Noisy oscilloscope measurements

[wsowa]

So I've build an non-isolated SMPS, an AC->DC buck converter essentially. I run into some problems I described here: https://www.eevblog.com/forum/projects/acdc-buck-converter-noise-on-input-causes-heavy-ringing-on-output/msg1692356/#msg1692356. I have recently started some more experiments following suggestion in above thread that the noise might be picked up by probes/scope and I got to the point where I think I don't know how to use oscilloscope properly. Thus a thread in "Beginners" section.

I thought I'll measure my SMPS using poor's man diff probe, i.e. use just two channels per measurement since the suggestion about ground and EM coupling problems. I have probes with the crocodile GND wire detached and no probe ground connected with "spring".

So I have mains conected into the SMPS' input on a PCB and also to the mains L and N at the PCB terminal are connected to 2m long pair of wires open at the end with the switch on the hot wire. See the diagram below for a schematics and the scope measurement taken when I engage the switch: See scope1.png attachment. Clearly the output of the SMPS is noisy.

Then I though "I'll measure just the mains and check what happens when I click the switch". But I left two other channels (for no reason really) connected to the PCB which in turn was not connected to anything. It was physically 50cm away with no power connected to it, just with probes on the output. Here's what I got: See scope2.png attachment
Still noise at SMPS output!

That brings be to the conclusion that whatever I've seen on the scope and posted in the thread I linked at the beginning was likely a measurement noise. So the big question is what am I doing wrong?! Why do I see probes on remote unpowered PCB picking up a noise from switch of unloaded mains line? How to avoid it?

[wstrzalka]

Have you tried putting the circuit in the metal can (Faraday cage)?
This will tell you whether it's the circuit ot the scope picking the noise ...

[CatalinaWOW]

There are two general ways this noise can get into the scope, ground problems and air paths (call it EMI).  For things like your setup an air path seems most likely.  I suspect you have the unused channels in high impedance mode. If your scope supports it try putting them into 50 ohm mode, or just put 50 ohms across the inputs.

All measurements require wires which act as antennas, and load the circuits being tested.  Drawing a circuit with a simplified model of the scopes input (usually just a resistor and capacitor, but maybe an inductor also) can help you understand these effects, and help you make the compromises that are always required when hooking observation lines to a circuit.