The PS-4 is a variable high-voltage bench supply that was manufactured starting around 1960. It is regulated up to 400 VDC at 100 mA, and it also has a separate low-current negative output from 0-100V in case you like make a manual curve tracer or whatever, as well as a 6.3V AC output for lighting up filaments. It appears to be electrically identical to the Heathkit IP-32 supply, but I don’t have one to compare it to. I regularly use this one as a steady voltage supply for my early one-tube oscillators like this one since the regulated voltage tends to keep the frequency steady. However, it took a little work to get it there, and I figure I might as well write about it in case it help (or just entertains) anyone looking to do the same.
I got this with a caveat: “It powers up, tubes light, didn’t test further.” Now, “testing further” with a power supply doesn’t take much more than a voltmeter, but it’s just as well…I’ve heard enough stories of people turning on old electronics and of people plugging things in to see if they work and finding that the transformer burns up. So I negotiated the “it’s broken” price and figured if nothing worked I would rob it for the 6L6GC’s to some audiophool to break even.
When I got in home, I saw the 6X4 was missing, but the 6L6GC’s were still there. That’s odd. Perhaps someone really needed a 6X4, but why would you take it from the inside of a piece of bench equipment? Clearly if someone was in to harvest tubes for resale they would have chosen the two 6L6GCs next to it…but we’ll come back to that.
The PS-4 happens to contain all three common rectifier technologies at the time:
- A 6X4 vacuum rectifier
- A halfwave rectifier made from first generation “fuse holder” type 1N1081 or similar diodes.
- For excitement, a pair of selenium rectifiers in a half-wave rectifier for the screen voltage supply!
With three high voltage DC circuits there are plenty of electrolytic capacitors to replace, as this had all the original ones. Once I got the thing wired, and took a deep breath and tuned it to “standby”, and watched all the tubes glow. So far so good. I put the B+ voltage to zero, switched to “on” to apply the plate voltage and watched the 6X4 buzz and show some fireworks!
This was the day I learned to check the previous owner’s wiring.
Okay, so the 6X4 was missing for a reason!
Although I had replaced the caps, being careful to observe the polarity markings, or so I thought. I went back to a picture and saw that I had put them in the right way…or rather, the same way the original ones were wired. What I failed to do was check the polarity against the schematic. One of the caps was wired backward, probably arced internally, and the 6X4 did the same.
I fixed backward cap, rechecked all the rest, and plugged in a new 6X4. This time there were no fireworks and I’m happy to say its still there.
Now that things came alive I started checking the output voltages. The B voltage was at 40V when the dial was set to zero, and C-voltage output was actually at zero. Moving the B-voltage adjustment, did nothing; I got 40 volts all the time. The C voltage output worked like it should, so I know at least part of this beast was working.
Lots of staring at a schematic and tracing wires eventually lead me to a subtle wiring issue at one of the tube sockets:
After counting correctly an moving the wire everything worked wonderfully.
After seeing the backward cap and this wiring error I knew for a fact this power supply never worked, and the 6L6GC’s were essentially new tubes. A brand new, never-used Heathkit PS-4! I was able to calibrate the output for 0-400V continuous and I checked the regulation under load and it did as it was supposed to. I feel bad that someone gave up on it, but sixty years later it was finally fixed right.
Replacing the solid state rectifiers
The silicon rectifiers in the PS-4 were of the old cartridge type that fit in a fuse holder. The reason they sit in a fuse holder is that, like fuses, there’s an expectation to replace them. The early diodes only had PIV ratings of a few hundred volts or less, which is why two are used in series in each leg of the full-wave rectifier. The reason you expect to replace one of these is because they aren’t terribly reliable. As big as they are, they are only rated for a few hundred mA forward current, and back then the surge current was not very high. So, in the end they worked like fuses too…don’t worry, there is a real fuse upstream in this thing.
Nowadays a 1N4007 is rated 1000V PIV and 1A forward current, as well as 30 A momentary surge. I just wired four of these across the fuse holders where the originals had been. Normally I try to use original parts in vintage equipment, but this isn’t something with aesthetic effect, and I really intended to use this thing a lot, so I needed it to be reliable. A diode failure in the middle of an antique radio operating event would keep me off the air, and that’s no fun. If a museum curator wishes to restore it back to the original
fuses diodes I haven’t hindered them from doing so as long as there are soldering irons in A.D. 3019 and the nanny state hasn’t banned all electronics that contain small amounts of lead.
The little selenium rectifiers would probably work for a long, long time as there is hardly any current passing through them. In my mind, this is still something that I would feel better replacing. I don’t like the idea of a burning selenium rectifier stinking up the place with quasi-garlic poison gas. If that’s not enough, a more immediate reason to replace them is that such an event would also not endear me to the real brains of the household economy. We don’t want to draw negative attention to the “musty things” that might sneak into the house via the garage entrance, do we?
Selenium rectifiers are usually replaced with a diode and a resistor. The resistor is there to simulate the voltage drop of the selenium piles as well as limit inrush current to the replacement silicon. The original rectifiers in the PS-4 were potted black cubes, rather than the more obvious finned piles, and they were bolted into the chassis. This served as the heat sink. (If you drop a few volts at some amount of current, then Ohm’s Law and the conservation of energy tell us that heat has to go somewhere. (This is why high-current diodes have convection cooling fins even if dropping less than a volt.)
I’ll admit to not using my brain when replacing these. I see something bolted to the chassis, and I automatically think there must be significant heat involved. These things probably drop a few volts each (about one volt per pile, but I don’t know how many piles are in each), and Ohm’s Law says there must be heat generated. The usual rule for Se-rectifiers is to allow for a couple watts of heat dissipation in the resistors that replace them. Not having any 5-watt resistors in the 50-100 ohm range, I made a four-watt one out of four 47 ohm resistors.
I should have looked at the circuit more closely. This “rule of thumb” is based on typical tube voltages in HV circuits of consumer electronics. That usually means a plate circuit drawing 20-100 mA or so. However, a visit to the schematic shows that, in this application, the Se-rectifiers are in the screen bias of the 6L6’s. A few milliamps at the most? So I really overdid it with the power dissipation in the series resistor. The the old selenium rectifiers were bolted to the chassis solely because they doubled as a mechanical terminal strip, not for any real heat dissipation. The good news is that a few volts difference in voltage drop in a silicon diode vs. the Se-rectifier was not really a concern at the nominal 270 volts of this supply, and the 50 ohms is sufficient to limit the inrush on the diodes. Oh well, in another 100 years someone will have a story to tell when restoring this thing when they see my big fat resistors.
Finally, I replaced the cord with a the three-wire polarized version and grounded the chassis, making sure that the hot side of the outlet was fused and switched.