how to test a power diode

[coppercone2]

So you might have come across a diode measurement problem in broken switch mode power supplies you are trouble shooting. I wonder how many people got hung up on this.

I was doing spot testing on components and I thought I found a broken diode until I carefully read the datasheet. It was measuring 0.3V on a diode tester but the Vf for the diode was 1.25V at 8 A.

So I took a power supply, shorted the diode across it and cranked up the current limiting to 1A. When there was 1A flowing through the diode, the power supply correctly read the voltage drop as 1.24V. Much better.

So this is an easy way to check power diodes on power supplies in case you are stuck on component diagnosis. Power components are annoying because your meter does not have enough juice to bias the test correctly, luckily a lab power supply can act as a current source while also reading voltage. This is actually a quick and practical test for trying to pin point faults.

[Kim Christensen]

The "power supply method" works quite well. I always take the multimeter "diode test" results with a grain of salt.
Had one a while ago that checked OK on diode check, but when you put any current through it, the Vf rose to 5V and beyond.

[coppercone2]

Yeah, its actually a great technique. I wanted to make a adapter that plugs into it with some aligator clips to make a elegant 'attachment' for testing through hole parts, and maybe something with pins for SMD parts too. Totally unnecessary but I think it will be nice and accurate.. plus its a fun DIY project. Would be nice to make some fixture while not worrying about RF like I have been.

I am kind of wondering too what can be done for power transistors with a dual power supply, like an adapter that uses one side of PSU for gate control and other to force current. Poor man curve tracer made with pin receptacles for TO220 and larger parts quick testing, maybe with a heat sink and heat sink clip (screwless construction).

I know you can probobly breadboard or solder something up.. but it would just be streamlined. That can have an oscilloscope BNC hookup too, and be custom tailored to 'fit' the test equipment wall. That would be nice for people who are a little 'sick of it', it meaning hookup work. Altoid tin plug in maybe.

[Kim Christensen]

I am kind of wondering too what can be done for power transistors with a dual power supply

I do this at home as a quick transistor sanity check, but it's usually ad hoc on a breadboard with a resistor to limit the base current. Small piece of metal and a binder clip works as a heatsink in a pinch. There are also those clip-on heatsinks normally used when soldering. Usually I keep the on time short enough that my built in fingertip heat alarm doesn't trigger. I do fantasize about making DIY curve tracer now and then but I don't think I'd use it much.
 
I'll also use a current limited power supply and multimeter as a very crude 4-wire ohm meter. ie: Set current limiting to 100mA and measure mV across the resistance and multiply the reading by 10 to get milliohms. Or use 1A and then millivolts = milliohms. Depends on the DUT.

[coppercone2]

I kinda wanted to load test these things though, unless you just mean resistor on the gate?

The diode is OK I guess since its only a 10A but when it comes to like 200A transistors it gets annoying to trust them.. I kinda need to fix this problem because my repair pile ended up filled up with suspicious switchers, the amount of effort required to deal with thermal grease, conformal coating, bolts, shitty heat sink positions, heavy PCB with crappy de laminating traces.. I would like to be very sure about the transistors before I properly install them. It is extremely fucking annoying.

[Kim Christensen]

Well, 200A is a lot...
What commercial curve tracers have is a pulse mode. Instead of driving the gate/base with a DC ramp, they pulse it at a really low duty cycle which lowers the power dissipation significantly. Kind of made it hard to see on the old CRT, but with a modern digital scope/tracer it would be much more user friendly.

[Kleinstein]

One has too keep in mind that the forward votlage depends on the current. With a large diode and relatively small small current the forward voltage can turn out quite small. Usually modern silicon diodes are relatively close to the ideal Shockley diode formular (exponential current / voltage curve) over a large range of currents.

I had just measured a diode in a 6 A rectifier at more current points. I started with the low current (8.4 nA) and was a bit suspicious at first, but it turned out OK. The straight line fit in the graph is mainly based on the last points (due to the log scale). With the more suitable fit on log of data the line better fits the center part and is a little off at the ends - than the slope is also OK for the temperature.