I've restored an IBM 5160 XT vintage PC (https://davidegironi.blogspot.com/2023/12/vintage-ibm-5160-xt-pc-restoration.html).
It's based on MS-DOS 3.3, sometimes I boot it up, just to check if it works. After a few month the boot process ends up with an error. It seems the 25Mb Hard Drive has not been recognized, so the O.S. it's not been loaded. The only way to boot it is to turn the PC off and on a few times within minutes.
The first thing I've done was to investigate the Hard Drive. I though it was a problem related to capacitors on the Hard Drive electronic. Soon I've realized it has nothing to do with this. Indeed the power lines of the supply are heavily out from recommendation.
With PSU disconnected I get
- +9.7V (on the +12V line)
- +4.7V (on the +5V line)
With all connected (and PC powered on)
- +10.5V (on the +12V line)
- +4.4V (on the +5V line)
Negative lines are ok.
I've posted an help request on vcfed forum, https://forum.vcfed.org/index.php?threads/ibm5160-psu-output-voltage-issue.1244144/. With the help on people of the forum I've learn that usually that era PSU has to be loaded to pull out the correct voltage. But even loading it does not help, this means the IBM loads are ok (motherboard and peripherals).
So I disassembly the supply on my bench and start the repairing adventure.
My supply is a 130W IBM 5160 Schrack Elektronik EG07168-E PSU.
First step was to check capacitors, they are all ok. I've even momentarily replaced those with new one. Nothing helps.
I've then posted an help request on the more specialized forum, the eevblog forum: https://www.eevblog.com/forum/repair/vintage-130w-ibm5160-schrack-elektronik-eg07168-e-psu-repair/.
People on the forum guide me on tests. During tests I've made a giant wiring mistake putting the mains on the DC rectified high voltage input of the PSU. Luckily I've "just" fried the BUT11A main switching transistor and the input 2.7ohm resistors. I've then replaced the main transistors and input resistors but this not fix my under voltage issue, it works again but the issue is still there.
Diodes also seems ok.
It's pretty difficult to check this board cause I'm not an expert in swathing PSU and also there is no schematic available. I've ask Schrack for a schematic without luck.
There's a "mysterious" board on this PCU, which have some SMD component, and it seems double sided with carbon printed resistors. I try to reverse engineer it. It seems this board is the one that drive the switching of the transistors, and so the output voltage.
Also I've try to reverse engineer the output stage of this PSU. It's difficult cause there are unbranded inductors with multiple wires and without de soldering it or experience on switching supply, drawing the schematic of the board can be difficult.
Then the turning point. I try to heat the board up. Heating the "mysterious" SMD board makes the output voltage decrease, and it makes sense looking at my reverse engineered schematic.
The heating test effect, in conjunction with the help on the eevblog forum post helps me finding the issue. It was all related to the 1N823 thermally compensated zener diode mounted on the SMD board.
I've changed it with a standard 1N4735A 6.2V zener diode and the voltage goes back to nominal values. Heating it up does not makes the voltage changes too much, so I'm happy with that diode. Below the testing diode, later I've soldered it in a better way.
Fixed issue. After reassembly all the things the Hard Drive keep working again, and the PC can now boot.
Find the complete story on the eevblog forum. Again thanks to the help of people on the eevblog forum and vcfed forum this IBM vintage PC is working again.
Notes
- read risk disclaimer
- excuse my bad english