I recently picked up an old Hallicrafters power supply at a garage sale. The person selling it had tested the output voltage and labeled the supply with it. He also showed me around his radio lab before I left with a full trunk of test equipment. This particular HP-1 was producing almost double what the front panel said. This is true of both ranges: the 8V range maxes out at 14.29V and the 16V range at 28.15V with no load. That seemed unlikely to be a coincidence given the similar ratios, and we both speculated that the supply had been modified.

Once I got the HP-1 home and started looking up the schematics, I found out that the HP-1 was the kit model of the HPW-1, available for a discounted price of $50 versus $75 for the factory wired unit. The power cord was a good hint that some changes had been made, having a grounded 3-prong plug with some electrical tape around the base. When I cracked the chassis open, there was a piece of what looked like the original line cord still being used inside. The new line cord was connected to the old one with wing nut connectors, and the cord’s ground lead was taped off and hanging out doing absolutely nothing.

This is an impressive unit, with more than half of the interior volume filled by the transformer and choke. The whole thing weights about 19.4 pounds. The transformer is made of a large ferrite core with a relatively heavy gauge magnet wire around it. The variable tap uses a pair of metal contacts and an exposed section of the winding, which could use a little bit of cleaning but still makes good contact. There are not many other components in this unit, mostly the two meters, one capacitor, a choke, and the neon lamp. I considered replacing the lamp, but with a fuse in place before the switch, failure at the lamp should not pose a risk to the transformer.

My first guess was that someone had tried to convert this into a 30V supply later in its life, so I took some time to trace out the connections from the line cord, through the windings, and into the rectifier. My resulting schematic, drawn from the actual circuit, matched the original perfectly. There were no modifications beyond the line cord. Confused at why the voltage was so much higher, I built a similar power supply in CircuitJS. I don’t think it can do a pair of variable taps like this transformer uses, but the other components are all supported. The simulated supply has the same behavior and puts out very similar voltage.

Having confirmed that all of the internals were working, I decided to keep the extra voltage and upgrade the tank capacitor to match it. The original electrolytic was a Sprague rated for 25V, and while it was still working, I don’t know how much longer it would have kept working at the new voltage or in general. It’s a nice capacitor, with a foil wrap inside of a plastic shell, and can still produce a good spark. I washed out the chassis with isopropyl alcohol, then wiped down the front panel and inside. It didn’t get all of the dust that had accumulated, but at least you can tell which wire is which color again. The wire is all solid stuff, cloth wrapped and still in good shape.

The new capacitor is much smaller than the old one and did not come with screw terminals, so I added a pair of leads with crimped and soldered terminals for easy replacement later. All put together, the new capacitor with leads is the same height as the old one. I added a bleeder resistor to drain the capacitor once the supply is off, 1W at 470k, and a pair of X1/Y2-rated safety capacitors to help filter out any noise coming in the line. Between the transformer, choke, and huge tank capacitor, I can’t imagine this unit would be very susceptible to noise in the first place. The original design did include a fuse, but the fuse is on the neutral side after the switch and indicator bulb, which is not ideal. I think they may have done that to use it as a mounting point, since there is nothing else in the front panel that is connected to the neutral side.

I recorded most of the repair process and put it on Youtube:

The revised schematic and simulation source are on Github.

New Parts

I only added a few new parts: