Looks cool. I've never had any experience with these early IBM PCs. I don't think I've even used one - I saw some common desktop models occasionally as a kid but that's it.
Several years ago I was in a local computer shop and they had an old IBM that they were giving away for free. I left it there, but kind of wish I had taken it.

The capacitors might all have a common ground. If so, you could have probably probed to find which pads were in common to each other and the PSU ground connection. However - that method could backfire with any caps on negative voltage rails, which I suppose is pretty likely for a system of that era.

Ambitious project raymangold 😀 I have wanted to recap some of my boards but have never gotten around to it. I know next to nothing about the early IBM PS/2 models. 1987 and 1024x768 @ 256? damn....must have cost a small fortune 😀

When its complete what do you plan to use it for?

Love the case, cant really tell but is it full AT tower size?

shamino wrote:

Looks cool. I've never had any experience with these early IBM PCs. I don't think I've even used one - I saw some common desktop models occasionally as a kid but that's it.
Several years ago I was in a local computer shop and they had an old IBM that they were giving away for free. I left it there, but kind of wish I had taken it.

The capacitors might all have a common ground. If so, you could have probably probed to find which pads were in common to each other and the PSU ground connection. However - that method could backfire with any caps on negative voltage rails, which I suppose is pretty likely for a system of that era.

Well one nice thing is that the cards are on +5 (since there's no ISA, -5v was never incorporated in MCA PS/2 PSUs). Near the end I was getting impatient and just wanted to finish the project, and I had a pretty good feeling I was right. I think the + / - / + variants are very uncommon.

Do you recall what model it might have been?

rgart wrote:

Ambitious project raymangold 😀 I have wanted to recap some of my boards but have never gotten around to it. I know next to nothing about the early IBM PS/2 models. 1987 and 1024x768 @ 256? damn....must have cost a small fortune 😀

When its complete what do you plan to use it for?

Love the case, cant really tell but is it full AT tower size?

I was just thinking the same thing with that card... probably was more than a grand...

Still haven't really figured it out, I was thinking maybe as an OS/2 test platform, or banking. I could also throw Windows 95 on it like the other PS/2s, but maybe that's cliche.

I'm not sure if it could be classified as AT-- it has 8 card slots and only two 5 1/4" bays at the front that are sideways (and this also sacrifices one of the HDD bays).

Looks like a major project! Very impressive! 😀

Nice job on the recapping! Every board I try to do that to breaks. Will you benchmark it?

meisterister wrote:

Nice job on the recapping! Every board I try to do that to breaks. Will you benchmark it?

Thanks for bringing that up. I did have a special process I used:
#1 You need an iron at least at 400C to melt the old solder. 300C won't be able to get all of the solder out.
#2 you MUST see the joint in a melted state and not stiff-- I introduce new solder to help the joint to become molten (again the old solder has trouble heating up from age, and lack of flux-- the flux in rosin core solder will assist)
#3 use a good solder sucker, I love the Hakko FR-300. It is the absolute best tool for through hole, there is NOTHING better.
#4 after the solder is removed, rock the tantalum side to side to free up the legs. Often times I have to rest the tip on the lead at the other side to heat it up while rocking it-- this ensures when it 'unsticks' from the via, it doesn't tear it.

Yeah I plan on doing Phil's benches on these computers.

PeterLI wrote:

Looks like a major project! Very impressive! 😀

Yeah.. doing an entire cleanup and recapping takes a fair bit of time-- used up most of my weekend, but it was fun and gave me something to do.

Very impressive. 😀

I have one of those PS/2 1.44meg drives in my IBM 5170. The drive was originally a 720k model that was available in late XTs and ATs, but I found that it was mechanically identical to the ones used in the PS/2s so i was able to mount it in my bracket. Unfortunately the drive doesn't work. I tried recapping the PCB (a common problem) and using a cleaning kit but I just couldn't get the damn thing going. If you find any information on how to repair these drives let me know. Of course for me using a standard drive is quite easy, but I like the look of the IBM parts.

I also have an IBM 486SLC2 upgrade in my 5170. However, it does not work with my adaptec AHA 1542C SCSI adapter. You mentioned there was a switch on your blue lightning module. By chance do you know what it does? If it just changes a CPU register setting it may be possible for me to get my upgrade working.

Anonymous Coward wrote:

I have one of those PS/2 1.44meg drives in my IBM 5170. The drive was originally a 720k model that was available in late XTs and ATs, but I found that it was mechanically identical to the ones used in the PS/2s so i was able to mount it in my bracket. Unfortunately the drive doesn't work. I tried recapping the PCB (a common problem) and using a cleaning kit but I just couldn't get the damn thing going. If you find any information on how to repair these drives let me know. Of course for me using a standard drive is quite easy, but I like the look of the IBM parts.

I also have an IBM 486SLC2 upgrade in my 5170. However, it does not work with my adaptec AHA 1542C SCSI adapter. You mentioned there was a switch on your blue lightning module. By chance do you know what it does? If it just changes a CPU register setting it may be possible for me to get my upgrade working.

On the back of the blue lightning card (the side with the pins), there'll be two DIP switch modules switch bank #1 (goes 1-4) and switch bank #2 (1-6). We're focusing on switch bank #2. I quote from IBM's documentation:

Switch 6 ON | Flush cache on INT 15d OFF (Default) | Do not flush cache on INT 15d If you have a SCSI hard disk drive, this swit […]
Show full quote

Switch 6
ON | Flush cache on INT 15d
OFF (Default) | Do not flush cache on INT 15d
If you have a SCSI hard disk drive, this switch must be set to ON.

Try setting switch 6 to ON. Please let me know if it works-- I'm curious myself as I might be adding SCSI to my system as well.

As per your FDD, I'm the same way. Sure I could hack another drive in there, but I like the machines to be aesthetically 'original' or complete.
Can you provide me some more details as to its behaviour? Does it power up, does it attempt to read the drive, etc. It could be an alignment issue. Also let me know what brand it is (mitsubishi, ALPs, sony, etc).

My CPU upgrade module is made by Evergreen and I don't think it has any switches, but I do remember the software being highly configurable. I'll look into this Int 15d thing, there's probably a way to turn it on through software.

I do believe my drive is made by alps. It powers on and it tries to read the disk but returns something like "sector not found reading drive a:". It could very well be an alignment problem. I think I have the drive here with me in China, but Sadly my AT has been in storage in my parent's basement for over 4 years. I'm going to try to have it shipped to China this summer. But...I think I have a spare AT motherboard here as well as the upgrade module and SCSI card, so in theory I could more or less recreate an AT for testing purposes.

VERY NICE WORK!
Always like seeing old PC's getting the full service like this. Your electronic skills are well past my own but look forward to the day I can spoil my machines in the same way. Plus I haven't had much to do with IBM's so was an interesting read as well

I second that; Very interesting project, as is your other pizzabox IBM PS/2 486. That IBM stuff makes the ordinary hardware look cheap.

raymangold wrote:

I'm deathly paranoid of tantalums exploding

Hmm, I have to look up which hardware here uses these tantalums. Never knew what to think of these little bulbs. Also recently acquired a desoldering tool, but unlike yours it is a manual pump. It should work well too.

I opened up my Acer and earlier and the s3 trio has little yellow tantalums all over it 0.o should I be worried?

raymangold wrote:

I'm deathly paranoid of tantalums exploding

gerwin wrote:

Hmm, I have to look up which hardware here uses these tantalums. Never knew what to think of these little bulbs. Also recently acquired a desoldering tool, but unlike yours it is a manual pump. It should work well too.

I don't understand why these capacitors need to replaced. Sure, tantalums tend to fail in a dead short, but modern tantalums are very reliable so failures are quite rare. And how often do they explode, assuming they haven't been installed backwards or aren't being abused in some other way? Even in the extremely unlikely event an explosion occurs it's not like they're going to penetrate the casing and be a physical hazard. As far as I know they'll remain good right up until they fail completely, unlike aluminium electrolytic capacitors, so why not just leave them in?

ODwilly wrote:

I opened up my Acer and earlier and the s3 trio has little yellow tantalums all over it 0.o should I be worried?

If they're surface mount ones and aren't 20+ years old, I'm pretty sure they'll be fine for now. Replacing them with better capacitors may improve the video output, though.

lissajous wrote:

I don't understand why these capacitors need to replaced. Sure, tantalums tend to fail in a dead short, but modern tantalums are very reliable so failures are quite rare. And how often do they explode, assuming they haven't been installed backwards or aren't being abused in some other way? Even in the extremely unlikely event an explosion occurs it's not like they're going to penetrate the casing and be a physical hazard. As far as I know they'll remain good right up until they fail completely, unlike aluminium electrolytic capacitors, so why not just leave them in?

Tantalums actually do use an electrolytic solution as one of the plates between the oxide layer-- so they do have a limited period of life; but generally are longer-lived due to the fact they are not [solely] based upon an electrolytic solution. Industrial tantalums will not contain an electrolytic solution at all, but those are not the type you deal with on consumer electronics-- and are definitely not the ones we're dealing with on Personal System/2s.
The reason why they explode is when the oxide layer reduces to metal shorting both [polarized] sides! So that is *exactly* like hooking up a capacitor backwards, but internally.

Yes fresh new tantalums are generally okay-- but we're replacing vintage ones which already having an oxide layer that's breaking down. To put it in perspective, these tantalums are older than me.
--> if they explode they can send shrapnel that can short out something in the system. If they short out in general, they can damage ICs on the motherboard. It's not good at all...
The whole point of this is to eliminate any unpredictable failures or damages to these electronics because it is not easy to source replacement parts.

High quality electrolytics won't have a problematic oxide layer. It is well documented of having exploded tantalums in vintage computers-- I've seen it occur as well. Heck, even uxwbill already had a three-leaded tantalum explode with a huge flash & bang in one of his PS/2s: https://www.youtube.com/watch?v=WVODl9NpROE

EDIT: Also inadvertently found a video with an Amiga video toaster that had a blown tantalum that completely scorched the PCB: https://www.youtube.com/watch?v=Wagp2pPDi4Q

Disclaimer: I am not an electrical engineer, I'm simply basing my observations on hands-on experience.

-----------------

A quick update, I did recap the PS/2 Model 80 power supply (I had the rivets drilled out). To my dismay there were SAMWHA capacitors, and some DAEWOOs. DAEWOO is okay, but definitely not a brand I'd put my money on. Unfortunately some of the solder pads were screwed up due to the fragility of the PCB, and I had to jump a few things to fix it as a result. It seems to have worked...
I didn't bother taking any photos after the recap, advanced apologies. I was just really frustrated after working on that thing. Regardless every single capacitor was replaced with a high quality rubycon or chemicon capacitor.

--> if you have a Plessey model 80 PSU, they are very difficult to open. Not recommended unless you're good at fiddling with things. It's unlikely anyone else on the entire universe will be recapping one any time soon, but it's worth adding that notice regardless. You have to bend some metal tabs in and pull the PCB out at an angle from one side-- placing it back in the reverse manner.

raymangold wrote:

Tantalums actually do use an electrolytic solution as one of the plates between the oxide layer-- so they do have a limited period of life; but generally are longer-lived due to the fact they are not [solely] based upon an electrolytic solution. Industrial tantalums will not contain an electrolytic solution at all, but those are not the type you deal with on consumer electronics-- and are definitely not the ones we're dealing with on Personal System/2s.

You are talking about the old style wet slug tantalum capacitors which stopped being used in the 1970s AFAIK. Since then dry slug tantalums have entirely replaced them in every market. These don't contain a liquid electrolyte, have no known wear out mechanism and are almost certainly what your computer contains. Look here for some relevant information: http://sequence15.blogspot.com/2010/12/fear-a … f-tantalum.html

lissajous wrote:

You are talking about the old style wet slug tantalum capacitors which stopped being used in the 1970s AFAIK. Since then dry slug tantalums have entirely replaced them in every market. These don't contain a liquid electrolyte, have no known wear out mechanism and are almost certainly what your computer contains. Look here for some relevant information: http://sequence15.blogspot.com/2010/12/fear-a … f-tantalum.html

A few things:
#1 vintage syntheizers use tantalums as many film capacitor technologies that exist today were not available then (which is also why they use ceramics in some places, that's also not great). Also they opted to use cheaper components in some cases for cost rather than what's 'proper'.
#2 tantalums can explode with age without having the need to be reversed-- as I already sufficiently described, the oxide layer can break down in more ways than just reversal or ripple; and once that happens, bad things happen. It's more than well enough documented that they become volatile with age, so there is zero reason to keep them in.
#3 The article states:

No electrolyte is an advantage, but many synths are loaded with electrolytic caps and they seldom leak, and even if they do, it's usually not that big a deal.

That's a rather bold statement considering tantalums do not have the same properties of some electrolytics. There are a lot of specialized electrolytic capacitors like Rubycon RX30 (130C tolerance!) that have unique characteristics which filter out line noise which I often use in PSUs and Mixers-- and stated many times, Nichicon MUSE bipolars. Tantalums are terrible at filtering as we'll get to in a bit.

I'm certain the tantalums in the PS/2s are of the electrolytic variant-- but regardless they still suffer from the oxide problems.

Since that article is targeting synthesizers and not motherboards (which is odd as most professional synth techs don't use tantalums for other reasons not pertaining to vintage computers)-- I can deviate and explain why they are terrible for synthesizers: tantalums are not good for decoupling or for voltage sensitivity. You'd want to use film capacitors.

To quickly explain why tantalums suck as decouplers, I'll quote from here: (http://www.bext.com/replace.htm):

I'll be dogmatic about this: tantalum capacitors stink for audio use: they should never be used in any signal-carrying circuit. They sound tubby in the bass, harsh and gritty in the top end, and compressed in their dynamics—mirroring their poor engineering performance in such factors as dielectric absorption. If you find them in your equipment, replace them forthwith. You'll hear the improvement. (One exception: a 1F tantalum capacitor is often used at the output of a voltage regulator chip. Leave it there.)

And it appears the fellow you quoted from praising tantalums doesn't understand what the 1uF tantalum regulators are for, as he states:

(Nonetheless, some power supply makers use tantalums as filter caps in their supplies, and it works. How they get this to work, I don't know.)

As I already mentioned earlier with the Nichicon MUSEs, you need bipolar for a lot of the outputs in which case finding a 10uF+ bipolar tantalum is going to be problematic. Also in the case of my synths with VCOs, I cough up the dough and stick in polypropylenes-- tantalums have so many issues from the dielectric leakage. If you look at the 'tolerance' rating of many tantalums (and a handful of electrolytics), they'll be at the 25% mark, which is very inaccurate.

And when the author states that

Lately I see posts in various places from people who are anxious to go through all of their synths and rip out any tantalum caps that they find. Completely unnecessary, provided that the person who designed the circuit that the tantalum cap is in was designed by a competent engineer, and that it was installed properly.

may be so but that doesn't mean the best passives were used in proper situations, or that cheaper components were used for lower costs-- let's take the Moog Polymoog and Roland Jupiter-4 into consideration.
A) The Polymoog uses *ceramic discs* for the tuning of each polycom card, that is a terrible idea and causes mechanical pickup from the keys.
B) The Jupiter-4 originally used electrolytics for the VCO tuning which causes instability.
--> In either case the only good capacitor would be a polypropylene and nothing else. Tantalums would not be appropriate either. So if we were to take the author's opinion that whatever type of passive was put in place is canon... we wouldn't be able to improve the circuit for spot-on tuning stability and eliminate issues with ceramic discs placed where they shouldn't be.

Also IBM has reversed tantalum capcaitors on some video cards-- showing that Engineers *do* make mistakes. Humorously enough the reversed tantalums on those video cards didn't seem to explode, not enough current and voltage I wager.

-----

There's a lot of varying opinions about capacitors-- but I am basing everything what I do on observation. Why didn't I replace the old tantalums with new tantalums?
#1 I don't want a capacitor with an oxide layer, horrible idea
#2 they are considerably more expensive and offer more problems than advantages & I already had a bulk stash of nice eletrolytics
#3 If I went solid state I would go 'all the way' with a more reliable OSCON polymer or film capacitor

raymangold wrote:

A few things:
#1 vintage syntheizers use tantalums as many film capacitor technologies that exist today were not available then (which is also why they use ceramics in some places, that's also not great). Also they opted to use cheaper components in some cases for cost rather than what's 'proper'.

Film capacitors have been around for as long as tantalum capacitors. Suitability is generally the reason tantalum capacitors are used in preference to them, not cost. Also, although the largest available values of film capacitors are comparable in capacitance to typical values of tantalum capacitors they are physically big and extremely expensive. For these and other reasons they are often impractical for use in applications where tantalum capacitors are ideal.

#2 tantalums can explode with age without having the need to be reversed-- as I already sufficiently described, the oxide layer can break down in more ways than just reversal or ripple; and once that happens, bad things happen. It's more than well enough documented that they become volatile with age, so there is zero reason to keep them in.

Some tantalum capacitors might become volatile with age but this certainly isn't guaranteed. I will take the fact that the ones in your computer still looked to be in good condition despite being several decades old as a good sign.

I'm certain the tantalums in the PS/2s are of the electrolytic variant-- but regardless they still suffer from the oxide problems.

They definitely don't look like wet slug tantalums. All of the ones I've seen used sealed cans, like aluminium electrolytic capacitors.

Since that article is targeting synthesizers and not motherboards (which is odd as most professional synth techs don't use tantalums for other reasons not pertaining to vintage computers)-- I can deviate and explain why they are terrible for synthesizers: tantalums are not good for decoupling or for voltage sensitivity. You'd want to use film capacitors.

Ceramic capacitors are a better choice than film capacitors for decoupling at high frequencies, i.e. the kind of frequencies you see in computers. These sort of capacitors would be my first choice for this application since they are now available in high values which are physically small and inexpensive.

To quickly explain why tantalums suck as decouplers, I'll quote from here: (http://www.bext.com/replace.htm):

I'll be dogmatic about this: tantalum capacitors stink for audio use: they should never be used in any signal-carrying circuit. They sound tubby in the bass, harsh and gritty in the top end, and compressed in their dynamics—mirroring their poor engineering performance in such factors as dielectric absorption. If you find them in your equipment, replace them forthwith. You'll hear the improvement. (One exception: a 1F tantalum capacitor is often used at the output of a voltage regulator chip. Leave it there.)

You need to differentiate between power supply decoupling and signal decoupling. I wouldn't use a tantalum for signal decoupling in an analogue circuit either, but they work very well at power supply decoupling.

Also IBM has reversed tantalum capcaitors on some video cards-- showing that Engineers *do* make mistakes. Humorously enough the reversed tantalums on those video cards didn't seem to explode, not enough current and voltage I wager.

It's much more likely that this is a manufacturing error than a design error.

There's a lot of varying opinions about capacitors-- but I am basing everything what I do on observation. Why didn't I replace the old tantalums with new tantalums?
#1 I don't want a capacitor with an oxide layer, horrible idea

But aluminium electrolytic capacitors also have a oxide layer which can fail.

#3 If I went solid state I would go 'all the way' with a more reliable OSCON polymer or film capacitor

You would use a film capacitor for power supply decoupling in a digital circuit? That doesn't make sense.