It turns out I had mistakenly interpreted the role of the capacitor in question here as an EMI filter for line noise. It looks like a capacitor in that position actually serves as a snubber for high-frequency rectifier switching spikes/noise. Although perhaps unlikely to be conducted to later stages of the circuit (being contained by the reverse-biased rectifier diode), without snubbing that high-frequency noise might be radiated.

I'm specifying an EMI bypass cap for placement on the secondary winding of a line transformer. It feeds a simple linear supply that acts as the bias supply for a DC Lab Power Supply I'm designing as a learning project, (C5 in the schematic below):

enter image description here

I'm pretty sure the right answer is a roughly 0.1uF 100V (maybe 250V) film capacitor because that's what I've seen used in other circuits. But I want to understand how to go through the design process myself, even if I end up with the same answer in this case.

Here's what I have so far:

  • The capacitor shouldn't be bigger than necessary, because it will shunt a significant level of the wanted 60Hz AC current. For example, a 1uF capacitor would shunt about 25mA RMS.

  • The frequency response of the cap needs to be good so it gets rid of high-frequency noise well into the MHz range.

  • If the capacitor is too small, it might not bypass lower frequency noise very well as its impedance at a given frequency is inversely related to its capacitance (C).

Here's what I don't know:

  • The practical range of EMI frequencies we'd be looking to bypass in this sort of situation. I'm not sure what's typically around, and perhaps not all of that gets through the transformer because of its natural inductance or something.

  • Whether the working voltage of the cap needs to be bumped up to accommodate possible high-ish voltage transients that might get through.

  • Why a plain-old ceramic cap wouldn't get it done just fine.

Can someone help me understand the design space for this component?

  • \$\begingroup\$ "I've seen used in other circuits" - do you have a link(s) to another circuit that uses this method? \$\endgroup\$ – Andy aka Dec 3 '15 at 10:32
  • \$\begingroup\$ The Agilent E361XA series bench supplies have caps in these posistions, schematic here a couple pages from the end of the PDF. \$\endgroup\$ – scanny Dec 3 '15 at 16:03

You don't want just capacitors, you want to install an RC snubber on the transformer secondaries. The reason is that when the secondary voltage drops far enough that the diodes stop conducting, you would get significant ringing due to the interaction between the diode's open-circuit capacitance and the transformer's inductance. Here is an appnote on snubber design but it's fairly technical and assumes you know more about the transformer than you are likely to actually know.

A capacitor alone is likely to make the ringing worse. A resistor is necessary, as it is the component that dissipates energy and therefore reduces the Q of the LC ringing.

The exact choice of R and C depends on your selection of bridge rectifiers and transformers and the most accurate way of designing the snubber seems to be empirically, because of all the unknowns in buying a random transformer from the shop without a super-detailed datasheet.

Have a look at this diyaudio thread, where they describe a device to drive the transformer with impulses and thereby measure the effect of the snubber.

My vague recollection of when I last designed a transformer snubber was that the choice of R is important in terms of getting Q down to where you want it but you don't need to be super-precise, and the C (in series with the R) exists merely to prevent the R from dissipating loads of power at 50/60Hz. So you might be looking at a few hundred ohms (totally dependent on the transformer!) and about 100nF of C, but the exact value of C doesn't matter much at all. C just needs to be big enough so that the RC impedance is very close to R at the ringing frequency, and "high" at 50-60Hz.

  • \$\begingroup\$ I think you've hit it on the head William! I was mistaken in thinking the examples I'd seen were EMI filters, rather they were simple snubbers. The PDF you linked to was very helpful and also put me onto the search terms I needed to find a lot of additional resources :) \$\endgroup\$ – scanny Dec 9 '15 at 13:45

I have a few half-baked ideas - they tend to revolve around the fact that there are none of these types of capacitors on the primary side so, I'm tending to think they are there to prevent dv/dt problems on the bridge rectifier inputs causing the possibility of an actual negative voltage been seen on the output. This may occur because of two simultaneous reasons and not individual reasons: -

  • The bridge diodes have relatively high self capacitance and
  • The high current/low current switch operating and generating large (ish) flyback voltages.

Here's my comments on an extract of the circuit: -

enter image description here

  • \$\begingroup\$ Ahh, very interesting. So I'm taking it that it's not common to place a cap in this position (on the secondaries) for EMI bypass and it may be used in this case because of the winding switch. My design doesn't use that sort of thing, so maybe I don't need a cap across the secondary. Another bench supply I've studied does have a bypass cap across the primary, it was a ceramic disc IIRC. Maybe that's the way I should go instead :) \$\endgroup\$ – scanny Dec 3 '15 at 18:38
  • \$\begingroup\$ It's a tricky one and fairly speculative and may come down to know failure mechanisms of the bridges due to dv/dt (somehow)? \$\endgroup\$ – Andy aka Dec 3 '15 at 18:41
  • 1
    \$\begingroup\$ I read something recently that mentioned that you could reduce this capacitance by using fast diodes (low trr). The problem with them is that they tend to have high forward drop thus high dissipation. So this may have something to do with the diode capacitance you mention and some amount of shoot-through that occurs at zero-cross. \$\endgroup\$ – Daniel Dec 6 '15 at 6:24
  • 1
    \$\begingroup\$ ac-dc.power.com/design-support/circuit-ideas/… \$\endgroup\$ – Daniel Dec 6 '15 at 6:33
  • \$\begingroup\$ Ok Andy, your answer really set me on the right track. I've done a little more research based on your direction and I've come to the conclusion these capacitors are very likely serving as snubbers. I hadn't realized it before, but the switching action of the rectifiers when they go from conducting to the non-conducting (reverse biased) state, when coupled to the inductance of the transformer secondary, can produce some pretty wild ringing (like 100V of it in this case based on a quick simulation). I have a lot more to learn about it, but thanks very much for setting me on the right track :) \$\endgroup\$ – scanny Dec 9 '15 at 13:38

The cap across the secondary has basicley the same function as 4 caps one across each diode.The traditional reason is for EMC or "PARD ' from the diodes.In the old days when I was thin and had thick hair AM radio was common just like mains transformers.If capacitance wasnt present you would get a terrible buzz when a station was tuned in .This was at the time referred to as modulation hum.Also you would sometimes get a terrible speaker popping turn off plop when you switched the thing off.This was due to stored inductive energy in the power transformer resonating with parasitic capacitance and radiating into sensitive analog circuitry.The cap doesnt damp this but it lowers the resonant frequency to lessen the radiation.I always put the caps in because its just not worth it to have those hassles.On your linear PSU it could do something stupid like outputting a voltage spike at switch off .


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.