CGA and also CGA Redux circuit is sensitive to the series of IC used. The readme from github says:

U4,U5 and U101 should all be 'S174s
U9,U10 are 74153 chips.
U28 actually an 74S10

It also says that the author tried to use the 74LS series instead of them and everything worked for him, but this does not mean that it should work for you. In addition, if you use and order chips from China with Aliexpress, then now there are many fakes of the 74LS series from the 74HC series (remark).
So, chips marked as 74S on circuit must be 74S series, not 74LS or 74HC or whatever.

I have seen similar behavior on my board:

To fix it, I pulled down the /LCLK signal to ground with a 1K resistor:

You can try this, but my board is built on Soviet chips and therefore this fix may not work for your case.

Good day ! My board is also almost entirely built on Soviet microcircuits (90% -1533, 5% - 555, 5% - 74LS, RAM - РУ5Д). So I will definitely try your method. As soon as I have free time. Thank you in advance !

Good evening !
I also suspect that the problems with my CGA Redux card lie somewhere in the /RAS, /CAS, or /WE signal generation. The fact that you had an almost similar situation only strengthened my suspicions. Because I was still not sure about the control signals of the address registers /EN_CPU_CAS_ADDR, /EN_CPU_RAS_ADDR, /EN_CRT_CAS_ADDR, /EN_CRT_RAS_ADDR. But, probably, it's not about them.
So, I couldn't stand it and today I tried to pull-down the /LCLK signal, as you advised. And ….. it got even worse…. (see photo).

About the demandingness of CGA Redux to the types of chips used. I couldn't even think that in some schematics there could be a situation that the microcircuits used would turn out to be too fast! Because, usually, the opposite happens - you have to look for the fastest possible ones. And that is why I keep a stock of microcircuits K1533 (74ALS) and K1531 (74F). I try to keep even K555 (74LS) to a minimum. Let alone K155 (74) and K531 (74S).
@Tronix : I looked carefully at the photo of your CGA Redux. You used almost the entire set of ex-USSR TTL logic - K555 (74LS), K1533 (74ALS), K531 (74S). And even somewhere they found such a rarity as K1554TB9 (74AC112). Although, it seems, is it CMOS? And although most of the microcircuits you use are K555 (74LS), you still put K531 (74S) in the most "responsible" places. Even on panels - surely you have experimented with them?
I have almost no K155 (74) or K531(74S) series chips. So I started "randomly" changing the control logic to K1531 (74F) chips. Just for an experiment. With each replaced chip, the image became more "dirty" until CGA Redux stopped rendering images at all. So, it may really be a matter of the type of microcircuits used, especially those responsible for generating dynamic memory control signals.
So I quickly ordered a large handful of different K531 (74S) series chips online. Moreover, they can still be found at a ridiculous price (about 10 cents per piece). Will receive an order - will continue my efforts to complete CGA Redux. But I still hope that CGA Redux can work on 74ALS microcircuits. So I would appreciate any advice.

Vlodek_d wrote on 2023-08-28, 17:49:

@Tronix : I looked carefully at the photo of your CGA Redux. You used almost the entire set of ex-USSR TTL logic - K555 (74LS), K1533 (74ALS), K531 (74S). And even somewhere they found such a rarity as K1554TB9 (74AC112). Although, it seems, is it CMOS? And although most of the microcircuits you use are K555 (74LS), you still put K531 (74S) in the most "responsible" places. Even on panels - surely you have experimented with them?

I assembled the board from the parts that I had at hand, so I had to put one K1554TB9 CMOS chip. And yes, I also tried different types of IC in some of seemed to me important sections of the circuit until I achieved a result. I tried to follow a simple rule - install those series of IC that are indicated in the original scheme on the github, that is, all the main chips of the 74LS (K555) series and in some special places the 74S (K531) series. The 74S series has a lower propagation delay compared to the 74lLS series, i.e. they are faster.

Unfortunately, I don't know how else I can help. In any case, I wish you a successful launch of your board.

Tronix wrote on 2023-08-28, 19:10:

And yes, I also tried different types of IC in some of seemed to me important sections of the circuit until I achieved a result. I tried to follow a simple rule - install those series of IC that are indicated in the original scheme on the github, that is, all the main chips of the 74LS (K555) series and in some special places the 74S (K531) series. The 74S series has a lower propagation delay compared to the 74lLS series, i.e. they are faster.

It's strange.... If so, then why did everything get worse for me as I installed 74F microcircuits? 74F is better than 74S in everything, only they consume a little less..

Thanks to Tronix for the tip. The solution to the problem turned out to be exactly as he described. Replacing several microcircuits with 74S (K531) helped, the board works flawlessly. All the same, I don't understand why it didn't work when I replaced the same microcircuits with 74F (K1531)...