IBM PS/2 model 9595 Pentium 90 processor complex problem

[microtask]

I have 2 type 4 Pentium P90 processor complexes, however neither of them work. I suspect that the problem is down to the capacitors, but I don't know where to start. I don't know what the capacitor values should be, I don't have a circuit diagram, which would allow me to test them in-circuit. The cards are both FRU 07H7095, one came from a Server 500 and the other from a 9595. Can anyone help me with a circuit diagram, or preferably somewhere I can send it to have the capacitors replaced? I'm in the UK but there's no problem in me sending these anywhere abroad. BTW I only want 1 of these, so I'm willing to give the other to anyone who's interested.

 

[NeXT]

First question is: Are you trying to use the boards in a system that you know is fully working with another processor complex, or is the entire system stone dead with no power?

 

[GiGaBiTe]

For the capacitors, why do you not know the values? Are they not printed on the capacitors? Almost all capacitors have markings on them telling you what they are, the only real exception are MLCC SMD capacitors.

MLCC capacitors rarely go bad unless they're mechanically stressed and crack.

 

[microtask]

First question is: Are you trying to use the boards in a system that you know is fully working with another processor complex, or is the entire system stone dead with no power?

I do have a working 9595 that works with a 486DX2 processor complex, and one of the P90 complex's did work in there some time ago. The other one was pulled out of a Server 500 before I got rid of it.

 

[microtask]

For the capacitors, why do you not know the values? Are they not printed on the capacitors? Almost all capacitors have markings on them telling you what they are, the only real exception are MLCC SMD capacitors.

MLCC capacitors rarely go bad unless they're mechanically stressed and crack.

The markings seem to be so faint, and some so tiny, that I can't see what's on them. I think that the age of these capacitors might be the problem, these boards were sold in the mid-1990s, and I've seen that even solid state capacitors can fail over time.

 

[ardent-blue]

First question is: Are you trying to use the boards in a system that you know is fully working with another processor complex, or is the entire system stone dead with no power?

"however neither of them work"

Describe happens when you push in the power button.

Type 4 "Y" Processor Complex and 8595 / 9595 XP Planar and 9595A / Server 500 Planar should be a start.

2[x] type 4 Pentium P90 processor complexes
suspect that the problem is down to the capacitors
FRU 07H7095, one came from a Server 500 and the other from a 9595.

WHAT system is your 9595? Is it a Single Serial / Single Parallel or Dual Serial / Dual Parallel planar? Type-Model, please [like "9595-OMT"] This alone does not explain everything, but it helps.
The 95A DS/DP planar has "LogicLock" and a jumper to allow functioning.

FRU 07H7095 is the complex with a "fixed" P90.
The BIOS level may be of any level and the system will boot and POST.

 

[microtask]

The 9595 is a dual serial/dual parallel planar. I can’t tell you which model it is because I replaced the case with one from a 9585. It isn’t a 9595A due to it not having a RAID controller. With these complexes installed and when I press the power button it starts the PSU, but there’s nothing displaying on the front panel. As mentioned earlier, the system works ok with a DX2 66 complex.

 

[GiGaBiTe]

The markings seem to be so faint, and some so tiny, that I can't see what's on them. I think that the age of these capacitors might be the problem, these boards were sold in the mid-1990s, and I've seen that even solid state capacitors can fail over time.

Got any pictures of the capacitors?

Solid capacitors can indeed fail from old age, but the enemy in 90s electronics are SMD capacitors. The rubber plugs fail and let all of the electrolyte leak out the bottom.

 

[microtask]

I've added some pictures of the capacitors on one of the boards, they're both identical in appearance.

 

[GiGaBiTe]

Those all look to be 22uF 20v SMD Tantalum capacitors. Sort of an odd voltage, I doubt they're being pushed anywhere near their voltage limit. All of the chips around them are 74 series TTL logic, so they're probably running in the 5v range.

Tantalum capacitors don't go bad like an electrolytic cap does, where an electrolytic cap can have a variety of different failures, tantalums almost always short out. They can go out in spectacular fashion by burning or exploding.

You can check the tantalum capacitors with a multimeter to see if any of them are shorted, but I doubt you'll find anything. If you've powered the board up, they're most likely fine, you would have seen one go off by now.

Instead of looking at the capacitors, I'd be looking at the 74 series logic chips. It's more likely that one of them has an issue rather than the capacitors.

 

[microtask]

Those all look to be 22uF 20v SMD Tantalum capacitors. Sort of an odd voltage, I doubt they're being pushed anywhere near their voltage limit. All of the chips around them are 74 series TTL logic, so they're probably running in the 5v range.

Tantalum capacitors don't go bad like an electrolytic cap does, where an electrolytic cap can have a variety of different failures, tantalums almost always short out. They can go out in spectacular fashion by burning or exploding.

You can check the tantalum capacitors with a multimeter to see if any of them are shorted, but I doubt you'll find anything. If you've powered the board up, they're most likely fine, you would have seen one go off by now.

Instead of looking at the capacitors, I'd be looking at the 74 series logic chips. It's more likely that one of them has an issue rather than the capacitors.

Thanks for your post, it's always useful to know what to look at with these boards. It's over 40 years since I did anything with TTL, and then it was through hole rather than SMT. Do you know of anyone who might be able to test/diagnose/repair this board without needing a 9595 to do any testing? I really want these repaired so I'm prepared to send them anywhere at any cost (within reason of course)

 

[GiGaBiTe]

Not having the machine really limits the testing that someone can do on the CPU card. You really need the CPU card installed and powered on to do any diagnostics on it.

 

[microtask]

Not having the machine really limits the testing that someone can do on the CPU card. You really need the CPU card installed and powered on to do any diagnostics on it.

I think the only option would be to provide the 9595 to whoever does the diagnostic/repairs. Shipping that will be expensive

 

[GiGaBiTe]

You may check around locally and see if there are any vintage computing groups, someone may be able to help diagnose it. You could also use the machine as a learning opportunity, buy some diagnostic equipment and attempt to fix it yourself. You'd need a logic analyzer and an oscilloscope that have sufficient speed and bandwidth to measure signals on that machine. I'd recommend something that can measure at least 200 MHz to be on the safe side.

I definitely wouldn't recommend shipping something that big, especially if it's already broken, it could just make it worse.

 

[luckybob]

Off the top of my head... Are you using the correct RAM? I know the 486 complexes are much more forgiving, where the Pentium is limited to just a handful of compatible simms.

 

[vossi]

I have/had 8x P90 complex and all are working with the original tantal caps.
So I dont think you have to replace the caps.
Tantals are normally ok till they explode but they don't dry out like electrolytic caps.
Mostly tantals explode that are only a small bit above the voltage limit used (5150, 5160 boards).
But IBM used 20V tantals - so they go on the safe side!

Christian

 

[microtask]

I have/had 8x P90 complex and all are working with the original tantal caps.
So I dont think you have to replace the caps.
Tantals are normally ok till they explode but they don't dry out like electrolytic caps.
Mostly tantals explode that are only a small bit above the voltage limit used (5150, 5160 boards).
But IBM used 20V tantals - so they go on the safe side!

Christian

Do you have any would like to sell?

 

[microtask]

Off the top of my head... Are you using the correct RAM? I know the 486 complexes are much more forgiving, where the Pentium is limited to just a handful of compatible simms.

One of the Pentium processor complexes worked in the same machine, so I would think the RAM is not really the problem. The fact that nothing is displayed on the front panel suggests to me that it's properly damaged.

 

[vossi]

Do you have any would like to sell?

sorry, I have only left two in my 9595's and a spare for each.

 

[ardent-blue]

The 9595 is a dual serial/dual parallel planar. I can’t tell you which model it is because I replaced the case with one from a 9585. It isn’t a 9595A due to it not having a RAID controller. With these complexes installed and when I press the power button it starts the PSU, but there’s nothing displaying on the front panel. As mentioned earlier, the system works ok with a DX2 66 complex.

It's hard to differentiate the planars in a quick style, a short-hand is 95 and 95A. The 95 is used in 8595 and non-Pentium 9595 [xLx and xMx]. The 95A DS/DP planar supports ONLY Type 4 complexi.

You are correct that a "true" 95A[rray] model would come with a SCSI RAID controller.

I have a Franken95A made from a 9585, the 85 sysboard had three electrolytics pop and the on-board SCSI F/W had spooge over chip leads Dog robbed a wall fan from my stash.

By "DX2-66" I assume you mean a Type 4 "N" complex.

9595A / Server 500 Planar should show some details.

>IF< you did not touch any memory -OR- jumpers, then the Pentium complexi -SHOULD- work. All Type 4 complexi use interleaved FPM -or- ECC of the same specs, so an N will accept / tolerate the same memory specs used by a Pentium complex.

Processor Complex Installation has some nuggets.

85/95 & 3511 Power Supply has more nuggets

PSU spools up. Nothing on OP panel.

Hmmm. Perhaps you must burn down the 'ville in order to save it. Adapters "cocked" in an MCA slot can prevent booting. SIMMs installed backwards [yes... a friend... did this...]. A complex might not be fully seated. PSU correctly seated?

Power down, unplug cord from PSU.
Pull all adapters.
Blow out MCA slots with computer duster, inspect for spring damage or crud.
Pull all memory except the lone pair in B1/A1. Reseat both SIMMs.
Unship complex. Inspect component leads for damage.
Inspect MCA edge contacts on complex for trace damage or corrosion. If needed, buff edge contacts on complex with a double-folded dollar bill.

Properly and carefully re-install complex.

Plug in power cord. Power up. Is the system alive? Beeps? Anything on OP panel?