rgb_panel project built successfully but nothing shows on the panel

[Hazeline2018]

I used the exactly same setup as the IceBreaker ICE40 dev board with 64x64 panel. I was able to build rgb_panel project with Pattern mode successfully, but nothing showed up on the panel. it's pure blank. I read through the RTL, didn't see anywhere specifies the scan mode, eg 1/32 or 1/16. The only parameters I can see as follows in "top.v":

// Params

ifdef BOARD_ICEBREAKER_SINGLE2X localparam integer N_BANKS = 4; else localparam integer N_BANKS = 2; `endif localparam integer N_ROWS = 32; localparam integer N_COLS = 64; localparam integer N_CHANS = 3; localparam integer N_PLANES = 10; localparam integer BITDEPTH = 16;

localparam integer LOG_N_BANKS = $clog2(N_BANKS);
localparam integer LOG_N_ROWS  = $clog2(N_ROWS);
localparam integer LOG_N_COLS  = $clog2(N_COLS);

Since I'm using 64x64 panel, should I change N_ROWS to 64? Tried, but still nothing shows. I also noticed that in the "pgen.v" module, these paras are hard-coded:

module pgen #( parameter integer N_ROWS = 64, // # of rows (must be power of 2!!!) parameter integer N_COLS = 64, // # of columns parameter integer BITDEPTH = 24,

// Auto-set
parameter integer LOG_N_ROWS  = $clog2(N_ROWS),
parameter integer LOG_N_COLS  = $clog2(N_COLS)

)(

I tried other simple examples from other projects, the panel light up just fine, altering color blocks to test the panel and setup fine. Please any help would be highly appreciated!

[smunaut]

64x64 panels are still 1:32 and that's specified using N_ROWS (which is the number of multiplexed rows). But it pushes 2 rows at once (upper and lower part of the panels) which is called N_BANKS here. So the default config is good for 64x64 panels.

A few questions :

• What kind of panels do you have ? (especially which driver chip)
• Are you sure the connections are right ? It's easy to plug something wrong ... either feeding the panel output, or having the header the wrong way around ...

[Hazeline2018]

Thanks so much for your prompt response! I bought the 64x64 panel from Crowdsupply along with the iCEBreaker FPGA board. I checked the panel driver ,it's ICN2037 shift driver. Yes, I'm quite sure the connections are good. I verified quite a few times. I even tested with another very simple project to have the entire panel alter R/G/B blocks which works fine.

[Hazeline2018]

I checked the panel driving logic, it is 1/32 scan. The row driver is GM7258 and col driver is ICN2037 which are all standard LED panel drivers.

[smunaut]

Can you try the attached bitstream ? (to exclude any build or toolchain issue)

rgb_panel.zip

[Hazeline2018]

An update, I have managed to get the project working by trying an older version of hub75_top.v although didn't really figured out what was going wrong with earlier attempts. I then resyn the whole sandbox and tried again and everything works fine now. So I moved on. :-)

However, when I'm trying the video mode, I was able to flash in my own video contents (over 30s long). But when playing back, it's playing backwards, like rewinding. Tried the up/down button function, that works as expected. I guess it might be due to wrong frame counter calculation? Since the iCEBreaker FPGA board I use has on-board 16MB flash, which is large enough to fit in mins long video content, so it's not the memory overflow/wrap-around issue for sure. Any suggestions? thanks!

[smunaut]

So it works with the latest version ? Or did you have to stay on the older one ?

For the play direction, yeah look like I might have swapped them ... Look in vgen.v and change :

frame_sel <= cnt_rep_last ? 2'b10 : 2'b00; to frame_sel <= cnt_rep_last ? 2'b11 : 2'b00;

[Hazeline2018]

The latest repo works.

Okay, will try your suggestion and update soon. Thanks a lot!

Sent from my iPhone

On Jan 16, 2021, at 11:20 PM, smunaut notifications@github.com wrote:

 So it works with the latest version ? Or did you have to stay on the older one ?

For the play direction, yeah look like I might have swapped them ... Look in vgen.v and change :

frame_sel <= cnt_rep_last ? 2'b10 : 2'b00; to frame_sel <= cnt_rep_last ? 2'b11 : 2'b00;

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[Hazeline2018]

Yep, that swapping direction fixes the problem! Thanks a lot for the great help!

I have a few other questions though and like to discuss with you further. But I'll create another thread so you can close this one if you like. Thanks again!

[smunaut]

You should instead join the 1bitsquared discord channel to discuss things. ( https://1bitsquared.com/pages/chat ) or alternatively the ##openfpga IRC channel on freenode.

Trying to keep the issues here for actual bugs.

[Hazeline2018]

all right, I'll join !