VOGONS


Early P5 + P54C (Socket 4 + Socket 5) mainboard

Topic actions

First post, by LocalBus

User metadata
Rank Member
Rank
Member

Hi! After going "down" the ranks with DOS compatible machines (started with AMD64 with ESS-Solo1 soundcard) - I am finally back to the roots where my journey began back in 1994. While my machine was a classic VLB 486 DX2 spec, I got this mainboard handed down to me a couple of years later when it was time for upgrade. Long story short, I went for a Cyrix 6x86 PR166 CPU instead, so this mainboard have been collecting dust ever since.

I have been trying to wrap my head around the manufacturer behind it, but so far to no avail.

What is very rare about it, is the Socket 4 + Socket 5 combo, where the printing on the mainboard covers jumper settings for P60/P66 (P5) and P90/P100 (P54C). I know the first owner bought it with the original P90 (Gold heatspreader) with the FDIV bug in all its glory. Sometime later this was swapped for the P100 that is currently mounted.

I'm now on a quest to resurrect this piece of history (have to source some 72-pin FPM SIMMs and some kind of Multi I/O card first). Since it does lack all the legacy I/O connectors on the mainboard I have to go the ISA route for the I/O card and possibly a complementary PCI IDE controller. I guess technically there were never any PCI Multi I/O cards? I just remember the huge VLB Multi I/O card on my old 486.

It carries the early Pentium chipset from OPTi:

82C571 + 82C572
And
82C822 (this one surprised me) - VLB-to-PCI bridge - makes me wonder if it would be possible to retrofit a VLB port 🤔

I'll try to keep this topic open as I go, hopefully it will POST once more!

Cheers,
Linus

20211127_181251.jpg
Filename
20211127_181251.jpg
File size
1.91 MiB
Views
1643 views
File license
Public domain
20211127_181251.jpg
Filename
20211127_181251.jpg
File size
1.91 MiB
Views
1643 views
File license
Public domain
20211127_181339.jpg
Filename
20211127_181339.jpg
File size
1.6 MiB
Views
1643 views
File license
Public domain
20211127_181352.jpg
Filename
20211127_181352.jpg
File size
1.58 MiB
Views
1643 views
File license
Public domain
20211127_181401.jpg
Filename
20211127_181401.jpg
File size
1.79 MiB
Views
1643 views
File license
Public domain

Attachments

Reply 1 of 283, by Tetrium

User metadata
Rank l33t++
Rank
l33t++

PCI560-01 and made in Taiwan, didn't find anything in google (got wrong results).
The board is damaged at the top left at the edge, but could still work.

Very rare sight indeed, I've only ever come across a single computer with Socket 4 and Socket 5 and that one wasn't a standard AT board (it needed a riser) so I didn't take that board.

You'd better remove that varta battery btw before it starts leaking and ruins the board.

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 2 of 283, by dionb

User metadata
Rank l33t
Rank
l33t
LocalBus wrote on 2021-11-27, 17:42:

[...]

And
82C822 (this one surprised me) - VLB-to-PCI bridge - makes me wonder if it would be possible to retrofit a VLB port 🤔

Quite likely (electrically; obviously no place physically on the motherboard).

The 82C571 is a hybrid 486/Pentium chipset - yes, there are boards with So3 and So4 with it as well. And I strongly suspect it doesn't have a PCI controller integrated, so it talks to the outside world over what is essentially a VLB bus. That 82C822 is needed to connect those PCI slots to that. Probably didn't help performance...

Reply 3 of 283, by LocalBus

User metadata
Rank Member
Rank
Member

Thanks for the comments and input! Yes a very rare board indeed, I searched through here https://www.ultimateretro.net/en/motherboards/search/ (both specs and pictures but nothing with both Socket 4 and Socket 5).

Yes there is a slight cosmetic damage in that corner but no damaged traces what I can see.

Luckily no signs of leaks from the Varta 3 battery but I have ordered NiMh replacement for the same (so I can still charge it with mainboard current).

I hope it will not be too picky with memory.

Keep you POSTed!

Cheers,
Linus

Reply 5 of 283, by LocalBus

User metadata
Rank Member
Rank
Member
Doornkaat wrote on 2021-11-27, 19:24:

That's very cool!
Could the missing oscillator be the culprit?

Well spotted!

Still waiting for parts so I cannot test it yet.

Looks like there is another Xtal next to it. I doubt that this have been removed, but who knows 🤔

What kind of oscillator would normally sit there? Related to Real Time Clock? Or maybe this one is key for 66 MHz FSB. Not sure how fancy clock generator circuit this board is carrying but it does support changing between 1x and 1.5x multiplier and 60/66 MHz FSB.

Good guess, 33 MHz oscillator maybe?

I cannot find any Dallas module or similar on the board...

Last edited by LocalBus on 2021-11-27, 19:50. Edited 1 time in total.

Reply 6 of 283, by LocalBus

User metadata
Rank Member
Rank
Member
Doornkaat wrote on 2021-11-27, 19:24:
That's very cool! Could the missing oscillator be the culprit? […]
Show full quote

That's very cool!
Could the missing oscillator be the culprit?

20211127_203420.jpg
Filename
20211127_203420.jpg
File size
1.83 MiB
Views
1576 views
File license
Public domain
20211127_203451.jpg
Filename
20211127_203451.jpg
File size
1.88 MiB
Views
1576 views
File license
Public domain

Reply 7 of 283, by LocalBus

User metadata
Rank Member
Rank
Member
Tetrium wrote on 2021-11-27, 18:14:
PCI560-01 and made in Taiwan, didn't find anything in google (got wrong results). The board is damaged at the top left at the ed […]
Show full quote

PCI560-01 and made in Taiwan, didn't find anything in google (got wrong results).
The board is damaged at the top left at the edge, but could still work.

Very rare sight indeed, I've only ever come across a single computer with Socket 4 and Socket 5 and that one wasn't a standard AT board (it needed a riser) so I didn't take that board.

You'd better remove that varta battery btw before it starts leaking and ruins the board.

Some cosmetic damage:

20211127_194042.jpg
Filename
20211127_194042.jpg
File size
1.27 MiB
Views
1571 views
File license
Public domain

Battery will be replaced with NiMh counterpart:

20211127_194108.jpg
Filename
20211127_194108.jpg
File size
1.21 MiB
Views
1571 views
File license
Public domain

Reply 8 of 283, by TheMobRules

User metadata
Rank Oldbie
Rank
Oldbie

Very interesting board, I love this 486-to-Pentium transitional phase!

You may want to check the area below as it seems to be damaged and a trace could be affected:

PCI560-01.jpg
Filename
PCI560-01.jpg
File size
51.92 KiB
Views
1559 views
File license
Public domain

Regarding the oscillator, I assume it would be a 60 or 66MHz part depending on the bus speed you want. Do you have any documentation about the board?

Reply 9 of 283, by LocalBus

User metadata
Rank Member
Rank
Member
TheMobRules wrote on 2021-11-27, 19:49:
Very interesting board, I love this 486-to-Pentium transitional phase! […]
Show full quote

Very interesting board, I love this 486-to-Pentium transitional phase!

You may want to check the area below as it seems to be damaged and a trace could be affected:

PCI560-01.jpg

Regarding the oscillator, I assume it would be a 60 or 66MHz part depending on the bus speed you want. Do you have any documentation about the board?

Thanks!

Yes 60/66 MHz oscillator makes sense considering what CPU this mainboard is intended for.

Unfortunately the manual is long gone (I never had it). I will try my luck with similar OPTi boards from that era.

This predates PLL chips 😀

Clock Chip = Chrontel CH9007E

Need to find this datasheet...

20211127_210959.jpg
Filename
20211127_210959.jpg
File size
1.88 MiB
Views
1540 views
File license
Public domain

Thanks for spotting the potential trace damage, looks OK but I check for continuity and protect with some new solder mask.

20211127_205432.jpg
Filename
20211127_205432.jpg
File size
1.58 MiB
Views
1546 views
File license
Public domain

Reply 10 of 283, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie
LocalBus wrote on 2021-11-27, 19:58:

Need to find this datasheet...

I can't find a datasheet, but did find a reference, and an output table for the CH9007b on page 23 here: https://core.ac.uk/download/pdf/4405047.pdf

Maybe JP28 is frequency selector?

Oh, and the RTC is probably in the 206: http://bitsavers.trailing-edge.com/components … eets/82C206.pdf section 4.5

Reply 11 of 283, by LocalBus

User metadata
Rank Member
Rank
Member
snufkin wrote on 2021-11-27, 20:57:
I can't find a datasheet, but did find a reference, and an output table for the CH9007b on page 23 here: https://core.ac.uk/down […]
Show full quote
LocalBus wrote on 2021-11-27, 19:58:

Need to find this datasheet...

I can't find a datasheet, but did find a reference, and an output table for the CH9007b on page 23 here: https://core.ac.uk/download/pdf/4405047.pdf

Maybe JP28 is frequency selector?

Oh, and the RTC is probably in the 206: http://bitsavers.trailing-edge.com/components … eets/82C206.pdf section 4.5

Brilliant! Yes the CH9007b seems to work similar as more "modern" Frequency/Clock generator chips, like this one:

https://datasheetspdf.com/pdf-file/556937/Int … s/ICS9169C-27/1

What they have in common is that they use the 14.318 Mhz Xtal as reference, and then produce various frequency/clocks based on this.

This leads me to believe that the socketed oscillator is optional, but I guess it could be used for more fine granular control maybe? Nevertheless, I have just ordered a 66 MHz oscillator in DIP-14 package (harder to find than I would expect).

Regarding the jumper settings, you are completely right. JP28 is for setting 60 or 66 Mhz FSB:

20211127_221715.jpg
Filename
20211127_221715.jpg
File size
1.63 MiB
Views
1503 views
File license
Public domain
20211127_221742.jpg
Filename
20211127_221742.jpg
File size
1.64 MiB
Views
1503 views
File license
Public domain

Additionaly, JP33 is for setting the 1.x or 1.5x multiplier (not sure why it is located so close to the 3.3V voltage regulator, but maybe it enables/disables the P60/P66 CPU for the Socket 4 as well - using 1x multiplier.

20211127_221850.jpg
Filename
20211127_221850.jpg
File size
1.6 MiB
Views
1503 views
File license
Public domain

Reply 12 of 283, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
LocalBus wrote on 2021-11-27, 21:33:

Additionaly, JP33 is for setting the 1.x or 1.5x multiplier (not sure why it is located so close to the 3.3V voltage regulator, but maybe it enables/disables the P60/P66 CPU for the Socket 4 as well - using 1x multiplier.

20211127_221850.jpg

This seems somewhat plausible, since P60/P66 doesn't have a 1.5x multiplier and Socket 5 CPUs don't have 1.0x multiplier.
I wonder how the board would work having 2 different CPU interfaces. There are also s370/Slot1 combo boards and obviously these were not made to work simultaneously.

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 13 of 283, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
Doornkaat wrote on 2021-11-27, 19:24:

That's very cool!
Could the missing oscillator be the culprit?

If this would be the case, it would make me wonder why someone would even remove only the oscillator if the board was trashed? (I would have at least removed all the jumpers as well if it was a dead board)
If I had to guess, I would have guessed this missing part was probably optional. Still doesn't mean this is for sure not the problem of course 😜

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 14 of 283, by Doornkaat

User metadata
Rank l33t
Rank
l33t
Tetrium wrote on 2021-11-28, 09:52:
Doornkaat wrote on 2021-11-27, 19:24:

That's very cool!
Could the missing oscillator be the culprit?

If this would be the case, it would make me wonder why someone would even remove only the oscillator if the board was trashed? (I would have at least removed all the jumpers as well if it was a dead board)
If I had to guess, I would have guessed this missing part was probably optional. Still doesn't mean this is for sure not the problem of course 😜

It's a guess based only on seeing the empty socket. I guess I'd just try and follow the traces to figure out what it might do. Maybe it's for <60MHz FSB CPUs?
Why would somebody pull this out? Why not? 😉 People disassemble things for no reason. Maybe because it's shiny or looks interesting? Personally I do lots of stuff for no apparent reason. 😅

Reply 15 of 283, by dionb

User metadata
Rank l33t
Rank
l33t
Tetrium wrote on 2021-11-28, 09:50:
LocalBus wrote on 2021-11-27, 21:33:

Additionaly, JP33 is for setting the 1.x or 1.5x multiplier (not sure why it is located so close to the 3.3V voltage regulator, but maybe it enables/disables the P60/P66 CPU for the Socket 4 as well - using 1x multiplier.

20211127_221850.jpg

This seems somewhat plausible, since P60/P66 doesn't have a 1.5x multiplier and Socket 5 CPUs don't have 1.0x multiplier.
I wonder how the board would work having 2 different CPU interfaces. There are also s370/Slot1 combo boards and obviously these were not made to work simultaneously.

Slot 1 and So370 are at least electrically identical, so easier than this. Agreed that this could be a voltage jumper mislabled as 'multiplier'. Could also just be the BF0 1.5 vs 2.0 mutliplier for So5 in an odd place.

Reply 16 of 283, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
Doornkaat wrote on 2021-11-28, 10:26:
Tetrium wrote on 2021-11-28, 09:52:
Doornkaat wrote on 2021-11-27, 19:24:

That's very cool!
Could the missing oscillator be the culprit?

If this would be the case, it would make me wonder why someone would even remove only the oscillator if the board was trashed? (I would have at least removed all the jumpers as well if it was a dead board)
If I had to guess, I would have guessed this missing part was probably optional. Still doesn't mean this is for sure not the problem of course 😜

It's a guess based only on seeing the empty socket. I guess I'd just try and follow the traces to figure out what it might do. Maybe it's for <60MHz FSB CPUs?
Why would somebody pull this out? Why not? 😉 People disassemble things for no reason. Maybe because it's shiny or looks interesting? Personally I do lots of stuff for no apparent reason. 😅

Yes, that's kinda true, that's definitely a possibility. Could be that someone needed that particular part for whatever reason and thought something like "I'll put it back later" and forgot about it.
Though there was quite a lot of missing stuff on mainboards those days. Even completely missing slots with only the solder dots visible or missing sound chips etc.

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 17 of 283, by LocalBus

User metadata
Rank Member
Rank
Member

I am fairly confident that any potential external oscillator is redundant in this case.

I found the datasheet for an earlier version of the Chrontel clock generator chip CH9001 (486), which indeed only relies on the 14.318 Mhz Xtal for generating the wanted frequency/clock outputs:

http://13.209.45.252/pdf/download.php?id=9789 … e=O&term=CH9001

In fact, the datasheet specifies that the Xtal input shall be shorted to ground if an external oscillator is being used as reference instead, and in such case the oscillator should match the Xtal frequency.

Time will tell, waiting for a shipment of SIMMs to test with along with a Cirrus Logic GP-5446 PCI graphics adapter. I was also told why this board was scrapped, and that was due to it being incompatible with the Matrox Mystique PCI 2.1 card 😄

Reply 18 of 283, by LocalBus

User metadata
Rank Member
Rank
Member

This datasheet is more or less a perfect match with the CH9007E sitting on the board (did a sanity check of the pinout configuration in continuity mode):

https://www.datasheetarchive.com/pdf/download … e=O&term=CH9008

Possibly on the 'E' revision I have the possibility to push it up to 80 Mhz FSB if playing with the FS0,1,2,3 jumper configuration.

The empty socket for the oscillator have no connection with the clock generator more than sharing Vdd and GND. Definitely not used as external Fref input and in that case it must be a value between 1 to 20 Mhz; possibly it could produce other frequencies out of that.

Next step is to look for BF0 and BF1 traces and to which jumpers they go to on the board.