Finally I am back to the 2200 restoration. I managed to get the switch mode PSU running before the project pause. One of the reasons for the pause was that I concluded that it would be rather hazardous to work on this machine since there are live voltages coming from the PSU module in the back going to the control logic which is on a board sitting in the backplane so some signals in the backplane are live.
I thought I would hook up the machine to a regular PSU and found an old ATX PSU that I pulled out of my fathers old machine when I recapped it. The PSU also got a re-cap at the same time. The capacitors were really bad. Then I tried it. For some reason it delivered 5.8V on the 5V line. The 3.3.V was fine. Even with quite some load it still showed to high 5V. I tracked down a schematic for the control IC and the example design in there matched very well with the PSU I had. I found a resistor in the 5V sense path that had increased from 11k to 15k. Changing this one and the PSU now gave 5V out.
The ATX supply had +5V, -5V, +12V, -12V that was needed by the 2200 machine, but the deflection board also want to have +25V and -25V. I found a transformer that was suitable and designed a quick little linear supply. All PSU leads were connected to the DD50 connector.
Now I had a PSU that could supply everything but the voltages required by the CRT so I could at least install all boards and check for some life-signs of the CPU.
But before that I wanted to make sure that the PSU filter and control circuitry board was OK. I feed voltages in it from a lab supply to check that the tantalum capacitor wouldn't short circuit. I hate when that happen. Nasty fire or spitting hot metal around.
I know that someone has worked on this machine before and while checking it for voltages my I fell on the new blue Rubycon capacitors that were installed. The probably should have been tantalum capacitors once upon a time since the rest of the +/- 12 V chain was using tantalums. But even worse, it wasn't even the right value. The schematic said 22uF / 15V while the installed blue Rubycons were 10uF / 63V. Why do people do things like this?
So those got replaced with tantalum capacitors with higher capacitance.
Next step was to check the control circuitry which do the PWM control of the four big switch transistors on the PSU module. The oscilloscope told me that the signal named 14KC DRIVE had a signal that was 10.5 kHz.
I then connected the 60V SENSE input signal to a lab power supply. At first the frequency increased to around 28 kHz and when getting close to 60V (which is the target voltage of the first buck converter) the frequency got lower again. Perhaps this is how the regulation should work? Not really a PWM signal that I would expect with constant frequency and varying pulse width.
Next step is to start testing with boards installed. First the CPU and then the rest one by one.