Working on power supply, discovering more issues with the power system of the IBM 1130

APPEARS THE POWER SUPPLY ISSUE IS NOT A WEAK CIRCUIT BREAKER

I had reset the circuit breaker and powered up the IBM 1130 to see if it ran for a bit and turned off, but it tripped immediately. Pulling the supply out and over to the workbench, I fed it power and applied a modest load. It immediately collapsed, the symptom being a dramatic drop in output voltage with a large demand for current. 

My power supply could only deliver 5A to the power supply thus I can only ask my electronic load to pull about 5A before my bench supply becomes the limiting factor. That was not enough to pop the circuit breaker, but it did hold for the 5A being consumed. 

If the output voltage had remained at +6 I would suspect the breaker couldn't handle greater than 5 but less than 24 amps, but the sag tells me the supply itself is faulty. I will turn my attention there.

This supply takes unregulated DC, 8 to 9 volts, with an unusual connection topology. The minus side of the raw DC is connected to the input of this power supply while the plus side of the raw DC is hooked to the regulated 6V output side! The supply works by floating the common (ground) side of the supply output to maintain 6V between common and output terminals. 

It is much more typical to have the minus side of the raw DC tied to the common output, with the plus side as an input.  This makes it very difficult to take a modern power supply and installing it as a substitute. When I hooked up a modern buck/boost unit this way it passed the full raw DC voltage to the output terminals. Not a good situation. 

BENCHWORK ON THE +6V POWER SUPPLY (REGULATOR)

I took apart the power supply and removed the six Motorola 108 transistors that comprise the output section. These are PNP Germanium power transistors thus they don't provide easy evidence of failure with a VOM, unlike Silicon transistors. The only good method would be to use a transistor curve tracer. 

The supply uses six aluminum heat sinks, each with a model 108 installed, wired in parallel. The bases are all tied together, same with the collectors. The emitters have a 0.1 ohm resistor connected to a common bus across all the transistors. The six heat sinks are electrically connected to the collectors of the transistors and aluminum bars connect them together. 

After unscrewing the connections, the six heat sinks are separated thus the collectors are not tied together but the base and emitter buses are hooked together with wires. Because of this, I can determine which transistor(s) are not working properly and replace them, because I can connect one collector at a time. 

I unwired them and now have six separated heat sinks with the transistors removed. I can install one transistor at a time in a single heat sink and test them. The supply is rated at 24A, thus each transistor is rated for 4A capacity. I can easily drive this level with my bench supply and electronic load.

PREPARING TO LOAD TEST THE SUPPLY WHEN NEW TRANSISTORS INSTALLED

In addition to testing out each transistor individually as described above, I want to load test the entire power supply after it is repaired but before it goes back into the IBM 1130. To do that I need to sink 24A at 6V, dissipating 144 watts of energy. A resistor to handle that would be 0.25 ohms but it is not reasonable to buy one that can handle 144 watts or more. Instead, I will use parallel resistors to divide the load and allow the use of modest sized resistors.

I chose to use 16 10W ceramic resistors of 1 ohm value. I will wire pairs together in series to form 2 ohm units that handle 20W. These pairs can be wired together in parallel, with eight pairs giving me 160W capacity at 1/4 ohm. I can start with two pairs, then four pairs, six pairs and finally all eight. That will let me start with a 6A load, building up in steps of 6A additional until I reach full capacity at 24A. 

AC WIRING IS SCRAMBLED IN THE IBM 1130 I AM RESTORING - UNFORTUNATE'

I was investigating the lack of 7.25VAC lighting power in the system, tracing all the wires and checking for voltage at various points. This power supply is fed by fuse F7 which should be a 1A cartridge. The fuse holder was empty, but I put in a cartridge to test. 

To understand what follows, keep in mind I was carefully staging my tests, section by section, inserting fuses only when I tested that portion of the machine. Thus, this test was done with only two fuses inserted - F5 to provide the 24VAC that sequences power-up and F7 to deliver 115VAC to the lighting power supply.

I worked with the wiring diagrams and beeped out each connection in order to find the fault. When I beeped back from the lighting power supply transformer to the SAC power distribution terminal block, I found the wire was hooked to the wrong place!

Instead of taking power from the output of fuse F7, it was tied to the convenience outlet fuses. Thus it would have had power if all the fuses were installed but I caught it because I isolated sections in my testing. The worst part of this is the lack of protection from overload this causes. The convenience outlet fuse is 6 1/4 A but the lighting supply should be protected with a 1 A fuse. 

I had previously found the cooling fans turning on when they shouldn't, going directly to the 230V supply instead of coming off the 115V stepdown transformer. I still don't know if the fans were replaced with 230V blowers or they are just running on borrowed time at twice the rated voltage.

Still, with two wiring deviations discovered already I think I need to go back to square one and validate every link. I want this put back to the correct wiring before I proceed further. Good power is everything when you deal with a vintage system.