Blew bench switching PSU, did I make a mistake probing?

[rwgast_lowlevellogicdesin]

I have a couple cheap switching PSU's and I have been working on trying to quite them down. I don't know a lot about how larger switching supplies are designed, I believe this type is a flyback supply? I had done a pretty decent job of getting most of the common mode noise off the output and dampening the ripple. Today I decided to try adding gate resistors and maybe some caps to the FETS. I wanted to probe the FETS to see how bad they ring. The AC side of the supply  ties the ground directly to chassis, does not touch the circuit board at all. That means everything on the PCB is isolated correct? I connected my scopes clip to the chassis/earth ground and then probed the gate of one of the FETS and the whole PSU died on me. Admittedly I might have slipped and shorted the gate and drain with my probe, which could have killed the supply. So did I kill the PSU because I probed something wrong, or because my hand slipped?

[wraper]

That means everything on the PCB is isolated correct?

Nothing is isolated on primary side. You should be happy you (supposedly) did not blow your oscilloscope. Not to say the vast majority of oscilloscope probes cannot probe voltage this high, even if you used isolation transformer. Most of them top at 300V DC, that drops with frequency increase. Flyback PSU can easily have over 500V at MOSFET drain.

[rwgast_lowlevellogicdesin]

Ok but my clip is on earth ground, so everything is at the same ground reference? Why would there be any issues?

Also if the mains earth lead does not connect to the board wouldnt that mean its not using earth ground as its reference? The dc ground post is definatly floating.

[wraper]

Ok but my clip is on earth ground, so everything is at the same ground reference? Why would there be any issues?

Also if the mains earth lead does not connect to the board wouldnt that mean its not using earth ground as its reference?

I edited my post to add more info while you typed this. No, it's not OK even by attaching earth ground (not that you'd get any useable measurement anyway with such probing)

[rwgast_lowlevellogicdesin]

I measured the voltages on the fet legs and they were at about 160v referenced to earth. Why wouldnt i get usable measurements with the ground lead on earth ground, and if my voltages arent swinging above 400v wouldnt i be good?

Btw there were no sparks no smoke nothing the psu just turned off and wouldnt come back on.

My bad i guess the primary side wouldnt be floating due to the neutral ground bond, but still im not understanding why an earth referenced probe would cause an issue.

[George Edmonds]

Never ever go near to the primary side of a switch mode power supply with a class 1 oscilloscope.

What you think is the 0V/ground of the power supply  primary side is actually sitting at half utility supply voltage and capable of proving many amps to the  oscilloscope ground.

If you have a death, or wish to blind yourself carry on/

G Edmonds

[magic]

It's not at half mains voltage, but at full mains voltage for half of the time. If in doubt, do a SPICE or pencile-and-paper simulation of what happens on the output of a rectifying bridge connected between line and neutral.

Not sure how the PSU died, but there are at least a few problems with such probing:
- as others said, you are exceeding the CAT rating and possibly the voltage rating of typical 10x probes
- your small gate drive voltage is superimposed on large half-wave rectified mains voltage, plus whatever noise that's there
- to the DUT the probe appears as a capacitor connected to half-wave rectified mains plus noise

On the upside, you haven't connected the ground clip to the negative rail of the primary - this would be sparks and smoke.

[rwgast_lowlevellogicdesin]

Ok so would I have been ok to make a measurement like this using a differential probe, if not what is the best way to see how bad the fet is ringing?

[George Edmonds]

As the poster is in the US what appears to be the supply 0V/common is usually at about 160VDC above true ground as in the US the bridge rectifier is frequently configured as a voltage doubler to give a supply DC voltage of about 360VDC across the bulk storage capacitor.  Mains voltage times 2.828.

G Edmonds

[MathWizard]

In a desktop computer PSU, the mains earth GND wire comes in, and it is already externally tied to the neutral wire that will go to part of the full-wave rectifier.

Then after the rectifier, the local primary DC GND, is usually the negative pin of the biggest cap. All the local circuits on the primary side will lead back to there. But then usually, there are high power, flame-resistant, current sense resistors, between this local DC GND on the bottom of the bulk capacitor(s), to the negative terminal of the rectifier. Then some chip will monitor the current returning to the rectifier.

And usually there are some very high voltage, small unpolarized caps, between the mains-earth-GND, and this local DC-GND on the primary side.

But then, on the secondary side, the local DC-GND is connected to the mains-earth-GND. So that just like with audio equipment, the GND on stereo input/output, is also mains-earth GND. And you can hook a stereo into a computer motherboard or soundcard nop problems.


In general any time I'm working on circuits, especially mains powered ones, I'll solder on a few wires, I'm sure that's saved me a bunch of headaches. Now I also have an isolation transformer, so I could touch one hand to the primary side of some DUT, and not have current flow through me and into the floor, and back to some mains-earth GND wire. But touch two wrong places on the primary side, and then I would still get a shock, between the parts of my body touching the primary side. So that's another reason to still solder on test wires.

Any time I work on SMPS, I always have a big GND wire on both the primary and secondary sides. And always remember, only probe one side or the other, and make sure to unhook any probes from opposite side. So far, I haven't damaged myself or my equipment. But that's with an isolation 1:1 transformer.