A few months ago I came across a faulty programmable power-supply (PS) with a 60V/12A maximum rating on each of its two channels. The exact model is DTPS6012 from Horizon, a company I’m familiar with as I’ve used and owned a few of their linear PS’s (such as the DHR40-1). The problem that was observed during initial check at the seller’s location was that upon power up one of the channels behaved as expected, while the other wasn’t regulating the output voltage. The voltage just kept on rising until it was ~10% over the 60V rating, at which point the over-voltage-protection (OVP) kicked in and switched off the entire unit except for the front panel. Because the unit was faulty the price was quite low, so I’ve decided to purchase it and try and fix it. At the very least this could be an opportunity to have a look inside and learn how these things were built back then.
I should note that such a high power rating PS is more than I will probably ever need for my projects. However, I have had some projects in the past where the 2x3A rating of my existing PS’s wasn’t enough, even when I’ve used two such units. Therefore, a more capable PS, even if its not as low noise and ripple, is always welcome. Additionally, as I’ve noted earlier, I have owned and used elsewhere other PS’s from Horizon. I was always happy with the build quality and performance, especially for the price these things could be had on the used market.
Took me a few weeks after purchasing it, but I was eventually able to find some time to work on the unit. My first step, before taking it apart, was to first understand a little better the behavior of the unit. Therefore, I’ve loaded the faulty channel with a programmable DC load and turned it on. What was interesting to see is that at currents of over ~0.5A it was operating as expected. At lower currents of >0.1A, it was able to work down to 2-3V, but would go no lower. At lower currents it would behave just as it did without any load connected to it. This gave a few possible directions to look at, the first of which was that perhaps an internal minimum load path wasn’t operating properly. At this stage it was time to open the unit and have a look inside. Here are a few pictures of the teardown:
The unit is very compact for its power rating obviously, but as you take it apart you can really appreciate just how much stuff goes inside of this very moderately sized PS. There are loads of wires going from one board to the other. Each of them is color coded, and they are all nicely tied together to keep it organized. Date code on most parts is 94-95, the sticker on top of the unit says it left Horizon’s facility on 8-96, so this unit is over 20 years old.
Trying to find faults in PS’s, and many other types of electronic instruments actually, should always start with a few basic steps. First, visual examination for anything out of the ordinary. This can be a leaking capacitor, blown parts, marks of over-heating/fire on the board, and any other number of things. In this case I didn’t notice anything note worthy. Next step is to find some points of reference such as test-points, or voltage regulators that could be used to measure DC voltages. To be able to do this without the unit going into protection mode I’ve loaded the output with a high power 8ohm resistor I typically use for amplifier testing, to provide the minimum load needed for operation. As can be seen in Fig. 8, there are a few voltage regulators, including a 7805, a 7815, and a 7915. All these regulators behave as expected with the output voltages within the expected ranges.
Guess there’s no luck for me this time, the problem won’t present itself without digging deeper and trying to actually understand what’s going on there. My initial guess was that the fault would be in the minimum load path of the regulator, which would either be a high power resistor, or a high power transistor (or a combination of the two). Therefore I’ve started looking for such devices on the regulator board. I didn’t find any resistors with significant enough power rating, however, there are 2 transistors mounted to the heatsink along the 78XX/79XX regulators. The 2 devices are an IRF640, and a larger BUK436. By looking at the board and following the path with the continuity mode on the DMM it was easy to see that the IRF640 had a 0.51ohm resistor connected between its source and GND. The BUK436 had a 0.47R connected between its source and a negative supply of ~-3.3V. Negative voltage? sounds like the minimum load path was indeed found. On the working channel this resistor had 350mV over it, or ~700mA through it. On the faulty channel, 0V across this resistor, indeed no minimum load is available on this channel which causes the problem. Next step is to find what is causing this path to misbehave.
Since this is a small degeneration resistor placed at the source of a high power MOSFET, it fairly reasonable to assume the MOSFET is dissipating the power, while the resistor is there to allow the current to be regulated by means of translating it into a voltage. Resistor checks out with the correct value, so it must be the MOSFET or the control circuit that drives the gate. The MOSFET had no shorts, but since I was too lazy to take the MOSFET out and check it by itself to make sure it is working, I’ve instead measured its VGS while the circuit is active. This was close to 0V, which means the problem is probably somewhere in the control circuit. Therefore I’ve had to trace the gate of the transistor back to the circuit that was driving it. The control circuit is built around one of the channels of an OP400FY operation amplifier (opamp). The interesting thing about this circuit is that in the faulty channel the DC voltage at the positive input of the opamp was already significantly different than on the working channel. Since this voltage is set by a simple voltage divider of the supply rails, this seemed strange. Both resistors, as well as the filtering capacitor measure well, so what else is there? must be the opamp itself then.
Since these opamps are socketed, I could take the opamp out of its socket, and swap it with the one from the working channel. Guess what? the problem follows the opamp, original faulty channel is now working well. Looks like I got lucky after all, as its easy enough to fix. A new opamp of the same part # was ordered (despite the very high price of ~17$ per unit), and indeed after replacement both channels are now working well.
Overall the price of the unit and replacement part was just ridiculously cheap (~60$) and it only took a couple of hours of work. Its high power rating adds additional capabilities to my working bench, so I couldn’t be happier. Now all that is left to do is get a new knob instead of the one missing from the front panel. Getting one new (463-8457/020-3520 + cap + nut cover) including international shipping would end up costing almost as much as I’ve spent on the unit itself. Therefore I think I’ll wait until my next visit to the junk yard, and see if I can find one that I can salvage from another Horizon PS 🙂
The visit to the junk yard proved fruitful indeed, a matching knob from a cheaper and faulty Horizon PS was found and installed on this unit. It was even free of charge, so it was worth the wait instead of ordering a new set of knobs.