VOGONS


Cirrus Logic CL-GD5429 VLB corrupted display

Topic actions

Reply 20 of 38, by mkarcher

User metadata
Rank l33t
Rank
l33t

The 74F32 is primarily used to provide a sufficient address space for 386DX and 486 systems. The CL-GD542x family of graphics chips only supports address lines up to A23, providing just 16MB of address space. The 542x chips can not distinguish an access to video memory at A000:0 from an RAM access cycle to 16MB + A0000 without external help. The external help is another input pin that enables the address decoder only if it is low. You could just connect this pin to A24, which means the CL-GD542x would only respond to 0..16MB, and disabled at 16MB..32MB. It will still interfere at 32MB + A0000. This was deemed insufficient by Cirrus Logic when they suggested the VL card design (and they needed some OR gates for another purpose anyway), so instead they use one gate of the 74F32 to detect whether both A24 and A25 are low, increasing the supported address space to 64MB. This means the card still interferes with RAM access at 64MB + A0000, but many system designs at that time were limited to 64MB address space anyway, so this is an acceptable compromise. This way of interfacing the CL-GD542x to the local bus thus allows operation in systems with up to 64MB address space, but as the Cirrus chip is completely taken off the bus at 16MB..64MB, the linear frame buffer, if enabled, is always located in the first 16MB. As this makes little sense, Cirrus Logic doesn't recommend to use the simple 74F32 solution if LFB usage is desired, but suggests a more complex solution involving a custom programmed PAL for high address decoding instead.

Except for keeping the 542x off the bus at 16MB..64MB, the 74F32 is also used to generate the BIOS read signal. We know that BIOS reads are working if the system is posting with something on the screen, as the BIOS is required to set up the graphics card to generate any kind of output at VGA frequencies. Furthermore, it seems more like data is bad than the cirrus chip responding or not responding at the wrong time, so I think the 74F32 is unlikely to be the culprit. The 74F32 only sees the top two address bits to decide whether the 542x should be enabled, so a failure of that chip would not explain why the text can be written to the screen memory, but the character shapes or character code numbers are messed up.

Reply 21 of 38, by HwAoRrDk

User metadata
Rank Member
Rank
Member

I've been trying to probe the signals to/from the 74F245s, but its proving to be tricky. Because of course the VLB side sees all the data bus activity from the CPU, seeing whether what's going in is also correctly coming out means one has to also look at the /OE and DIR pins, and I only have a 2-channel 'scope. ๐Ÿ™ I tried using the /OE pin with the third external trigger input - assuming the buffers will only be enabled when the GD5429 either wants to talk or listen - but it's still tricky to tell what's going on.

What I can say is that when doing that, there's a bunch of activity during POST, but then the GD5429 seems to go 'silent' at the point the system doesn't progress any further in the boot process and no further activity.

Reply 22 of 38, by tony359

User metadata
Rank Oldbie
Rank
Oldbie

I believe this video card ended up on my desk. ๐Ÿ™‚

I made a repair video as usual. If you don't want to watch the video, below is what the problem was:

Spoiler

It was the RAM chip corresponding to the socket the OP mentioned earlier on. Touching those lines would increase/decrease the corruption on screen and even cold spray would confirm that. Swapped the chip and it now works! ๐Ÿ™‚

https://www.youtube.com/watch?v=X3Cw34WQEXM

My Youtube channel: https://www.youtube.com/@tony359

Reply 23 of 38, by HwAoRrDk

User metadata
Rank Member
Rank
Member
tony359 wrote on 2026-11-04, 18:04:

I believe this video card ended up on my desk. ๐Ÿ™‚

Indeed it did. Great job Tony identifying the problem part so quickly! ๐Ÿ‘ Thanks again.

Reply 24 of 38, by tony359

User metadata
Rank Oldbie
Rank
Oldbie

Ok., I need some help from the community before I shred this card in a thousand pieces ๐Ÿ˜€ (Sorry @HwAoRrDk !)

So, a little plot twist and also spoiler, after I made the repair video I went to test the card again and it was preventing the system from booting. Which was the original fault that could be seen UNDER the video corruption. But it worked fine during my final tests! Most of the time it wouldn't say "Starting MS-DOS", sometimes it would say it but stop there. Other times it would boot and then end up in lots of errors.

Very very long story short, I found that if I warmed up the 244 which is connecting the VGA BIOS to the 8-bit ISA BUS, I could make it boot and I could run a graphic benchmark indefinitely. But once I'd quit the test and go back to the OS, I'd have HDD read/write corruption almost immediately.

So I replaced the 244 and it started working fine.

However, the video output was flickering - that is, the brightness was flickering - whenever the HDD was being accessed and also whenever some CPU activity happened. That has always been there, I thought it was my capture device but no, it happens on a monitor too.

I thought it was a third fault. I left the card alone for 3 days, went to test it again as I wanted to make sure the "unable to boot" issue was resolved first. And it wasn't. The card worked for 5 minutes and then started misbehaving again.

So I removed XT-IDE and installed a 16-bit ISA IDE controller card with the same Compact Flash card.

Now the system worked fine. I ran hours of RAM and HDD tests with Check-it, zero issues whatsoever. Even with the "bad" 244 transceiver.

But, the video output still flickered - but not as bad as with the XT-IDE in the system! ๐Ÿ˜€

On the Cirrus Logic, the video output is set by the IREF pin of the main chip. It's a constant current connection. To make that happen, there is a TL431, a transistor and a resistor. I cooled down the TL431 while the card was running a memory tests and the flickering went away. ๐Ÿ˜€

The attachment SCR-20261113-uack.png is no longer available

However, if I check with my scope, I don't really see a lot of voltage instability on the 2.5V of the 431, it's barely visible with 100mV resolution. And with XT-IDE in the system, the flickering is always there, regardless of cooling down the 431 or not.

If it helps, I've noticed that the flickering happens when CheckIt says "Lower Address Lines" only. I'm just joining the dots: the XT-IDE is an 8-bit device, the IDE controller is a 16-bit. Maybe the card has an issue with the lower bits? But that's ONLY the BIOS on this card.

I have checked on an actual monitor and the flickering is there too. Also, a different VLC board in the same system doesn't show the issue.

My question: is this Cirrus Logic card known for this bad video output? Does any of what I've said make any sense to any of you? Is the XT-IDE known for conflicting somehow with a VLB card?

I don't have another VLB motherboard to test it with but with another VLB card I have zero issues.

Help ๐Ÿ™‚

My Youtube channel: https://www.youtube.com/@tony359

Reply 25 of 38, by mkarcher

User metadata
Rank l33t
Rank
l33t
tony359 wrote on 2026-11-13, 23:19:

Very very long story short, I found that if I warmed up the 244 which is connecting the VGA BIOS to the 8-bit ISA BUS, I could make it boot and I could run a graphic benchmark indefinitely. But once I'd quit the test and go back to the OS, I'd have HDD read/write corruption almost immediately.

If you enable video BIOS shadowing, the 244 that connects the VGA BIOS to the ISA bus is not required to drive the bus anymore. So if you get issues caused by that chip after the system is booted, this is because the chip is driving the bus when it shouldn't. A way to test that is using a VL IDE controller instead. With a VL IDE controller, something running haywire on the ISA bus is not supposed to disturb hard disk operation. You might also want to scope the /OE and /DIR pins of the 244 to make sure it is controlled correctly.

tony359 wrote on 2026-11-13, 23:19:

However, the video output was flickering - that is, the brightness was flickering - whenever the HDD was being accessed and also whenever some CPU activity happened. That has always been there, I thought it was my capture device but no, it happens on a monitor too.

This usually indicates insufficient filtering of the reference current. The schematic you show below doesn't look like it depends on +12V/-12V, but there are VGA cards that use a 12V line to generate the reference current. As there is only a single +12V and a single -12V pin on the AT power connector, so if that pin is making bad contact, reference current may get unreliable. I don't necessarily think that's the issue on your card, though. Nevertheless, you should keep that in mind when troubleshooting ISA VGA cards.

tony359 wrote on 2026-11-13, 23:19:

The card worked for 5 minutes and then started misbehaving again.

Now the system worked fine. I ran hours of RAM and HDD tests with Check-it, zero issues whatsoever. Even with the "bad" 244 transceiver.

This has all the symptoms of an ISA bus conflict: The graphics card 244 and the IDE device or IDE interface card competing over D0..D7. Depending on chip type and temperature, the "desired" chip wins or not. The cirrus logic card is supposed to leave D0..D7 on the ISA bus alone unless unshadowed VGA BIOS reads happen. Thus, I'd still suggest to enable VGA BIOS shadowing, and then check with a scope whether the 244 is ever enabled to drive towards the ISA bus while the system is running. This must not happen.

tony359 wrote on 2026-11-13, 23:19:

But, the video output still flickered - but not as bad as with the XT-IDE in the system! ๐Ÿ˜€

Assuming you actually do have an ISA bus conflict, the 244 will draw excessive supply current while it fights agains the IDE data transfers. It will draw that current from +5V, which is likely used on your card to generate the reference current.

tony359 wrote on 2026-11-13, 23:19:

However, if I check with my scope, I don't really see a lot of voltage instability on the 2.5V of the 431, it's barely visible with 100mV resolution. And with XT-IDE in the system, the flickering is always there, regardless of cooling down the 431 or not.

Possibly the reference source isn't the problem after all: Many video cards have ESD protection diodes on the R/G/B monitor lines, containing reverse-biased diodes to GND/+5V. If the 244 draws down the +5V line on your card while it fights the IDE data driver, this disturbance might be capacitively coupled directly into the analog R/G/B signals via the protection diodes.

tony359 wrote on 2026-11-13, 23:19:

If it helps, I've noticed that the flickering happens when CheckIt says "Lower Address Lines" only. I'm just joining the dots: the XT-IDE is an 8-bit device, the IDE controller is a 16-bit. Maybe the card has an issue with the lower bits? But that's ONLY the BIOS on this card.

Do not confuse "lower address lines" with "lower data lines". I don't expect the 8-bit nature of the XT-IDE to be the primary issue. On the other hand, as the XT-IDE is 8 bits, IDE transfers are slower, and thus the time the VGA card might conflict with IDE transfers is longer, creating more disturbance. Also, the bus driver on the XT-IDE might be stronger than on the 16-bit IDE card, causing more issues.

I'm sorry for just focussing on the bus conflict issue at the moment, but as all the symptoms line up, I'd like to have this cause excluded before trying to think of other possible causes for the same issue.

Reply 26 of 38, by tony359

User metadata
Rank Oldbie
Rank
Oldbie

No need to be sorry - in fact I appreciate the time you've taken to give me so many good ideas.

I was going to try the shadowing technique indeed. And check CE to see if the 244 is being driven at all when that is enabled. That is a great test.

The card I have seem to follow the reference schematics I have - the reference current for the DAC is generated from 5V. The "T" is linked to the 5V input.
I had to much noise on the scope yesterday when I was testing - The ground I was using was too far I guess (the bracket).

I'll try again but this time I think even just a multimeter might be enough. Across R7, so I can see if the reference current changes at all.

AFAIK the RGB lines are only connected to ground via resistors and capacitors. Attached the VLB drawings I found in the technical manual - so far, my card seems to follow it 100%.

I like the idea of "the 244 is using more current because it's fighting something on the ISA bus" - I think the shadowing idea will settle that.

I thought I had a VLB IDE controller but I don't unfortunately.

I'll come back with more diagnostic later on, hopefully! Thanks again!

My Youtube channel: https://www.youtube.com/@tony359

Reply 27 of 38, by HwAoRrDk

User metadata
Rank Member
Rank
Member

Yeah, before I even found the reference schematics, I figured that the only thing on the card that uses the 8-bit ISA bus is the BIOS ROM and its buffer. If anything's disturbing the ISA bus, it's gotta be those.

mkarcher wrote on 2026-11-14, 15:30:

Possibly the reference source isn't the problem after all: Many video cards have ESD protection diodes on the R/G/B monitor lines, containing reverse-biased diodes to GND/+5V. If the 244 draws down the +5V line on your card while it fights the IDE data driver, this disturbance might be capacitively coupled directly into the analog R/G/B signals via the protection diodes.

There don't appear to be any ESD diodes on the RGB outputs. Like Tony said, this card matches the reference design almost exactly - only the RAM configuration differs.

tony359 wrote on 2026-11-14, 16:03:

I was going to try the shadowing technique indeed. And check CE to see if the 244 is being driven at all when that is enabled. That is a great test.

I'd be surprised if your motherboard BIOS doesn't have VGA BIOS shadowing enabled by default. My 486 board with AMIBIOS does. But worth checking I guess.

tony359 wrote on 2026-11-14, 16:03:

I had to much noise on the scope yesterday when I was testing - The ground I was using was too far I guess (the bracket).

I'll try again but this time I think even just a multimeter might be enough. Across R7, so I can see if the reference current changes at all.

Well, the other side of R7 is ground, so that would be a better place to grab ground. ๐Ÿ˜€ I would suggest doing a multi-channel measurement: one probe measuring the IREF voltage at the transistor side of R7 (which should always be a stable 2.5V), and one probe measuring one of the VGA output channels (R, G, or B), perhaps at the top side of the 150ฮฉ R9/10/11. Then you can compare any deviations in IREF with any flicker in the image. I doubt a multimeter will be fast enough to notice any fluctuations.

I actually whipped up a little simulation of the IREF circuit, because I was puzzled because I was under the impression that the whole point of a constant-current circuit was to provide a fixed current even if the supply voltage varies. A bit of noise/ripple - no problem. But as you can see, if the voltage supply dip is fast and large enough, it will definitely affect the output! You can see that as VCC dips, the IREF falls, and that will affect the output signal voltage (even if they are supplied by a separate 5V), which if it affects all three RGB channels simultaneously would indeed appear as a flicker to black.

Reply 28 of 38, by tony359

User metadata
Rank Oldbie
Rank
Oldbie

I'll check later but I think all the shadowing are disabled by default - aren't those usually disabled?

The simulation indeed shows that but I see the supply goes under 2V for a moment. The 431 supply is at 3.2V and I don't see it moving. I will have to try and get a better ground - probably from the 431 itself so hopefully I can see a cleaner signal.

The RGB outputs are - obviously - very busy. I would need to use a third channel on the vertical sync to stabilise it a bit. But running memtest can affect the output every 3-4 seconds so that could be enough to have a repeatable setup.

Did I say that at some point, with the new 244, I cooled down the 431 and the flickering disappeared? ๐Ÿ˜€

Most of the resistors on the board are cracked and showing some rust. I wonder whether I should start replacing those.

My Youtube channel: https://www.youtube.com/@tony359

Reply 29 of 38, by HwAoRrDk

User metadata
Rank Member
Rank
Member
tony359 wrote on 2026-11-14, 18:59:

I'll check later but I think all the shadowing are disabled by default - aren't those usually disabled?

๐Ÿคท I mean, it's a pretty important performance optimisation, so I would have assumed it would pay to make shadowing of C0000h enabled by default. On my board that is the only one enabled by default, all the other addresses are disabled by default.

tony359 wrote on 2026-11-14, 18:59:

Did I say that at some point, with the new 244, I cooled down the 431 and the flickering disappeared? ๐Ÿ˜€

Hmm, so perhaps intermittently bad TL431? Certainly if the TL431 was failing short-circuit between cathode and anode, that would shut off IREF current completely and cause the RGB levels to go to zero, thus producing a blank display. TL431s are pennies, so might as well try replacing it. If you don't have any, they're quite commonly used in switch-mode PSUs, so you might be able to salvage one if you have any scrap PSUs around.

tony359 wrote on 2026-11-14, 18:59:

Most of the resistors on the board are cracked and showing some rust. I wonder whether I should start replacing those.

Oh? I never noticed. But then I wasn't able to look at it under a microscope level of magnification like you have. ๐Ÿ˜ I guess worth measuring them all to see if any aren't reading what they should be.

Reply 30 of 38, by tony359

User metadata
Rank Oldbie
Rank
Oldbie

You're right, Video Shadowing is enabled by default. So all my tests were with Video Shadowing enabled.

I've checked the CS pins of the 244. With video shadowing disabled, I have lots of activity whenever something goes on screen. With shadowing enabled, there is still some activity. Very little, Like every time I type something on screen there are between 1 and 4 pulses.

Is that expected?

Edit: I could clearly see the 2.5V across the resistor rippling a bit when the screen was being affected, I tried to capture that on camera and it stopped. In fact after a bit the flickering disappeared completely.

I swapped the 431 and the issue remains.

What if the issue is not the IREF but something else which is affecting the Cirrus Logic chip internally and as a consequence you have disruption on the RGB? I'm shooting in the dark here ๐Ÿ™‚

My Youtube channel: https://www.youtube.com/@tony359

Reply 31 of 38, by HwAoRrDk

User metadata
Rank Member
Rank
Member
tony359 wrote on 2026-11-14, 20:27:

I've checked the CS pins of the 244. With video shadowing disabled, I have lots of activity whenever something goes on screen. With shadowing enabled, there is still some activity. Very little, Like every time I type something on screen there are between 1 and 4 pulses.

Is that expected?

As far as the logic regarding what controls whether the ROM is selected, from looking at the schematic and reference manual, it appears the GD5429 chip is what does the address decoding to determine whether a memory read should enable the VGA BIOS ROM. It decodes A23..15 (off the VLB, I assume) to determine if a BIOS read is occurring (presumably by matching against 0CXXXXh), and sets EROM# active. According to the schematic, that is then gated by the 74F32 chip (U22 on schematic, U8 on this board) using the ISA bus' SMRDC# and BALE lines. I guess that's all working fine, otherwise there would probably be far bigger problems. ๐Ÿ˜€

Maybe some quirk/bug of the motherboard chipset? After a bit of research, I've read that some chipsets 'leak' some memory read/write signals to the bus even for shadowed areas; one of the causes apparently being if VGA BIOS caching is enabled (that is, the CPU is allowed to cache the shadowed RAM area in L1/L2 cache) sometimes a cache miss can cause a shadow-avoiding bus transaction even when it shouldn't. I dunno whether that's real or Google is making things up, though, so take with a pinch of salt.

Reply 33 of 38, by tony359

User metadata
Rank Oldbie
Rank
Oldbie

After a few hours of fiddling, I realised I had down to 4.5V at the ISA slow for the +5V line! ๐Ÿ™‚ That might not be good.

The PSU is outputting 5.0V at the ATX plug then I have an ATX to AT adaptor, 10cm of wire and the AT socket. The AT socket was not in the best shape and I realised my AT plugs were dirty after years of old, filthy motherboards.

The most important part is that I set the multimeter to "MIN" to record the minimum value and whenever I saw the most obvious patterns on screen, the multimeter would beep and go down a few millivolts - until eventually it found the minimum value.

I cleaned everything and I got to 4.67 - should work but still not great. (EDIT: No, the minimum for those chips is 4.75V so...)

I have a bunch of those PSUs - they're old but good quality. They're FSP from 2000ish. Pentium 4 friendly. I opened another one and I found a trimpot. I tried adjusting it and it adjusts +5V and +12V together - fantastic. Tomorrow I'll set the PSU with the MB and cards connected so the ISA bus reads something like 5.0V. If it turned out it was just my PSU...

Now, I've been testing my A7V333 board with a Radeon 9200 - all from +5V line. I ran benchmarks, games, Y-Cruncher. Zero issues whatsoever. But that is an ATX board so there are no adaptors.

I'd imagine the voltage issue is a combination of the AT adaptor and maybe the motherboard not doing a great job with the PCB? The adaptor has 18 AWG wires, it looks a decent one. Though any adaptors would introduce a little voltage drop. Maybe I should have my own "AT only" PSU set to a slightly higher voltage to compensate ๐Ÿ™‚

I'm not convinced that the issue will be addressed by fixing the main rails - but that needs to be done.

On the Shadowing, It'd be nice if I could test another motherboard indeed but unfortunately this is the only one I have.

BTW, here is a capture of the issue. https://1drv.ms/u/c/0f215fbb023f1ee9/IQAYwyuG โ€ฆ cv7-DA?e=87rc9N

The "jailbar" are always there I think, it's just that when it's not flickering the compression removes them? Maybe.

My Youtube channel: https://www.youtube.com/@tony359

Reply 34 of 38, by butjer1010

User metadata
Rank Oldbie
Rank
Oldbie

I think mkarcher told You guys that this could be problem (one of few) with current on at connector, in his second post. This guy is unbeliveable, and could solve any problem, just need to read what he says. Trust me... ๐Ÿ˜‰

Reply 35 of 38, by tony359

User metadata
Rank Oldbie
Rank
Oldbie

ok - I've looked into the issue again.

I feel the PSU was ok, maybe on the low side just below 5V (4.97V when wired to an ATX motherboard) but not the end of the world. I tweaked it to 5.1V, it can't hurt I guess, those chips can work safely up to 5.25V.

The drop was at the ATX -> AT adaptor -> AT socket. Those terminals have worn out over the years. I cleaned the connectors as much as I could, cleaned the motherboard AT socket too with fiber pen and re-crimped new terminals at the AT end of the adaptor.

Now I have 5.1V at the ISA slot which is more than I was expecting.

And... fingers crossed... the video out is fine now. Too early to celebrate because I've seen this before but so far so good.

Something I noticed is that the voltage was dropping a lot if I was squeezing the ATX->AT plugs. So I'd imagine all my "unreliability" of the past many days might have been caused by where that adaptor ended up being on my desk! ๐Ÿ˜€

Even the 244 was probably fine: Just working better when heated up because it was being powered with 4.5V ๐Ÿ˜€

I now want to know the voltages whenever I fix a board. Is there a Post Analyser card with voltage displays on it? ๐Ÿ˜€

PS: I did notice the voltages were on the low side but didn't pay too much attention to that. I assumed it was just bad connection with my probes and didn't think too much about that. That said, when I checked I was reading 4.8V, not 4.5. I checked in a "lucky" moment! Yes, 4.8 is low but I've always noticed 4.8V *at the AT plugs* and never realised the adaptor was eating up 200mV of energy!

EDIT: I ran 2 hours of memtest, no video issues. Then I rebooted and XTIDE refused to boot with "No System disk" ๐Ÿ˜€

I still had the "bad" 244 installed, I swapped it again and it worked - so maybe the original 244 was marginal after all. I cannot be sure as I've seen this 200 times: it works until it stops working! I'll test in 2 days and then in 2 more and then in 2 more... I think it's the only way to be sure! ๐Ÿ˜€

What a journey!

My Youtube channel: https://www.youtube.com/@tony359

Reply 36 of 38, by mkarcher

User metadata
Rank l33t
Rank
l33t
tony359 wrote on 2026-11-15, 16:29:

PS: I did notice the voltages were on the low side but didn't pay too much attention to that. I assumed it was just bad connection with my probes and didn't think too much about that. That said, when I checked I was reading 4.8V, not 4.5. I checked in a "lucky" moment! Yes, 4.8 is low but I've always noticed 4.8V *at the AT plugs* and never realised the adaptor was eating up 200mV of energy!

Keep in mind that the multimeter has a very high input impedance (likely 10 MOhm unless you use the cheapest stuff, which still has 1 MOhm), so there should be no noticable voltage drop in your probing setup. If you read low values that are stable (so the contact isn't "scratchy"), they are most likely real measurements.

Reply 37 of 38, by tony359

User metadata
Rank Oldbie
Rank
Oldbie

Thanks - No, my Fluke is not the cheapest one ๐Ÿ˜€

I meant the contacts, the whole board is pretty oxidised and I had to struggle to read stuff more than once.

In the end, it's my bad. The PSU is a "known good one" and it had just run a long session with a much more demanding system where I ran benchmarks and games, 100% from the 5V line with zero issues!
When I saw a low-ish value on the VLB board, I didn't pay too much attention!

Lesson learnt for the future - until I forget and it happens again!

Now, this is assuming it works now. I have seen this board "healing itself temporarily" in the past again so we shall see! ๐Ÿ˜€

My Youtube channel: https://www.youtube.com/@tony359

Reply 38 of 38, by tony359

User metadata
Rank Oldbie
Rank
Oldbie

Update.

I don't see the video flickering anymore and the system boots every time.

But I think more VRAM has failed.

Sometimes in DOS, scrolling the screen (hitting enter or while something happens) will mess up the whole screen. However, the system won't stop and CLS will clear the screen and the system will keep working.
Checkit splash screen has a missing line.
Checkit VRAM tests fail ๐Ÿ™‚

So this is what, a fourth fault with this card? ๐Ÿ˜€

To be honest, I've seen that before on poorly-stored card where you can see rust and corrosion a bit everywhere.

I'll work on that. So far, I only have 500GB of footage for 17 hours of recordings. it'll be fun.

My Youtube channel: https://www.youtube.com/@tony359