Best Capacitor for Bridge Diode Snubber?

[Ron AKA]

This thread suggests a capacitor for a snubber should be about 0.1 uF, and may be most effective with a resistor in series with it. One poster says a "lossy" capacitor may work the best due to the series resistance, and may not need a separate resistor. This National Semiconductor article also shows 0.1 uF but a film type.

Do you agree that a lossy ceramic would be best? If so what would be a good one for having a high loss? The cheaper the better? Or? It seems difficult to find anyone who admits to having high loss capacitors, and want to direct you to low loss types.

[Ron AKA]

Have done a little more research, and am wondering if the Class II X7R ceramic may be the best choice for this application? Seems to have a dissipation factor of around 3%, and is more stable than the Class III Y5V variety.

Ceramic Capacitor Basics

[DF96]

If you read that NatSemi article you will find that they admit that they have simply copied what they see others doing, and at least some of the extra capacitors they have sprinkled onto the circuit may be unnecessary. IMHO the article was written by people who don't actually understand PSUs in more than a superficial way so they just followed fashion.

[!]

^ define "necessary". Is it necessary to get rid of high frequency noise? No, you can build a PSU as quick, cheap and dirty as you want it, or any other combination you choose. Doing something merely because you are copying what someone else did, is not a proof that the thing someone else did was done out of lack of understanding. Did you or did you not at some point copy the idea of using a transformer to get mains voltage down to a lower voltage? That's what humans do, copy what came before them and tweak it.

Do you feel high frequency noise does not matter? If so, then you have justification not to try to filter it... or vice-versa.

Ron, we're talking about parts that cost pennies each, it seems the more conservative approach if you want the resistance, to use a resistor, though I tend to agree that since you don't need a high uF value, there is no compelling reason to use Y5V.

[Ron AKA]

!, I was planning on soldering the caps right on the bridge, and with the ones in parallel, plus across the feed, and potentially across the output, it is going to start looking like spaghetti even without the resistors. That is why I was trying to avoid the resistors. Cost, not the issue.

[!]

^ So you're doing it point to point? I could see leaving out the resistors or a lot of the caps if doing PTP for that reason, I'd tend to only do PTP with the bare minimum components needed to get a *job* done. Plenty of people will disagree with the following statement but me... I'd never do PTP on anything worth more than $30 total in parts even if I end up etching a double, even single sided board myself towards that end.

[John Luckins]

A well designed snubber network really requires two capacitors and a resistor, one of the caps in series with the resistor and the second capacitor across them both operating at a higher frequency to dissipate any ringing. If you measure the secondary inductance and know the capacitance of the diodes at switch off/on then the values can be calculated. There's a great article on it here.

http://www.hagtech.com/pdf/snubber.pdf

The tricky bit is measuring the inductance of the transformer. If you don't have a meter then a selection of capacitors and a sine wave generator can be used. The diode capacitance varies with the voltage across it but can usually be found on the data sheet. I've used this method many times and am confident in the theory though I haven't measured the improvement as I don't have the tools (though my ears don't object to the change).

John