Isolation transformer solves only isolating from distribution network. First moment you connect ANY part of circuit on isolated side of transformer to a ground clip of scope YOU and your circuit is NOT isolated anymore. You just referenced it to scope ground that is earthed... Also it doesn't solve anything if you want to measure on BOTH primary and secondary side of SMPS at the same time with same scope..
Or if you want to measure voltage on inductor AND output voltage at the same time.
I don't know why the dangers and difficulties of SMPS testing have take on such mythical proportions compared to other similar devices which have been around for many years. There's no special sauce, extreme energy levels beyond compare or complications beyond understanding of anyone with basic knowledge of the subject.
First, while isolation may protect you from touch voltage in a basic setup, I'm not using it that way here, so there is still a touch voltage hazard if you don't use the hands-off method. This is no different than using diff probes--if you are probing circuits by hand, they're live to touch. I'm intentionally referencing a point in the circuit to ground, the most common point being the negative terminal of the input bridge. And yes, in many cases you CAN measure both primary and secondary sides simultaneously with a single non-isolate scope with no voltage hazard or damage. I'll provide an example here, measuring "the voltage on an inductor AND output voltage at the same time".
I have an old Delta ATX SMPS, unfortunately it is non-PFC, which makes it a lot less interesting, but it will still serve. It's the only handy 'laying around' PSU I have right now. I take the case cover off and add two test points at the + and - of the bridge, where there just happen to be handy empty holes. I then install a jumper to tie the negative of the bridge to ground. Then the PSU gets plugged into the isolation transformer.
![Probing SMPS with oscilloscope 1030214-0](https://www.eevblog.com/forum/index.php?action=dlattach;topic=246472.0;attach=1030216;image)
![Probing SMPS with oscilloscope 1030218-1](https://www.eevblog.com/forum/index.php?action=dlattach;topic=246472.0;attach=1030220;image)
The ground clips are initially removed from the probes, and here I've connected one probe (1X) to the 3.3V output and the other (100X) to the +BRIDGE test point I installed. Unfortunately the 100X probe hook is obstructed in the picture and I've already put everything away. I turn the power supply on and observe the result on the scope, then set the trigger to capture the noise pulse on the 3.3V output. Note that CH1 is +350V, CH2 is 3.3V or so, one primary, one secondary and nobody died.
![Probing SMPS with oscilloscope 1030222-2](https://www.eevblog.com/forum/index.php?action=dlattach;topic=246472.0;attach=1030224;image)
![Probing SMPS with oscilloscope 1030226-3](https://www.eevblog.com/forum/index.php?action=dlattach;topic=246472.0;attach=1030228;image)
![Probing SMPS with oscilloscope 1030230-4](https://www.eevblog.com/forum/index.php?action=dlattach;topic=246472.0;attach=1030232;image)
Now I've said 'hands off' for safety, but in reality I often do some careful poking around. I'll bet the differential probe advocates do so as well and probably with both hands at times. As long as I use one non-grounded probe and keep my hand behind the guard, I'm pretty safe. The main danger is damage due to shorting two close points with the probe, a danger that has nothing to do with isolation or probe type--it can happen with a DMM as well. So I find an interesting point on the input side of the main transformer, attach a test point and hook it up.
![Probing SMPS with oscilloscope 1030234-5](https://www.eevblog.com/forum/index.php?action=dlattach;topic=246472.0;attach=1030236;image)
![Probing SMPS with oscilloscope 1030238-6](https://www.eevblog.com/forum/index.php?action=dlattach;topic=246472.0;attach=1030240;image)
![Probing SMPS with oscilloscope 1030242-7](https://www.eevblog.com/forum/index.php?action=dlattach;topic=246472.0;attach=1030244;image)
Well, there's some noise, which is what happens when you don't have ground clips. Clearly we can see what part of the CH1 signal is correlated to the noise (not that it was a big mystery), but if we want a more accurate and proportional signal, we can connect a ground clip
to the ground reference point. It's tempting to use a more convenient ground point, but that's where the errors can start to happen. Add the ground clip and the signal looks good.
![Probing SMPS with oscilloscope 1030252-8](https://www.eevblog.com/forum/index.php?action=dlattach;topic=246472.0;attach=1030254;image)
![Probing SMPS with oscilloscope 1030256-9](https://www.eevblog.com/forum/index.php?action=dlattach;topic=246472.0;attach=1030258;image)
And there you have it. This is, IMO, a completely safe--as safe as anything else with exposed high voltages--method of probing an SMPS. Yes, is requires some knowledge, planning and care--but differential probes are no more an excuse for carelessness than isolation transformers are. Stupid hurts, no matter how much gear you have. Obviously in this case, the ATX-type PSU has a grounded-common output, which may not be the case in other types of devices, so as my favorite math professor used to say, "
mutatis mutandis".
As for all of the other issues that have been raised, I don't want to bother quoting everyone, but every system has it's advantages and disadvantages. I can probe 4 points simultaneously this way with one scope--that would require 4 differential probes. A cheapo diff probe might be less expensive than my iso-variac, but then you start talking about 12-bit LeCroys and Rogowski coils. I'm not an SMPS researcher and if I was I wouldn't be asking someone like me for advice on EEVBlog, would I? I'm just fixing stuff and this is one way to do that.
EDIT: My pictures got out of order somehow, I'll maybe sort them later.