I'm on my last 7K repair project, another 7904. This one has been a PITA with a failure inside the power supply. I had gotten it down to suspecting something in the HV supply, and set it aside a while ago. I decided to try a new diagnostic method I've been thinking about, but had to at least partly complete another project first. I finally got it far enough along to be usable. It's an experiment driver power amplifier that's ultimately for the lock-in analyzer department. I have two other amplifiers for this purpose, with special characteristics, but not enough power for my plans on the 7904. This third one gets into the 100W and up range, so can easily do what I want. All of these are basically audio power amplifiers modified for the purpose. The two smaller ones (10 and 30 Wrms) have galvanically isolated ins and outs to avoid low frequency ground loops, and built-in ballast resistors and clamping for protection against any type of load. I had to forgo the isolation on this big one, so it's a little trickier to use.
I happened to have this nice medium power stereo amplifier made for driving lots of paralleled speaker load down to 2 ohms. It's specified for 60 Wrms per channel, but can do way more with higher distortion allowed. I modified it with output clamping and big ballast resistance (2, 5, 10, 20, or 50 ohms) selected by hooking up to the appropriate speaker terminals. It also has signal monitoring for the amplifier output voltage and current, and load voltage, on BNCs. The plan was to just drive it with a function generator, but since I had a crappy little junker ICL8038-based generator available, I removed and modified its board to build inside the amplifier, so it could stand alone too. This is only using one channel. I can picture adding both together with ballast if they can track well enough at high frequencies, but the total possible power can't be dissipated in the box anyway. I would have to repackage it all into a real test equipment chassis with enough volume and fan cooling just to use the one channel to full capability.
So it's not quite ready for prime time, but yesterday I finally got it together enough to take a shot at the 7904 PS. I could have rigged this all up easily in simpler form at any point, but I resisted that to keep the impetus on getting this thing mostly built. Really it's just a sufficient audio power amp with some series ballast resistance to be hooked up to a secondary winding on the SMPS, while monitoring the voltages and currents to figure out what's going on. The amplifier has plenty of juice to reach over 80 Vpp at 25 kHz with quite a few watts of load, but only about 40 Vpp is needed for this, so in any testing like this, it's important to closely monitor and control the output level.
Anyway, the 7904 PS was already opened up to expose the high voltage section. I first disconnected the primary winding of the main transformer for safety, and to avoid complications from it trying to start up during experiments. This prevents the big line to raw DC filter caps from charging up through the clamp diodes, which is hazardous. I hooked the amplifier up to the winding that makes the raw supplies for the +/-15V regulators, with 5 ohms of ballast, so it would drive either half of the winding with respect to ground (if it was isolated like my other amps, it could be end to end instead). I set up one of my modified (oil-filled and 500 meg input R) P6015s plugged into the Alden connector, and the scope set up to look at everything as needed. I used the internal 8038 generator set to 25 kHz, monitored with a counter. Then I turned on the amplifier output switch and adjusted the volume control for about 40 Vpp output.
It took me less time to figure out what was wrong than it did to set up the test. The current was way too high, when it should have needed only enough to magnetize the core and charge up all the DC outputs of the whole supply. When it's unhooked and taken apart this far, there should be virtually no real power load after the caps are up. I already knew that the low voltage supplies all came up to nearly normal during the ticking, which is why I suspected the HV or the main transformer itself. In this forced steady-state mode everything came up only about half way. Between the ballast and the circuit and transformer losses, the drive current was limited, and it could run continuously, so no tripping and ticking and trying to interpret fleeting events on the scope. It turns this section of the PS into a linear supply, like one running on a variac, but at 25 kHz instead of line frequency, and bass-ackwards powered from a secondary, so low voltage. Also, there's no need to put it all back together to fire up and see the results, and take apart over and over again to finally be done. I just left the test setup connected but turned off, then worked each thing until I got the answer.
I first tried it with the HV winding unhooked, and then all the LV raw supplies came right up just fine and normal, and the current was good. This showed that the main transformer was OK - a short there is the worst possible outcome as far as I'm concerned. It had to be the multiplier or the cathode supply. I was hoping it was just the first ceramic cap or a rectifier in the cathode supply, but no luck there - definitely the multiplier. I found a spare in the parts dept but of unknown condition, so I had to swap it in and hope for the best. It turned out OK, with the full 20 kV or so anode supply, and current still low.
So now I have to put it all back together and into the scope, so there's still plenty of things to check out, but this is the biggest step. I'm hoping it will be straightforward with minimal grief after all this. I'll also be swapping out this awful P11 CRT if possible. I have the P31 one I saved from that rodent-wrecked 7854 that was at the very beginning of my 7K mainframe repair saga some months ago. I don't know its condition since that 7854 could never be fired up, so I'm hoping for the best. If all goes well this unit should up and running and green by the end of the day.
Anyway, this scheme can be used for other stuff too, like in the LV rectifiers and caps, where it can be juiced up enough to track down problems in a safe, steady-state condition.
Ed
