I bought a cheap Chinese SMPS (Mains to 12VDC,30A). Metal-cased. Intention is to use it to drive an amateur radio HF transceiver. Works okay for that purpose but puts out lots of HF RFI.
I came across this interesting article on reducing RFI in SMPS
https://ka7oei.blogspot.com/2013/09/quieting-high-current-switching-power.html?m=1Scoping across the DC out + and - terminals, I could see an obvious 51Khz (presumably switching) signal and lots of ripple. Between chassis and DC - showed 1.56V p-p.
Opening up the case showed some basic input EMI suppression: a class Y cap, class X and a single common-mode choke. On the output, there was a single 10nF (iirc) class Y cap connecting the DC - to earth/chassis ground.
I pulled the three 3500uF output filtering caps (all which had dubious ESR) and replaced with a single 10,000uF 25V cap. I also soldered two 1uF 400V caps in parallel across the output class Y cap.
This has essentially solved the problem. The switching noise between the chassis and DC negative is around 20mV peak to peak and the ripple is substantially reduced scoping across the DC output.
My question is: how safe is this to do, bypassing the class Y cap? I'm on a 230VAC/50Hz mains supply. My (fairly beginners) thought process would be that if a fault were to develop (such as a power surge), the input side protection/transformer would fail before reaching my final output capacitor modification. Am I wrong?