(CPU cooler removed on purpose on PC3)

Chuck them in the dishwasher without soap.
Dry thoroughly.

I've had very dead things come good.
Eg a 100% dead rx580.

Is the missing cache board for PC3 obligatory? According to a manual I found, there's a jumper for selecting 256 or 512 KB cache RAM - not disabling it?
I don't have any cache modules, but it looks like it could be something like this: https://en.wikipedia.org/wiki/Cache_on_a_stick

I dumped the BIOS ROMs of the three systems and here are some of the strings I found:

PC1:
- (C) COPYRIGHT IBM CORPORATION 1981, 1994 ALL RIGHTS RESERVED
- 03/17/92

PC2:
- (C) COPYRIGHT IBM CORPORATION 1981, 1995 ALL RIGHTS RESERVED
- 12/08/95

PC3:
- SIS super VGA chip.
- Silicon Integrated Systems Corp.
- SiS 6205
- PS/1 POWER MANAGEMENT
- (C) COPYRIGHT IBM CORPORATION 1981, 1995 ALL RIGHTS RESERVED
- EL TORITO SPECIFICATION
- SurePath(tm) BIOS Version 0.02
- Copyright (C) 1993 I.B.M. Corporation., All Rights Reserved.
- 02/26/97

myne wrote on 2024-10-01, 14:32:

Chuck them in the dishwasher without soap.
Dry thoroughly.

We have extremely hard water here, so I'm not sure that's a great idea - also, I'm worried about the salt?

Rinse them with metho/IPA if you're that worried.
But realistically, the contacts are like grand canyon distances compared to the modern things I've washed.

So, I'm knee deep in the "Hardware Maintenance Service for Service Level A Machine Types 2134 and 2176", and I'm noting that..
- the motherboard has an extra 6 pin PSU-style connector marked "P11" that is only described as "3.3V" which is not connected.
- the ISA/PCI riser card (not visible on the pictures above) has the same style 6 pin connector on the back marked "J11", also not connected.

The PSU that came with the dead PC doesn't have anything except the two P1/P2 6 pin connectors. Maybe I need another PSU that has the 3.3V connector?

Most likely it's 12v+5v+3.3 and it goes out to the fdd/hdd

Got multimeter?

A couple of updates on PC3:

- Riser board and motherboard had two unused 6 pin connectors.
- Fitted the two regular connectors from the original PSU as well as the Rapid Resume 3-pin connector.
- Fan turned on when pushing power button. CPU ice cold.

I was unable to acquire the power connectors, so after extensive trawling of the 'net, I found and ordered this:
https://www.bicker.de/en/x1-116_atx_at_adapte … r_at_mainboards

When fitting the new adapter on the motherboard (NOT THE RISER BOARD) with a modern ATX PSU, it seems to work - but only if you change the setting on the white jumper underneath the regular power connectors. This apparently negates the requirement for the 3-pin Rapid Resume connector.

- System boots
- Identifies as Aptiva 3134-445 with Pentium 100 processor (which is weird, since clock and multiplier is set to 133, which matches the CPU)
- In a bizarre twist of events, the BIOS is in Swedish. Wonderful.

Without much thought, and probably blinded by success, I connected the remaning 6 pin power connector to the riser card, and the system refused to start. Not even the PSU fan.

- Fitted last 6 pin connector to riser card without thought - system dead.
- Measured header against PCI ans ISA connectors
- Discovered that the riser power connector needs to be INVERTED compared to one on the motherboard. Same 3x0+3x3.3V - just mirrored.
- System boots. ISA and PCI cards work.
- BIOS flashed to newer, English version.

Remaining issues:

- Power button doesn't work - turns off power LED and freezes system, but doesn't power anything off. Need to figure out how to do this correctly.
- CPU falsely identities as 100 MHz even though clock and multiplier seems to be correct.
- Consider more RAM and possibly a COASt (Cache-On-A-Stick) module.
- Case doesn't accommodate a modern PSU. Found another Aptiva case that is closer, but needs a little modification.

The two other systems are still dead.

Possibly something wrong with BF1 pin connection at socket or between there and jumper. Jumpers also go bad/dirty. Though run a benchmark check see if it comes up at 133, might just be the BIOS doesn't know anything faster than 100.

Then there is a rare situation with very early socket 5 and the initial implementation of multipliers, which only had one frequency BF pin, which could have values other than 0/1. Meaning only the first stepping of P133 would run at 133 on a board with that implementation, later ones would expect later multiplier scheme.

This is assuming that maybe the 133 is an attempted upgrade from 100.

Arrrrrrggggh! Turns out an ATX power supply will block the power connectors, since the board is apparently designed for a 42H2380 power supply. It came with a 42H2379, which is missing the 3.3V connectors.

Those "square" end PSUs are available refurbished, but at grotesque prices and primarily in the US, so postage to the EU is going to be insane. Bit of a setback. The case looked pretty after being in the dishwasher though 😀

BitWrangler wrote on 2024-10-09, 13:22:

Possibly something wrong with BF1 pin connection at socket or between there and jumper. Jumpers also go bad/dirty. Though run a benchmark check see if it comes up at 133, might just be the BIOS doesn't know anything faster than 100.

Then there is a rare situation with very early socket 5 and the initial implementation of multipliers, which only had one frequency BF pin, which could have values other than 0/1. Meaning only the first stepping of P133 would run at 133 on a board with that implementation, later ones would expect later multiplier scheme.

This is assuming that maybe the 133 is an attempted upgrade from 100.

Out of desperation, I tried numerous different combinations of multipliers and clock settings, up and including 200 MHz. The system booted without any issues, no matter the combination, and apart from one setting resulting in 90 MHz, all the others were 100 MHz. I'm thinking either a connection issue, or the CPU is somehow locked. When I removed and cleaned it, it looked very much like an original fit, bit I can't be sure.

Do an actual benchmark.
The bios might not have the capability to recognise more.

I assume you found this
https://www.manualslib.com/download/3128495/I … ptiva-2134.html

Though on closer look at board and CPU it's a 97 board and CPU is C step, so they should know each other, and CPUs up to 200 should be in the BIOS, so that might mean it's a bad connection problem on the multi pin to jumper somewhere.

But is this happening with a battery installed? Some boards, if they work at all with no battery, get stuck in crawler gear until fresh battery installed.

myne wrote on 2024-10-09, 13:42:

Do an actual benchmark.
The bios might not have the capability to recognise more.

I ran Norton SysInfo after every jumper change 😀

Oh something to be careful of in these IBMs, they use some of the jumpers sideways instead of up and down, double check you are actually jumpering 1-3, 2-4, if that's what it says not 1-2, 3-4.

I'm off on vacation in a little bit, but I thought I'd give an update. After consulting an expert in hardware cleaning, I decided to throw all three motherboards in the dish washer, and clean off with demineralized water. The working board, "PC3", was still working and now looks brand new 😀

Unable to find a "square end" PSU with 3.3V, I went back to the case with the ATX/AT-sized PSU and decided to go for an SFX-sized PSU. They're about 2/3 the size of ATX and you can get an ATX bracket for it. Seemed like the perfect idea, and it would leave enough room underneath for the power connectors/cables.

SADLY this is where I ran into issues. The power LED flickers when I power on the system, which seems to indicate that the PSU isn't delivering enough power. Primarily on the 5V rail.

The system is now clean as a whistle, I've worked around the Rapid Resume issue, I've found 64MB RAM for it, but I still have these outstanding issues:

- Pentium 133 CPU still at 100 MHz. Found a Pentium 200 MMX I'll try when I get back.
- ATX PSU blocks motherboard power connectors. SFX PSU perfect, but lacks power. "Original" (what came with the case) doesn't have 3.3V.

Pictures of the three different PSUs I've tried attached. See image descriptions.

If you happen to know how to get a hold of the original "square end" PSU *WITH* 3.3V, please let me know.

I'm like "It's really hard to underpower a machine of that spec..." *zooms into PSU spec label* oh shit, that's low, wouldn't have expected less than 12.

A couple of small updates, before I'm off to the airport:

PC 3: Bit the bullet and ordered an original PSU off Ebay. Shipping from the US is horrendous. The correct model number appears to be IBM 06H8822/06H8824. Now we wait.

PC 1: Dried from the dish washer adventure, jumpers, BIOS and CPU and riser card refitted. Threw in the ISA/PCI diagnostics card, and was greeted with all power good, but "00 AA" in the display. The manual only seems to suggest Phoenix, AMI and Award POST codes. According to the manual, "00" is only used by AMI and means "copy copying to specific areas done. passing control to int 19". For "AA", AMI says "Initialization completed. Displaying system config" while Phoenix says "Scan for F2 key stroke". Nothing happens after this point. CPU is luke warm, so is Cirrus Logic GPU.

Ideas?