Red background on hdmi with v9990 implementation on WT

[Msxinfo]

I have a bunch of Tang Nano 20 boards that I flashed with the tnCart Wonder firmware so I can use V9990 emulation on my WonderTang 2.00b cartridges.

I noticed that on quite a few boards I get a red background where it should be black. For instance with Eggbert and the Tiny Yarou v9990 demo. Any idea what the problem can be? I

[Msxinfo]

[Msxinfo]

[buppu3]

Does this happen every time?
Or is this a phenomenon that reproduces itself after certain operations?

[Msxinfo]

This happens every time, but only on specific Tang Nano 20K. Some work fine, but others have this strange behaviour.

[herraa1]

Is the issue related to specific Tang Nano 20k boards or to the WonderTang PCBs?
Does the same Tang Nano 20k board always show a red background irrespective of which PCB are you using (change the Tang Nano 20k board between your available PCBs)?
And the Tang Nano 20k boards that work fine, do they work fine on PCBs where other Tang Nano 20k boards show a red background?
Are those Tang Nano 20k boards that show a red background from different batches (or do they have a slightly different model number)?

[Msxinfo]

It happens both on 1.02d as 2.0b boards. It is not the WonderTang hardware, because when I replace the Tn20K the problem is sometimes gone or is still there. I have tested a lot of Tang Nano 20k boards from different batches (about 15). With or without the pin headers presoldered. The TN20k remains giving red backgrounds on different WT pcb's.

Could it be related to the flashing in Windows with certain Gowin programmer versions? What version are you using? One time I flashed a 'red' TN20k with another firmware and the problem was not there, but that didn't work on another TN20K. Could it be the incompleteness of the v9990 emulation?

[herraa1]

I don't think the programmer should be an issue.
I use Linux and the GOWIN EDA V1.9.9.03 Education Build(73833).

Are you using the .fs from this repo's main branch on 1.02d? Or are you building your own .fs?
Are the "red" Tang Nano 20K new or used?
I assume that the red background is just on the HDMI output of the V9990.

[Msxinfo]

All the TN20k are new. I use your wondertang2.0b firmware and also 1.02d firmware from this branch. I have also used a firmware provided from pakoto (with SCC enabled and a better mixing of the sound). One time I reflashed a TN20K with your firmware and that solved the problem, but I could not reproduce it on another TN20K. That's the weird part

Red background is indeed on the HDMI output only.

[herraa1]

I recall seeing a red background too on the very early stages of the project, when there were more apparent video issues, but not since then.

Does the red background happen on different MSX machines and different HDMI monitors?

Can you please try flashing this previous .fs on your 1.02d and see if something changes?
https://github.com/buppu3/tnCart/blob/9bbefbed26447f42592960f6ec74c0ec76d4bfba/rtl/impl/pnr/tnCart_board_wt102d.fs

[Msxinfo]

Yes, I'll try it tomorrow and will report my results. Thanks!

[Msxinfo]

So, I did a test of 8 WT boards. 3 of them are WT1.02d and 5 are WT2.00b. i installed the latest Gowin programmer on another computer and downloaded the firmware you provided here for the 1.02d and the one for 2.00b that you provide on your tnCart Wonder repo. I noticed that the Tiny Yarou tech demo is now showing a proper black background on all TN20K boards. So that is a win!

The game Eggbert only did show correct background colour with two TN20K boards. The other boards give a dark red, a dark blue or a green background. Not really a problem if it is only this game.

[herraa1]

Great. In order to narrow down the problem a bit more, could you please flash again the latest .fs for WonderTANG! 1.02d just to confirm that simply changing the .fs (and not the computer or the Gowin programmer version) you can reproduce the problem again?

[Msxinfo]

Oke, tried that. Tiny Yarou demo provides black background, but Eggbert and the msx-samoerai demo give dark red or green background.

[herraa1]

So if I understand it correctly you were not able to reproduce exactly the same problem as before, but you still get some background color issues. Isn’t it?

[Msxinfo]

Yes, that is correct.