• Please review our updated Terms and Rules here

IBM 5110 power supply schematics?

MattisLind

Veteran Member
Joined
Sep 30, 2013
Messages
1,158
Location
Stockholm, Sweden
A friend of mine got an IBM 5110 which was dead and I am trying to help out. It blows fuses. First he told me the transformer was short circuit. When I looked at the PSU I recognised that it was a switcher. And quickly concluded that one of the chopper transistors was a three pole short.

The chopper transistors were TI made in 1978 but marked T484. Probably some IBM marking which no one has the cross for.

I replaced them with nice high voltage high current TO3 transistors The only thing i could get from the original transistors except for the physical appearance was the polarity.

With a 5.6 ohm resistor on the 5V output I fired up the PSU with a 60W lightbulb in series with one of the mains leads. Nothing happened on the bases of the switchers until I got to around 190 VAC input.
At that point the base went high and stayed there. There seems to be very little power consumption since the bulb goes dark after the initial charging of the input capacitor of the PSU. There is a faint click sound heard from some part of the PSU when turning it on, but cannot really get exactly where. I suspect the sound comes from the transformer. If turning on using variac it happens exactly at the same time as the base voltage changes.

Have measured most of the semiconductors and they measure fine. Except for possibly one that look like a transistor but it hard to tell since it has just IBM number on it. Cannot detect any PN junction on it.

Of course I have no schematics for this IBM thingie. Does anyone have a schematic for the IBM 5110 PSU? I think I really need it to understand what is going on.

Tracing it out is an option but then all those square metal canned IBM ICs makes it very difficult...
 
If the power supply has an AC out plug, the click may be the relay closing to activate the output.

I'm not great with switching PSUs, but when I find that a primary side mosfet has blown up or gone short, it usually takes out whatever circuitry that's feeding it the chopping signal. Check passive components, especially resistors because those can go out of spec or open and cause the primary side to not start switching.
 
I want to hazard to guess the 5110 does not have an AC passthrough and that the relay has more to do with perhaps an anti-overload or crowbar circuit. Have you been able to see that your chopper oscillator is running? Are you testing the power supply attached to the logic or with dummy loads?

I'm working blind here as I've never touched a 5110 myself and google is a mix of front-only photos and deskthority. Snuci has one I think, as do a few other people here but unless Al had them hidden somewhere else I'm not seeing the CE schematics on bitsavers.
 
I add a picture to show what I am fighting:

MexGosO.jpg


I have checked resistors in circuit. There are none that are terribly out of spec. Of course I cannot only tell if they are out of spec on the high side in circuit, but they are quite well. Checked the driving transistor for the choppers and it at least seems to be all right.

It is really difficult to find where the oscillator is among all these IBM metal cans. IBM 22 ... seems to some kind of standard device which I have no clue what it is doing. Knowing that these could be small hybrids which could be just a few transistors it is really hard to tell how this bloody thing works. I am quite confident that the two 6 pin DIL are optocouplers used in the feedback path.

The fact that there seems to be something that measures the input voltage and seems to keep it from doing anything until it is within proper operating range makes it a bit difficult to work with. Even though I have a protection transformer I am a bit reluctant poking around in the circuit with 300VDC everywhere. It would be nice not disable that stupid sensing circuit so that it starts up directly, then I could power the control circuit from a bench PSU.

But all this is hard to do when there are no schematic and at the same time all the ICs and transistors have IBM markings. I really would love to have the schematic. If I cannot find a schematic for it I will probably tell my friend to replace the PSU with a modern PSU. Some PC ATX PSU can probably be made to work. Another possibility would be to replace all logic with a UC3842 or similar IC. But I have no idea what the proper switching frequency is. Given that the transformer looks like it has an air gap and that there is no diode for a core reset path I think this is a flyback design.
 
Hi Mattis,
it seems you've already done all the usual checks but I'm not sure if you tested the ESR of all (even the small ones) electrolytics?. I can't help with a schematic, but I can suggest first to check even the small electrolytics (I had an IBM PSU, in a monitor fail to start because of a 10uF dried out electrolytic in the control circui). Then try to power the output of the PSU (for example the 5V output) with a variable bench power supply, with the primary part unplugged from mains and check at what voltage the feedback path activates (it should activate one of the two optocouplers), of course try to check the phototransistor side directly. Drawing as much of the circuit as possible is another step I usually do when there're no other options. Often it gives an idea of the general blocks composing the PSU.
HTH
Frank
 
This is a challenging one. If I were repairing it I would copy out the schematic completely and for any device, like those IBM devices just draw them as squares with input-output connections etc. I think you have done very well so far.

If you are replacing any devices such as the chopper transistors be sure they are a type suited to the application. I would guess in a supply of this age the electrolytic caps on the output side of the supply would have gone high ESR and low capacity, probably the large capacitor on the line power input side is ok. With a supply of this power output, it will almost certainly have some inrush current limiter system, sometimes that was done with resistor on the input circuitry charging the main capacitor, and then a relay shorts that out sometime after the input capacitor is fully charged.

While it's fairly safe and easy working on the output (low voltage side of the supply) the input side is hazardous and more difficult . To get around this I use a Tek 222ps scope that has entirely isolated inputs and the probe "earth" clip can be connected anywhere. Another method, not quite as good is to power it by an isolating transformer, to be able to scope the chopper circuitry and chopper transistors on the primary side of the transformer. Some people have lifted the scope earth but I don't think that is a very safe idea.

It might be hard to find the schematic for this supply . IBM did not even publish the one for the 5155 computer and I had to copy the whole thing out. Without the schematic or at least 90% of it, it is a very difficult task for repairs sometimes.

In the worse case scenario it might be necessary to open up a suspect IBM can and document what is inside, could be made difficult if potted in resin in there (not impossible).

Fly back supplies generally have limited current output capability, normally for high currents requires two chopper transistors, one switched on for a % of the time for each half cycle. But it is an unusual looking supply, and it really would be good to get it running and document everything possible about it.
 
Last edited:
Thanks for all ideas!

I have replaced the chopper transistors with 2N6547 which are a quite common SMPSU transistor in DEC equipment. Should be quite oversized for this application I think.

I am already using a protection transformer. But still you can get hit even with a protection transformer if you are unlucky / clumsy. I prefer to be on the safe side..

I have checked the capacitors all over the PSU. They measure good both ESR and capacitance. The capacitors on the 5V out give out of range so they are more than 20000 mfd combined, which means that many of them are quite ok. I don't want to desolder them and I don't believe them to be bad.

This is not a terribly high current supply. In total maybe 200W power output.

Checking the output stage by back feeding voltages is a good idea. However I think I will have all 5 voltages in parallel since there seems to be some kind of analogue adding circuit in the output feedback stage.

Since we don't really know if the rest of the machine works or not I don't want to spend too much time on this PSU until I can figure that out. We will probably go for temporary replacing the PSU with a modern ATX PSU. Then if we get the machine running I can revisit this topic.
 
Thanks for all ideas!

I have replaced the chopper transistors with 2N6547 which are a quite common SMPSU transistor in DEC equipment. Should be quite oversized for this application I think.

I am already using a protection transformer. But still you can get hit even with a protection transformer if you are unlucky / clumsy. I prefer to be on the safe side..

I guess the thing to remember is that adding the isolation transformer doesn't in fact add any safety, what it does do is that it allows parts of the primary circuitry, on the line power side, to be connected to test equipment like an earthed oscilloscope's test clip ground wire, without tripping the RCD(ELCB). This of course grounds part of that circuit, raising the potential of another part of the circuit above ground, meaning then that a shock can be acquired by touching just one part of the circuit with your hand and, that current that flows via your body cannot trip the RCD, because of the isolation transformer. Rubber gloves are really helpful here.

So it pays to remember that isolation transformers defeat the RCD that normally would save a person getting a shock from say a one hand contact to a line powered circuit.

The worst case would be touching the body of the scope with one hand (earth) with the scope connected, and touching some part of the high voltage input circuitry with the other hand and the current goes across the chest. THe same applies to working on many high voltage equipments, where the high voltage side is derived from the secondary of a transformer which isolates the RCD.

This is one reason I have gone to the Tek 222ps scope for SMPS repairs, as the workshop's RCD remains operational, unlike the case with an isolating transformer.

It sounds like you know what you are doing with the supply. I really would like to see that circuit if you ever copy it out.Most SMPS's as you know use IC's for the switch mode control, it looks like IBM crafted their own circuit from those interesting modules and discrete transistors, so it's probably a very cool design.
 
I really would like to see that circuit if you ever copy it out.Most SMPS's as you know use IC's for the switch mode control, it looks like IBM crafted their own circuit from those interesting modules and discrete transistors, so it's probably a very cool design.

I happen to come across this picture which matches (i think, I have no idea of the IBM numbering scheme) four of the shiny metal cubes in the PSU:

IBMSLT.jpg


From this page.

Seems to be three transistors and three resistors in there.
 
I happen to come across this picture which matches (i think, I have no idea of the IBM numbering scheme) four of the shiny metal cubes in the PSU:

IBMSLT.jpg


From this page.

Seems to be three transistors and three resistors in there.

That is very good news in that the circuit can be drawn out and the resistors measured and on top of that the transistors could, if required, be replaced with modern surface mount types. So it's likely repairable and probably a logic gate configuration.Even a replica could be crafted with surface mount parts on a PCB as a replacement.

That kind of design with resistors screened onto ceramic and transistor crystals mounted on to it was a technique to make high reliability and temperature resistant modules. It's also possible some other module types contain surface mount capacitors in them too, depending what the module was designed for.

I wonder if IBM ever published a book with all their modules in it or whether they were top secret ?

I think as time goes by its going to be important that these modules get documented as they are are part of computer development and history and it will be required to have the knowledge of them to repair the vintage computers that use them.

I have also seen IBM modules , some much larger memory ones on a short ISA memory card in my 5155. Working currently, but if they fail it would be good to find a way to repair them or replicate their function and appearance.
 
The 5150 floppy controller used the SLT packages as well, which always seemed to me to be a bit strange, seeing as how IBM was working to use commodity parts for the 5150.
 
Hi,
Have you contacted owners of working 5100s? Ideally need to measure a good transistor and replace with compatible one.

To echo what Hugo wrote about oscilloscope safety, I would recommend a high voltage differential probe (I bought a Pintek one for about 150 pounds) or maybe find a scope with two high voltage probes that allows channel 1 minus channel 2 display (ie most DSOs).

Regards,
John
 
Back
Top