Smps capacitor across transformer windings

Ian0

Joined Aug 7, 2020
6,294
Do you mean the 2.2nF class-Y capacitor?
It’s there to reduce EMI.
The bigger it is, the better it works, but above 4.7nF the current through it exceeds safety requirements, so nearly everyone uses 2.2nF.
 

Tonyr1084

Joined Sep 24, 2015
7,093
Note to moderators: thanks for helping me download the page.

C7 would seem to block DC voltages from passing through from the source side to the regulated side. SMPS's are far beyond my ability to follow, so it's quite possible I'm wrong in my assumptions. I would also assume a potential danger of applying high voltage AC across the circuit. But again, remember, I'm in a realm where I probably don't belong.
It’s there to reduce EMI.
The bigger it is, the better it works, but above 4.7nF the current through it exceeds safety requirements, so nearly everyone uses 2.2nF.
This statement lends clarity to it, but I'm still unsure if it's a wise addition to the circuit. Granted, it's unlikely an air conditioner would become a hazard, but if the circuit were used elsewhere, depending on the usage, could potentially present a danger. OK, 2.2nF keeps it within the safe range. But what if it shorts? And no - I'm not trying to take over this thread, just posing a question for consideration.
 

Ian0

Joined Aug 7, 2020
6,294
It is ubiquitous .
There will be one in every switched-mode power supply that you own, and you’re not dead yet.
I think there is something in the design of a class-Y capacitor that prevents it from failing short-circuit. Someone more knowledgeable than me on capacitor design may clarify.
 

DickCappels

Joined Aug 21, 2008
9,317
Y capacitors are designed to fail open. X capacitors self-clear shorts by blasting the short out of the electrical path inside the capacitor.
 

Thread Starter

Missio2468

Joined Mar 18, 2022
33
Do you mean the 2.2nF class-Y capacitor?
It’s there to reduce EMI.
The bigger it is, the better it works, but above 4.7nF the current through it exceeds safety requirements, so nearly everyone uses 2.2nF.
Ok but what is the safety requirement you are talking about?
Also is there any calculation or theory. i know about emi noise but cannot calculate the values. Is there any formulas behind it that you can share??
 

DickCappels

Joined Aug 21, 2008
9,317
Also can you share any step by step process to design EMI filters
Having been there, there is none. Do what most engineers do -buy a ready-made filter and test it. The testing is the hard (Expensive) part.

If you can buy or rent US$50,000 to $100,000 worth of test equipment then you can design your own filter, and then you won't know until you try it in the actual product because the conducted EMI filter is not the whole solution.
 

Ian0

Joined Aug 7, 2020
6,294
I found "EMC for Product Designers" by Tim Williams ISBN 0-7506-1264-9 to be very helpful, but without a spectrum analyser you'll never know whether you have succeeded or you haven't.
Spectrum analysers are a lot cheaper than they used to be - a couple of thousand quid should get you one. That then should be enough to tell you that you won't be going home disappointed when you take the product to an official test house.
But the first stage is to design the product so that it doesn't leave you with a lot of interference you have to get rid of.
 

Thread Starter

Missio2468

Joined Mar 18, 2022
33
T
I found "EMC for Product Designers" by Tim Williams ISBN 0-7506-1264-9 to be very helpful, but without a spectrum analyser you'll never know whether you have succeeded or you haven't.
Spectrum analysers are a lot cheaper than they used to be - a couple of thousand quid should get you one. That then should be enough to tell you that you won't be going home disappointed when you take the product to an official test house.
But the first stage is to design the product so that it doesn't leave you with a lot of interference you have to get rid of.
Thank you
 

Thread Starter

Missio2468

Joined Mar 18, 2022
33
Having been there, there is none. Do what most engineers do -buy a ready-made filter and test it. The testing is the hard (Expensive) part.

If you can buy or rent US$50,000 to $100,000 worth of test equipment then you can design your own filter, and then you won't know until you try it in the actual product because the conducted EMI filter is not the whole solution.
Thank you
 

Irving

Joined Jan 30, 2016
3,133
Spectrum analysers are a lot cheaper than they used to be - a couple of thousand quid should get you one. That then should be enough to tell you that you won't be going home disappointed when you take the product to an official test house.
But the first stage is to design the product so that it doesn't leave you with a lot of interference you have to get rid of.
Having been there and done I concur, but even a good SA won't tell you the whole story - and BTW you'll need a big field in the middle of nowhere (hard to find these days) or a proper Faraday cage as well...

If you've ever watched those 'repair an iPad or iPhone' videos where they track down the fault to a 5v or 3.3v power line shorted by a MLCC (capacitor) which they remove and it works again but they don't replace it, and wondered why not? Its because the MLCC is there to remove (ie detune) a nasty low-level EMI spike at some weird frequency due to a PCB power track radiating at its 'self-resonant frequency'... that in reality probably never caused an issue but EMC rules & testing required it!
 

Ian0

Joined Aug 7, 2020
6,294
you'll need a big field in the middle of nowhere
I'm very fortunate on that front here in Lincolnshire, but just up the hill is a transmitting mast known by all and sundry as "the secret bunker" from which I get the internet connection by microwave, so even here is not radio silence.
 
Top