Some of the chips have arrived so I was finally able to replace all the damaged chips that are hooked to the PCI-bus: Buffer VL82C595AFC2, PCI IDE controller CMD PCI0640B, system controller VL82C594AFC2 and PCI-ISA bridge chip 82C596AFC2.

Everything is working fine again, except onboard video. The order seems to be stuck in customs.

Impressive amount of effort for a non Intel chipset (aka probably good 20% slower) socket 5 board with funny IDE controller 😮

It´s more the habit of fixing what I broke myself 😉 Plus it teaches you to be more careful...

The "Wildcat" chipset doesn´t perform _that_ bad.

It´s basically a P54C mainboard that can be modded/upgraded for P55. The Pebble Beach is collectable for me because it´s made by FIC and it is rare. And that´s why it´s important to get onboard video working again.

I finally replaced the onboard graphics chip (Cirrus Logic 5436) yesterday, but there´s still no video output.

When I got this mainboard there was no video output. While testing with an additional video card I picked the wrong riser card so all the chips got damaged. After that the video chip was overheating. The new chip´s temperature is normal, but still no output. I´m sure I can rule out bad soldering joints or shorts.

I now suspect the video BIOS contained in the Phoenix system BIOS is for the S3 chip that is used on most FIC / AT&T PSK-2000 and alike mainboards and the Cirrus Logic BIOS is not present.
The BIOS the PSC-2000 came with was AT&T / NCR 1.06 and I updated it to 1.07.

Can someone identify the video BIOS inside this BIOS and if it´s S3, could it be replaced??

majestyk wrote on 2026-09-15, 12:49:

[...]

but in a hurry I picked some different model with the same "EISA" connector.

It's called PISA and has a specified, standard pinout.

[...] […]
Show full quote

[...]

Same testing at the PCI pins showed a "floating" "AD30" line. With the original FIC riser the PCI "AD30" pin was routed to the mainboard correctly without interrupion and it was also connected to the "AD30" pins of the PCI-IDE controller, one of the memory buffer chips, the PCI-ISA bridge chip and the System controller chip. Since all chips are conneted to the "AD30" bus line a floating line means all of the chips´ I/O circuits for this line got blown up.
BUT WHY? Closer inspection revealed that the wrong riser card routes the AD30 pin of the PCI slots to the AD22 pins of the chipset - this shouldn´t cause any damage.
Vice versa it connects the incoming AD30 signal from the mainboard (not to the PCI AD22 pins) but to +12V. Obviously none of the chips survived this abuse.

Either the FIC stuff or the 'wrong' is clearly not sticking to PISA standard with magic smoke release results.

As I have some PISA stuff too, I'm pretty interested to figure out which is the odd one out here so I can avoid it like the plague. According to PISA spec, the upper row (eg. what you'd get if you stuck a regular ISA card in the slot) should have all the ISA signals and power, the lower row (the deep pins an ISA card wouldn't touch) should have all the PCI - and so should never carry anything over 5V. I can't quite make out from your post whether the 'wrong' riser connected a PCI line to the upper row (bad) or whether the FIC board is providing +12V on the lower row (bad).

The riser card at the top left of the picture (the one with the small round white sticker) belongs to a Mitac / Trigon PWA-PL5400VF mainboard from a "HP Brio" system. It routes 12V- exactly as I described.
The remaining two are FIC / AT&T risers that are (of course) working without issues.

majestyk wrote on 2026-10-01, 09:16:

The riser card at the top left of the picture (the one with the small round white sticker) belongs to a Mitac / Trigon PWA-PL5400VF mainboard from a "HP Brio" system. It routes 12V- exactly as I described.
The remaining two are FIC / AT&T risers that are (of course) working without issues.

That +12V is being routed to AD30 and blasting out the chips when combining FIC board with Mitac riser is clear, so one of the two isn't following spec - but it's not clear (to me at least) which is doing something badly wrong. Does the motherboard supply +12V on one of the lower pins or does the Mitac riser connect AD30 to one of the upper pins supplying +12V?

I just checked the connections on the riser cards:

FIC and AT&T riser cards: 12V ISA (pins B9) are connected to 12V PCI (pins A2). Both are connected to pin B9 in the upper row of the PISA slot. -> correct, these are the ISA contacts.

HP riser card: 12V ISA (pins B9) are connected to 12V PCI (pins A2). Both are connected to pin B9 in the upper row of the PISA slot PLUS to pin A14 of the lower row of the PISA slot. (This pin A14 is - in compliance with the PISA standard - connected to PCI A30 on the FIC /AT&T cards.) And that´s the problem with the HP riser.

HP obviously thought that the PISA standard is so important that everybody should make their own one.

so where does the HP riser route PCI A30 then? 😀

rasz_pl wrote on 2026-10-01, 11:52:

so where does the HP riser route PCI A30 then? 😀

To A22 of the mainboard (see earlier post).
Please don´t ask where A22 is being routed to... I´m sure there are more proprietary connections, which - as long as it´s address or I/O lines - does no damage.
But the A30->12V one is s deadly trap.

majestyk wrote on 2026-10-01, 10:15:

I just checked the connections on the riser cards: […]
Show full quote

I just checked the connections on the riser cards:

FIC and AT&T riser cards: 12V ISA (pins B9) are connected to 12V PCI (pins A2). Both are connected to pin B9 in the upper row of the PISA slot. -> correct, these are the ISA contacts.

HP riser card: 12V ISA (pins B9) are connected to 12V PCI (pins A2). Both are connected to pin B9 in the upper row of the PISA slot PLUS to pin A14 of the lower row of the PISA slot. (This pin A14 is - in compliance with the PISA standard - connected to PCI A30 on the FIC /AT&T cards.) And that´s the problem with the HP riser.

HP obviously thought that the PISA standard is so important that everybody should make their own one.

Tnx.

We knew some of the big brands had their own PSU pinouts (Compaq & Dell...), now we can add power lines on their PISA to the list 😦

Does any of you guys happen to be capable of looking inside the PSK-2000 1.7 BIOS to find out what video BIOS is inside?

Looks like Phoenix NuBIOS is compressed like Award bioses, you wont know whats inside without somehow extracting it, I have no experience with PHoenix bioses, or compressed bioses in general 🙁
Btw its another At&T Globalyst platform 😮 😀

Google points to "PHOEDECO" for uncompressing NuBIOS, but this seems to an OS2 program 🙁

https://www.sac.sk/files.php?d=13&p=17

The descript.ion file mentions:
phoedeco Phoenix BIOS unpacker (Linux)
phoedeco.com Phoenix BIOS unpacker (DOS)
phoedeco.exe Phoenix BIOS unpacker (OS/2)
phoedecw.exe Phoenix BIOS unpacker (Windows)

Thanks!
After decompression we have 6 single files:

The .ORO file is the S3 video option rom that needs to be replaced by a Cirrus Logic GD-5436 one.

do you have a packer tho? 😀

I have no idea so far what to use for repacking.
Phoenix Bios Editor does all the jobs like decompressing, editing, checksum readjusting and recompressing - for Phoenix BIOS versions it supports.
BPE 2.0.18 won´t open the AT&T BIOS ("Wrong BCPSYS version!").

Btw. the onboard video chip gets detected by speedsys along with the working S3 one, so it´s at least (probably) operational if the BIOS was present.

Im guessing one of those AIs could probably reverse engineer phoedeco.com and generate packer