I have been doing further analysis of the failure of my VAXmate’s H7270 PSU. To start with the schematics are now greatly improved. I am still not sure how correct they are, but here are the latest and greatest:

H7270 Primary Side

H720 Secondary Side

I am fairly sure that the problem is that the primary side is detecting an overcurrent situation. What I am less sure about is whether the overcurrent is real or not. Below is an oscilloscope trace that I think shows this. In this trace the channels are set up as follows:

1. Ch1: NE555 output, trigger for the one-shot on the negative edge.
2. Ch2: Vcc (pin 7) of the UC3842 PWM.
3. Ch3: SCR gate of D19, between R14 and R15.
4. Ch4: Source of Q1

Clearly the gate of the SCR (Ch3) goes high and the 555 stops oscillating. I am a little unsure if I should really be seeing so much spiking on channels 3 and 4.

I decided to look at how the UC3842 PWM power supply is behaving. Accordingly I set up the channels as follows:

1. Ch1: NE555 output, trigger for the one-shot on the negative edge.
2. Ch2: Drain of Q1 (pin 2)
3. Ch3: Cathode of D6
4. Ch4: Anode of D7

This is what I got:

I am seeing a lot of large spikes on the drain of Q1. I wonder if they are enough to make R13 detect an overcurrent? I don’t know enough about switched mode power supplies to know if this is expected behaviour.