VOGONS


First post, by geiger9

User metadata
Rank Newbie
Rank
Newbie

I was very fortunate to find a 386 locally. I can't stress enough how rare it is to find this locally. I've been searching for a long time and I don't know if I would ever come across this again. The PSU died the day after I bought it. I turned off the machine, I heard a pop and crack, then all kinds of smoke came out. Here are albums full of pictures of the PSU both the top PCB and the bottom.

https://imgur.com/a/KECP06m

https://imgur.com/a/jYw1Qpe

I spoke to some people in a Discord chat about this issue and they identified three blue RIFA caps on the bottom PCB that had blown and suggested they be replaced. I have no soldering skills so I had to take this to someone to get the work done. They replaced the 3 blue caps but when I power this on now, the fan spins up for not even a second and then stops. I tried connecting a HDD and when I flick that big red power switch, the drive and the fan spin up and then promptly spin back down. The guy at the electronics store seemed very hesitant to fix this. When I first brought it in he told me that they would not do any diagnosis. I would have to tell them exactly what I wanted replaced and then they would only do that. He didn't even test it after he did his work to make sure the PSU worked. I won't be bringing it back there so now my only options are two acquaintances who are familiar enough with soldering who say they will try to help me. They probably would diagnose but they are not used to working with PSUs. They can solder though.

The way the case of my computer is designed (it looks like a clone of the IBM AT 5170), the PSU's red switch sticks out the side. I REALLY don't want to replace it for monetary and aesthetic reasons. To buy another one like this on eBay would cost a couple hundred.

Can anyone out there please help with this? I will send you more pictures or a video, or whatever you need to help me diagnose this.

Reply 1 of 15, by FAMICOMASTER

User metadata
Rank Member
Rank
Member

It is really best not to use these supplies and to instead swap to a more modern ATX type supply and use an ATX-AT converter for the power switch and cabling. This is just for safety.

On the other hand, I'm a serious advocate for keeping old PSUs like this around, so I will give you a few basic tips on repairing it:
Make sure you didn't kill the board first, please. Test it with another supply if you can. If you killed the board this project is basically worthless.

Disassemble and clean EVERY piece of that power supply. leave no surface uncleaned.
Do not just replace the parts which have failed - All the parts in that supply are theoretically the same age or older and all have been under similar stress their entire lives.
If you are serious about keeping it around, replace ALL capacitors and replace probably MOST resistors.

Attach a voltmeter and leave the supply running with a high load on it (Old MFM hard disks make great loads. If you have them, I want them) for a few hours to make sure it's stable. If nothing goes bang, you're golden and it will probably last another few decades. If it does, throw it in the trash because a coil inside is likely damaged and you will not want to spend any more time or money on such a supply anyways.

Reply 2 of 15, by Caluser2000

User metadata
Rank l33t
Rank
l33t

I'd just transfer the innards of a later clone AT psu in. In fact I'm going to do that on an old style psu with a huge red switch as a spare for my XT-Turbo system.

There's a glitch in the matrix.
A founding member of the 286 appreciation society.
Apparently 32-bit is dead and nobody likes P4s.
Of course, as always, I'm open to correction...😉

Reply 3 of 15, by FAMICOMASTER

User metadata
Rank Member
Rank
Member
Caluser2000 wrote on 2021-06-09, 04:26:

I'd just transfer the innards of a later clone AT psu in. In fact I'm going to do that on an old style psu with a huge red switch as a spare for my XT-Turbo system.

This is also a good idea. I'm just a weirdo.

Reply 4 of 15, by TheMobRules

User metadata
Rank Oldbie
Rank
Oldbie

Looks like that unit was well built for its time, it's full of Japanese capacitors.

In any case I agree with the post above, you first need to disassemble it and give it a good clean before you do anything else. Problem is, you may need to desolder some wires in order to do that so if you don't know how to do it it's gonna be tricky.

After cleaning, look for obvious signs of damage: burn marks in the PCB, blown components, etc. If everything seems OK we can move to a more in-depth troubleshooting.

geiger9 wrote on 2021-06-09, 01:34:

The guy at the electronics store seemed very hesitant to fix this. When I first brought it in he told me that they would not do any diagnosis. I would have to tell them exactly what I wanted replaced and then they would only do that. He didn't even test it after he did his work to make sure the PSU worked.

It's becoming more common in recent times to hear about electronic repair shops that are so afraid of working with power supplies. That doesn't give a lot of confidence in their work...

Reply 5 of 15, by canthearu

User metadata
Rank Oldbie
Rank
Oldbie

Yeah, it is a shame that technicians do not want to touch PSUs at all.

Fixed a small bluray home theater system for a friend after finding out all that was wrong with it was a fused bridge rectifier.

There is always a risk of shock, but you can minimise it by keeping everything powered off and testing the large switching PSU capacitors for voltage before working on it. And only test it while it is in its case and on a earth leakage protected circuit.

Edit: But yeah, this isn't permission to go digging in your own PSUs if you have no electronics experience at all. It isn't a good place to start.

Reply 6 of 15, by FAMICOMASTER

User metadata
Rank Member
Rank
Member

Working on power supplies is typically frowned upon because it can be dangerous in some cases, and has become an insurance liability. Most avoid it where possible.

For troubleshooting, have you remembered the first law of diagnostics?
"Thou shalt check voltages"

I.E. have you turned the supply on and checked the +12/-12/+5 outputs? Given your description of what happened and the fact the board has tantalum capacitors... It may not have been the supply which quit.

Reply 7 of 15, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie

First, that PSU is filthy inside. I would not even work on it in that state, not only all that dust and grime and cigarette smoke residue are harmful to your lungs, they can also cause PCB corrosion or even shorts if that stuff gets dump enough on a rainy day. Step 1 is always disassemble and clean, just wash it soap or dish washing liquid with a soft brush (long enough to reach everywhere). Leave it to dry out in room temperature for few days at least.

Second, the start and immediate stop is almost always an indication of short on the output. Probably not the 12V line, that would make the fan not spin up. Could be internally shorted capacictor or, sometimes, dried out caps can also cause that by not smoothing the output properly and then the spikes will trip the controller.

In general though if one is not comfortable with working on line PSU I'd suggest to look for a different one - soldering skills aside, there is enough energy stored in the input capacitors to kill, not to mention causing serious damage to any measurement equipment attached to the primary side without proper care (and knowledge on how to do that properly in the first place).
Modern PSUs do have one major issue though, except those really high-end 750W+ units these are geared towards 3V3 and 12V line loads. A typical modern "old PC" will not even have a mechanical HDD anymore, or just one, so the 12V is almost not used, and 3V3 isn't at all obviously. The simplified way cheper ATX PSUs regulate output means the 5V line will tank, 12V line will be a bit high, and in general you can't really draw too much from 5V rail anyway. That being said most of my PCs don't even get close to 40W at wall socket, only the Pentium 200MMX has higher power draw of just shy of 50W. You could maybe just stick a modern, passively-cooled +5/+12 PSU there and add a small DC-DC converter to make -12V out of 12V. If you need -5 it can be easily obtained by slapping a 7905 regulator on the -12 line, that's what most old AT PSUs did.

Reply 8 of 15, by canthearu

User metadata
Rank Oldbie
Rank
Oldbie
Deunan wrote on 2021-06-09, 09:03:

Modern PSUs do have one major issue though, except those really high-end 750W+ units these are geared towards 3V3 and 12V line loads. A typical modern "old PC" will not even have a mechanical HDD anymore, or just one, so the 12V is almost not used, and 3V3 isn't at all obviously. The simplified way cheper ATX PSUs regulate output means the 5V line will tank, 12V line will be a bit high, and in general you can't really draw too much from 5V rail anyway. That being said most of my PCs don't even get close to 40W at wall socket, only the Pentium 200MMX has higher power draw of just shy of 50W. You could maybe just stick a modern, passively-cooled +5/+12 PSU there and add a small DC-DC converter to make -12V out of 12V. If you need -5 it can be easily obtained by slapping a 7905 regulator on the -12 line, that's what most old AT PSUs did.

Yeah, In theory, a modern ATX power supply is too 12V focused for old 5V systems.

But 386's use so little power overall that it doesn't really matter. You can get ISA cards that create a -5V rail for the odd ISA cards that need them.

Reply 9 of 15, by geiger9

User metadata
Rank Newbie
Rank
Newbie

I was hoping it was a very obvious thing that you could point to in the pictures but it appears not. I don't have the skills to troubleshoot a PSU so I think I will attempt is to use another AT PSU on this motherboard. The case won't look right without the giant red switch but perhaps it is within my skillset to take the big red switch off the old PSU and connect it to the other PSU.

Thank you for all the information.

Edit: Okay the form factor of the PSU in this IBM 5170 clone is VERY different from this other AT PSU I tried putting in its place. The power cord and fan don't sit at the right places and are obscured by the case.

Last edited by geiger9 on 2021-06-11, 19:40. Edited 1 time in total.

Reply 10 of 15, by mkarcher

User metadata
Rank l33t
Rank
l33t
geiger9 wrote on 2021-06-09, 01:34:
I turned off the machine, I heard a pop and crack, then all kinds of smoke came out. Here are albums full of pictures of the PSU […]
Show full quote

I turned off the machine, I heard a pop and crack, then all kinds of smoke came out. Here are albums full of pictures of the PSU both the top PCB and the bottom.

https://imgur.com/a/KECP06m

https://imgur.com/a/jYw1Qpe

I spoke to some people in a Discord chat about this issue and they identified three blue RIFA caps on the bottom PCB that had blown and suggested they be replaced. I have no soldering skills so I had to take this to someone to get the work done. They replaced the 3 blue caps but when I power this on now, the fan spins up for not even a second and then stops.

Sounds like a good start. The blue caps (I hope the replacement caps have the X2 certification the original ones had, otherwise they might be a fire hazard) often die in old supplies, and if they die smoking, they don't harm anything else. So just replacing those blown caps is the right thing to do.

geiger9 wrote on 2021-06-09, 01:34:

I tried connecting a HDD and when I flick that big red power switch, the drive and the fan spin up and then promptly spin back down.

Did you connect just the hard drive? That might be too low load on the +5V line, and overvoltage protection might kick in and shut down the supply. A typical 386 mainboards takes one to two amps on the +5V line. Maybe the supply is fine (but clean it, anyway!), and you just need to connect a mainboard to the supply, or at least multiple hard drives to make sure it gets sufficient load.

Reply 11 of 15, by geiger9

User metadata
Rank Newbie
Rank
Newbie
mkarcher wrote on 2021-06-11, 19:27:

Did you connect just the hard drive? That might be too low load on the +5V line, and overvoltage protection might kick in and shut down the supply. A typical 386 mainboards takes one to two amps on the +5V line. Maybe the supply is fine (but clean it, anyway!), and you just need to connect a mainboard to the supply, or at least multiple hard drives to make sure it gets sufficient load.

No it was the motherboard and the HDD. Perhaps I can try connecting more to it and see what happens.

Reply 12 of 15, by mkarcher

User metadata
Rank l33t
Rank
l33t
geiger9 wrote on 2021-06-11, 19:41:
mkarcher wrote on 2021-06-11, 19:27:

Did you connect just the hard drive? That might be too low load on the +5V line, and overvoltage protection might kick in and shut down the supply. A typical 386 mainboards takes one to two amps on the +5V line. Maybe the supply is fine (but clean it, anyway!), and you just need to connect a mainboard to the supply, or at least multiple hard drives to make sure it gets sufficient load.

No it was the motherboard and the HDD. Perhaps I can try connecting more to it and see what happens.

Motherboard and HDD should be enough, if the supply is working properly. I'm afraid there likely is another thing wrong with your supply, which is most likely not directly related to the blue capacitors. Check the mainboards for shorts, though, especially on the -5V and -12V lines.

Reply 13 of 15, by geiger9

User metadata
Rank Newbie
Rank
Newbie
mkarcher wrote on 2021-06-11, 21:35:
geiger9 wrote on 2021-06-11, 19:41:
mkarcher wrote on 2021-06-11, 19:27:

Did you connect just the hard drive? That might be too low load on the +5V line, and overvoltage protection might kick in and shut down the supply. A typical 386 mainboards takes one to two amps on the +5V line. Maybe the supply is fine (but clean it, anyway!), and you just need to connect a mainboard to the supply, or at least multiple hard drives to make sure it gets sufficient load.

No it was the motherboard and the HDD. Perhaps I can try connecting more to it and see what happens.

Motherboard and HDD should be enough, if the supply is working properly. I'm afraid there likely is another thing wrong with your supply, which is most likely not directly related to the blue capacitors. Check the mainboards for shorts, though, especially on the -5V and -12V lines.

I still haven't had time to further troubleshoot this but I was down looking at the case last night and I noticed several wires that were connected to the chassis by a screw. I'm not electrician so I could be wrong when I think these are ground wires. I should have taken a picture beforehand. I wonder where the other ends are supposed to go?

Reply 14 of 15, by mkarcher

User metadata
Rank l33t
Rank
l33t
geiger9 wrote on 2021-06-15, 11:08:

I still haven't had time to further troubleshoot this but I was down looking at the case last night and I noticed several wires that were connected to the chassis by a screw. I'm not electrician so I could be wrong when I think these are ground wires. I should have taken a picture beforehand. I wonder where the other ends are supposed to go?

If the color of insulation of the wires is black or green/yellow, those wires are likely ground wires. Ground wires are mainly used for two reasons: Either for safety, to make sure that all the parts connected to ground are always safe to touch, or to suppress radio interference. A properly grounded part of the chassis is better in containing interference inside the computer case than an unconnected part of the chassis. It is likely that the ground wires you see are to provide better grounding to detachable parts of the chassis, like side covers. An alternate way of providing ground to side covers is putting brass springs between the parts of the chassis. As these springs are cheap and do not need to be mounted like cables, the springs are more common than grounding cables.

Side note: In PCs, not connecting a ground wire is mostly harmless. You might disturb an AM or FM radio inside the room, and you get slightly less protection against electrical shock. Cases in which these kind of ground wires are the last defense against a shiock are very rare, though. On the other hand, in CRT monitors, there is a very important ground connection from the horizontal deflection / high voltage board to the tube itself (the ground connection of the tube is often called DAG). If you operate a CRT monitor with that ground wire disconnected, you might destroy the electronics inside the monitor. So don't work on CRTs unless you know what you are doing!

Reply 15 of 15, by geiger9

User metadata
Rank Newbie
Rank
Newbie

Hey everyone. Good news! It seems the three caps that were replaced DID make the PSU work again. Here's what happened:

When I first got the PSU back from the shop, it was unassembled. The top PCB was outside of the PSU enclosure. I tried to push it back in but it's very tricky to put back in place because of all the wires. I keep feeling resistance when pushing it back in. I did manage to get it like... 98% back into place but that was a struggle and I felt like the wires had zero give left. This is how I powered on the PSU and it stopped after a second of being powered on.

Flash forward a week later. Someone responded to a Kijiji ad I posted about looking for someone to take a look at the PSU. This ad was posted before I took it to the repair shop. He says he would take a look at my PSU for free because he curious about it. He wanted me to go snap some pictures of the PSU so I did and while I was at it, I figured I would try and power it on. Now at this moment, the PSU was mostly disassembled. The top PCB was outside the PSU enclosure almost altogether. I hesitated a little but then decided what the hell, it's already broken. To my surprise, it turned on and stayed on. This was with only the motherboard plugged in. I tried connecting a HDD and FDD and it still worked fine. It seems the issue I encountered was with how that top PCB was inserted into the enclosure. I can't even begin to explain what happened here. Something about the wires that run from that top PCB to the bottom one I suppose? So it looks like I need to figure out how to re-assemble the PSU without shorting something out.