE 1
#define VOXEL_X_STRIDE VOXELS_Z
#define VOXEL_Y_STRIDE (VOXEL_X_STRIDE * VOXELS_X)
#define VOXELS_COUNT (VOXELS_X*VOXELS_Y*VOXELS_Z)

#define ROTATION_ZERO 286

#define CLOCK_WAITS 1

#endif

Per your advice in #2, the panels are shift register styled because --led-row-addr-type=3 was needed to get the demo working from rpi-rgb-led-matrix.

[AncientJames]

That's probably the timing. Try increasing CLOCK_WAITS - 5 worked for me, but you want it as low as it will go without turning to noise.

[puffpastri]

Thanks James, I've tried adjusting CLOCK_WAITS to various values and tested them with ./tesseract but it doesn't seem to be addressing the issue.

With CLOCK_WAITS 5:

With CLOCK_WAITS 10:

And with CLOCK_WAITS 100:

I also changed PANEL_MULTIPLEX to 4 for these tests, which seemed to reduce the noise but still produces an incorrect image. Do you have any other suggestions? The panels are also 1/32 scan, if that's relevant.

[p33tr-s]

Hey James,

I have been working on this with Sam and doing tests all day. Haven't been able to get any discernible image. The higher the clock wait, the less columns appear at once, though the columns are always a single colour
(one colour in top half of the panel, one colour in the bottom half).

All the pin mappings line up with the Adafruit Triple Bonnet, I cant seem to find any solution that changes how the image looks. Lowering the clock wait below five ca0uses the whole panel to light with flickering columns (like in the rotovox test pictured in the initial post). We are ale to successfully use RPI to get images using this line:

sudo ./demo --led-rows=64 --led-cols=128 --led-gpio-mapping=regular --led-parallel=1 --led-chain=1 --led-slowdown-gpio=4 --led-row-addr-type=3 -D 0

The panels we are using are P2 HD led video wall led screen module 128x64 hub75 smd3in1 full colour 256X128mm unit board cards LED, but we also have a pair of 81.875 128x64 HUB75E soft flexible LED panels. The latter we have not been able to get to work on RPI, and the sellers say they have only tested with a Novostar Controller. Here is where we are at:

#ifndef GADGET_H
#define GADGET_H

#define SPIN_SYNC 21

#define RGB_0_B1 7
#define RGB_0_G1 27
#define RGB_0_R1 11
#define RGB_0_B2 10
#define RGB_0_G2 9
#define RGB_0_R2 8

#define RGB_1_B1 6
#define RGB_1_G1 5
#define RGB_1_R1 12
#define RGB_1_B2 20
#define RGB_1_G2 13
#define RGB_1_R2 19

#define ADDR__EN_MASK (1<<ADDR__EN)

#define ADDR_CLK 22
#define ADDR_DAT 23
#define ADDR__EN 24

#define RGB_BLANK 18
#define RGB_CLOCK 17
#define RGB_STROBE 4
#define RGB_BLANK_MASK (1<<RGB_BLANK)
#define RGB_CLOCK_MASK (1<<RGB_CLOCK)
#define RGB_STROBE_MASK (1<<RGB_STROBE)

#define RGB_0_MASK ((1<<RGB_0_R1)|(1<<RGB_0_G1)|(1<<RGB_0_B1)|(1<<RGB_0_R2)|(1<<RGB_0_G2)|(1<<RGB_0_B2))
#define RGB_1_MASK ((1<<RGB_1_R1)|(1<<RGB_1_G1)|(1<<RGB_1_B1)|(1<<RGB_1_R2)|(1<<RGB_1_G2)|(1<<RGB_1_B2))
#define RGB_BITS_MASK (RGB_0_MASK | RGB_1_MASK)

static const int matrix_init_out[] = {RGB_0_B1, RGB_0_G1, RGB_0_R1, RGB_0_B2, RGB_0_G2, RGB_0_R2, RGB_1_B1, RGB_1_G1, RGB_1_R1, RGB_1_B2, RGB_1_G2, RGB_1_R2, ADDR_CLK, ADDR_DAT, ADDR__EN, RGB_BLANK, RGB_CLOCK, RGB_STROBE};

#define PANEL_WIDTH 128
#define PANEL_HEIGHT 64
#define PANEL_COUNT 2
#define PANEL_MULTIPLEX 2
#define PANEL_FIELD_HEIGHT (PANEL_HEIGHT / PANEL_MULTIPLEX)

#define PANEL_0_ORDER(c) (c)
#define PANEL_1_ORDER(c) (c)

#define PANEL_0_ECCENTRICITY 13.5
#define PANEL_1_ECCENTRICITY 0.375

#define VOXELS_X 128
#define VOXELS_Y 128
#define VOXELS_Z 64

#define VOXEL_Z_STRIDE 1
#define VOXEL_X_STRIDE VOXELS_Z
#define VOXEL_Y_STRIDE (VOXEL_X_STRIDE * VOXELS_X)
#define VOXELS_COUNT (VOXELS_XVOXELS_YVOXELS_Z)

#define ROTATION_ZERO 286

#define CLOCK_WAITS 5

#endif

Thanks for the help!

Edit: I've also found that changing from shift registry to ABCDE on the Vortex gadget produces practically the exact same result which doesn't make sense to me.

[AncientJames]

It looks as though these panels are different enough that you'll need to make some code changes.
I remember when I got my panels going, they didn't work straight off with rpi-rgb - I had to first set (or possibly clear) the 'C' ('ADDR__EN') pin. That's proably a good place to start - try inverting it (around line 621 in driver/vortex.c - move 'ADDR__EN_MASK from set_bits to clear_bits, and the other way at line 629).
Failing that, it'll come down to looking at exactly how rpi-rgb is doing it - once you have the sequence, it's not a big job to implement it.

[p33tr-s]

Thanks for the advice, I been going down this hole for a while to no avail. I took your advice about inverting ADDR_EN_MASK, but only seem to get a small flash then a black display. If you can remember, how did you find the exact sequence RPI was using to get your panels to work. Thanks!

[AncientJames]

I didn't do anything smarter than wiggling the bits in various orders until something displayed.
Looking at https://github.com/hzeller/rpi-rgb-led-matrix/blob/d8778b5f044b216acef2dec723bc9bdcd227cd8d/lib/framebuffer.cc#L245 (which I believe is 'type 3'), the comment says shift register row address setter that does not use B but C for the data. Clock is inverted. - you could try swapping the gpio mapping in the gadget header (A is addr clock, C is addr data, and maybe addr en isn't used?)

[mayakhalil-cmd]

i dont know if you resolved this yet but what is your chipset? some chipsets are not compatible with the hzeller library so double check if yours is! apologies if you already figured it out!

[p33tr-s]

I haven't resolved this yet :') I've been taking a break from it for a bit. I can't remember the exact chipset, but it was compatible with hzellers, we were able to get a lovely rotating square. I have a different set of panels that are ICN2037 that I would've loved to use, but they have proved incompatible. I think with the right code I can get my current panels to work on vortex, only problem is I know next to nothing about coding. ChatGPT will only take you so far.

[mayakhalil-cmd]

Good to know! I just got mine to light up recently so if I got the code working, I’ll come back here and share it with you! All the best in the meantime :)!

[mayakhalil-cmd]

@p33tr-s @puffpastri HI! hope youre good! i had a similar issue recently and someone helped me configure and calibrate it.

putting the code here in case it helps anyone :D

#ifndef GADGET_H
#define GADGET_H

#define SPIN_SYNC 2

#define RGB_0_B2 7
#define RGB_0_G2 27
#define RGB_0_R2 11
#define RGB_0_B1 10
#define RGB_0_G1 9
#define RGB_0_R1 8

#define RGB_1_B2 6
#define RGB_1_G2 5
#define RGB_1_R2 12
#define RGB_1_B1 20
#define RGB_1_G1 13
#define RGB_1_R1 19

#define ROW_A 22
#define ROW_B 23
#define ROW_C 24
#define ROW_D 25
#define ROW_E 15
#define ADDR__EN_MASK (0)

#define RGB_BLANK 18
#define RGB_CLOCK 17
#define RGB_STROBE 4
#define RGB_BLANK_MASK (1<<RGB_BLANK)
#define RGB_CLOCK_MASK (1<<RGB_CLOCK)
#define RGB_STROBE_MASK (1<<RGB_STROBE)

#define RGB_0_MASK ((1<<RGB_0_R1)|(1<<RGB_0_G1)|(1<<RGB_0_B1)|(1<<RGB_0_R2)|(1<<RGB_0_G2)|(1<<RGB_0_B2))
#define RGB_1_MASK ((1<<RGB_1_R1)|(1<<RGB_1_G1)|(1<<RGB_1_B1)|(1<<RGB_1_R2)|(1<<RGB_1_G2)|(1<<RGB_1_B2))
#define RGB_BITS_MASK (RGB_0_MASK | RGB_1_MASK)

#define ROW_MASK ((1<<ROW_A)|(1<<ROW_B)|(1<<ROW_C)|(1<<ROW_D)|(1<<ROW_E))

//static const int matrix_init_out[] = {RGB_0_R1, RGB_0_G1, RGB_0_B1, RGB_0_R2, RGB_0_G2, RGB_0_B2, RGB_1_R1, RGB_1_G1, RGB_1_B1, RGB_1_R2, RGB_1_G2, RGB_1_B2, ROW_A, ROW_B, ROW_C, ROW_D, ROW_E, RGB_BLANK, RGB_CLOCK, RGB_STROBE};
static const int matrix_init_out[] = {RGB_0_B1, RGB_0_G1, RGB_0_R1, RGB_0_B2, RGB_0_G2, RGB_0_R2, RGB_1_B1, RGB_1_G1, RGB_1_R1, RGB_1_B2, RGB_1_G2, RGB_1_R2, ROW_A, ROW_B, ROW_C, ROW_D, ROW_E, RGB_BLANK, RGB_CLOCK, RGB_STROBE};

#define PANEL_WIDTH 128
#define PANEL_HEIGHT 64
#define PANEL_COUNT 2
#define PANEL_MULTIPLEX 2
#define PANEL_FIELD_HEIGHT (PANEL_HEIGHT / PANEL_MULTIPLEX)

#define PANEL_0_ORDER(c) (c)
#define PANEL_1_ORDER(c) (c)

#define VOXELS_X 128
#define VOXELS_Y 128
#define VOXELS_Z 64

#define PANEL_0_ECCENTRICITY 13.5
#define PANEL_1_ECCENTRICITY 0.375

#define VOXEL_Z_STRIDE 1
#define VOXEL_X_STRIDE VOXELS_Z
#define VOXEL_Y_STRIDE (VOXEL_X_STRIDE * VOXELS_X)
#define VOXELS_COUNT (VOXELS_XVOXELS_YVOXELS_Z)

#define ROTATION_ZERO 160

#define CLOCK_WAITS 6

#endif

im using gadget_vortex btw and my photointerrupter is on gpio 2. best of luck and hope it gets resolved for you