rasz_pl wrote on 2026-10-26, 13:38:

EDIT: but there is caveat. I missed the PAL! 🙁 You mentioned PAL at the start but linked to a picture of card without one so Im confused.
All ICS2494-253 using cards do have the PAL, no idea whats the purpose, maybe to map frequencies by manipulating VCLK on the fly so ICS2494-253 output matches UM9502?

I made a mistake, someone told me it was a PAL and I was screwed, but it turned out to be a PLL.
Disregard PAL and just think PLL.
Sorry 😀

After the long read, my issue is that I have no 14.318MHz signal.
PLL is dead or crystal is dead.

I tried another crystal and no sugnal anyway.
By putting the PLL in reverse by mistake (at the very beginning, after cleaning it) I had +5V inputting to STROBE... fried anything?

Edit: by bodging a wire from another working 14.318 could work? Maybe it's just that easy?

Nexxen wrote on 2026-10-25, 23:39:

Take your time, but really thanks!

Here is the BIOS dump from my card:

I made a mistake in my previous post. I meant the two jumpers (J1 and J2) near the UM9502 chip, not UM85C408AF chip. Did you tried to change the frequency with them?

analog_programmer wrote on 2026-10-26, 16:08:

Here is the BIOS dump from my card: […]
Show full quote

Nexxen wrote on 2026-10-25, 23:39:

Take your time, but really thanks!

Here is the BIOS dump from my card:

The attachment UM85C480AF_VGA_BIOS.zip is no longer available

I made a mistake in my previous post. I meant the two jumpers (J1 and J2) near the UM9502 chip, not UM85C408AF chip. Did you tried to change the frequency with them?

Thanks!

I tried everything.
Unfortunately the PLL isn't issuing any frequency for memory and core... I have to figure that out, maybe it's just that and it would be a miracle 😀

Nexxen wrote on 2026-10-26, 14:06:

Disregard PAL and just think PLL.

no, I mean the UMC cards uisng ICS2494-253 that I found all have additional PAL chip
https://www.ebay.com/itm/324695209123
Looking more carefully on that ebay card PAL is somehow wired between vga chip and ram 😮 what the hell? is that one of those salvaged half broken ram cards? It is using some weird cl514512 with logo I never saw before. Your card has just a bunch of jumpers in same spot wired to same pins. Ok, for me that settles PAL chip has nothing to do with clocks. Those jumpers and usaeg of PAL makes me think they were doing funny smelly stuff using partial ram, like using couple bits from one ram chip and couple others from the next one !?!? :0
UMC was THE first Taiwanese IC Foundry and design house. They were manufacturing ram themselves so sellind a cheap solution with cheap controller, pcb and super cheap partial rams would be right up their alley. I remember mkarcher mentioning particular cards using partial ram, but I dont remember if it was UMC.

After looking once again I finally noticed MemClock is handled by jumpers, so no dynamic MClk switching, and can be set freely to even above 50MHz, one less thing to worry about.

Nexxen wrote on 2026-10-26, 14:06:

After the long read, my issue is that I have no 14.318MHz signal.

I would start by verifying that you can see crystal with your equipment at all, try measuring 14MHz on a crystal of another card, as I said measuring crystals is a dicey subject. Its a very delicate and miniscule signal that gets crushed by any stray capacitance/impedance like that introduced by scope probe

Nexxen wrote on 2026-10-26, 14:06:

By putting the PLL in reverse by mistake (at the very beginning, after cleaning it) I had +5V inputting to STROBE... fried anything?

VCC went to ground and ground to VCC - that usually means smoke inside the chip (fried silicon or bond wires)

Nexxen wrote on 2026-10-26, 14:06:

Edit: by bodging a wire from another working 14.318 could work? Maybe it's just that easy?

no. But your Can generators idea is good.
Any 30-60MHz to memclock pin and something around 25-30 MHz to VCLK and the card should start again.

rasz_pl wrote on 2026-10-26, 17:12:

Nexxen wrote on 2026-10-26, 14:06:

Edit: by bodging a wire from another working 14.318 could work? Maybe it's just that easy?

no. But your can generators idea is good.
Any 30-60MHz to memclock pin and something around 25-30 MHz to VCLK and the card should start again.

I tried yesterday to connect +5V to those square crystals I have.
Couldn't read anything.

My oscilloscope can read fine on any working board, 14.381, 66/60/1 MHz; this makes a bit confused on why I can't measure anything on my vid card.
I'll try my crystals on old 286/386 boards. If they are fakes and I got scammed, that's will settle it.

Yes, the vid card in undewhelming but I'd like to make it work again.

I have a card bearing this UMC chip *somewhere* although I am unsure if it also uses the UMC PLL chip but I can tell it is an extremely slow card, much slower than a Trident 9000 chipset. If I can find it again, perhaps it can find its way to a new home... (I moved in summer, a lot of stuff is still misplaced 🤣)

Tiido wrote on 2026-10-26, 17:25:

I have a card bearing this UMC chip *somewhere* although I am unsure if it also uses the UMC PLL chip but I can tell it is an extremely slow card, much slower than a Trident 9000 chipset. If I can find it again, perhaps it can find its way to a new home... (I moved in summer, a lot of stuff is still misplaced 🤣)

It's an option.
Moving is never funny unless you are a minimalist and all your belongings are in a duffle bag.

Tiido wrote on 2026-10-26, 17:25:

I can tell it is an extremely slow card, much slower than a Trident 9000 chipset.

It is very slow, because its BIOS lacks any VESA modes support. The videocard needs VBE TSR to be acceptable, but this eats some DOS RAM.

Im 90% convinced now those are ram partials. UM85C408 chip has two 8 bit busses, and those jumpers select which bits wire to which one of 3 ram chips 😮 so frickin cool! 😀 Like ZX Spectrum using factory defective ram, or SpecTek itself
Found one of the ram partials topics with a lot of info I posted about companies that did it Strange SIMM ram package
and mkarcher one with simm made from partials For your amusement: SIMMs with "interesting" design you can see similarity to this graphic card with jumpers for mixing bits.

rasz_pl wrote on 2026-10-26, 17:44:

Im 90% convinced now those are ram partials. UM85C408 chip has two 8 bit busses, and those jumpers select which bits wire to which one of 3 ram chips 😮 so frickin cool! 😀 Like ZX Spectrum using factory defective ram, or SpecTek itself
Found one of the ram partials topics with a lot of info I posted about companies that did it Strange SIMM ram package
and mkarcher one with simm made from partials For your amusement: SIMMs with "interesting" design you can see similarity to this graphic card with jumpers for mixing bits.

Also this: Re: [SOLVED] First attempt at "let's make some 30-pin FP memories"... turned out to "let's not make stupid things again"

analog_programmer wrote on 2026-10-26, 18:00:

Also this: Re: [SOLVED] First attempt at "let's make some 30-pin FP memories"... turned out to "let's not make stupid things again"

This was the topic I vaguely remembered and mentioned earlier, you are a legend for finding it! Its even the very same card 😁

rasz_pl wrote on 2026-10-26, 19:22:

This was the topic I vaguely remembered and mentioned earlier, you are a legend for finding it! Its even the very same card 😁

😀 Actually it was a thread started by me about some self-made SIMM RAM sticks, but also turned into a side-discussion about the chips on my ISA UMC videocard. The comment I linked above explains everything about its strange VRAM configuration.

Compared your BIOS with mine and there are 6008 differences.
Probably the origin of those artifacts at the beginning?

Dosdays Bios has only 18... could be nothing until I test them both.

Nexxen wrote on 2026-10-26, 20:14:

Compared your BIOS with mine and there are 6008 differences.
Probably the origin of those artifacts at the beginning?

Dosdays Bios has only 18... could be nothing until I test them both.

This is interesting. Could you share yours too?

analog_programmer wrote on 2026-10-26, 20:19:

Nexxen wrote on 2026-10-26, 20:14:

Compared your BIOS with mine and there are 6008 differences.
Probably the origin of those artifacts at the beginning?

Dosdays Bios has only 18... could be nothing until I test them both.

This is interesting. Could you share yours too?

There you go.
Btw, pin 1 reads as bad and must be forced. Probably a OTP rom.

To my knowledge Strobe is tied to Ground.
24 to VCLK and 32.768 to MCLK
After post screen hits all the errors + F1 or entering bios = wrong frequency and no display
I tried other combos but they all fail or just display a very wrong screen.

https://youtu.be/l9ku0hmzdvQ

we are back at broken ram? full success ! 😀 I would rip out those partials and the weird jumpers, trace how data (clearly distinguishable 2x8 tracks) route to ram and fit two good known chips
weird about strobe, maybe the NCR version of ics chip doesnt care

measure 4 VCLK pins when card generates mess, and what it changes to when picture fails, I bet it changes resolution and wants different clock
edit: or just measure Vsync and Hsync on vga connector, I bet the chip just generates video at wrong frequency and learning it will also tell us which one it wants 😀

rasz_pl wrote on Yesterday, 01:27:

we are back at broken ram? full success ! :-) I would rip out those partials and the weird jumpers, trace how data (clearly dist […]
Show full quote

we are back at broken ram? full success ! 😀 I would rip out those partials and the weird jumpers, trace how data (clearly distinguishable 2x8 tracks) route to ram and fit two good known chips
weird about strobe, maybe the NCR version of ics chip doesnt care

measure 4 VCLK pins when card generates mess, and what it changes to when picture fails, I bet it changes resolution and wants different clock
edit: or just measure Vsync and Hsync on vga connector, I bet the chip just generates video at wrong frequency and learning it will also tell us which one it wants 😀

I'll measure everything tomorrow.

One ram chip is noticeably hotter than the others. I'm inclined on replacing with some RAM chips I have but also socketing isn't out of the picture.
I need to wire things better as I was using long cables as a test. It was a mess and I didn't take a picture but it was kindergarten style 😀
Ripple needed a disclaimer, just looking at the oscilloscope screen made me dizzy - 🤣

This is the BIOS I found on VGA museum IIRC.

Nexxen wrote on Yesterday, 11:08:

This is the BIOS I found on VGA museum IIRC.

Thanks! This one is identical to my dump (minus 8 KB of empty space at the end). The BIOS from your card contains second OpROM starting at address 0x6800 to the end of the file (I have no idea for what is this one), otherwise it's identical to mine and to the one downloaded from the VGA Museum site. The checksums are correct for both OpROMs in the dump from your card.

So, it seems like the problem is not in the videocard's BIOS chip.