Why small caps across diodes?

[Axtman]

What is the reason for having small capacitors across the bridge rectifier diodes?

[Gingertube]

Supression of diode switching noise spikes.
Caps not required if using Ultrafast Soft Recovery diodes or Schottky diodes for the bridge.
Cheers,
Ian

[Axtman]

Thanks Gingertube. Would 1N4007 diodes be considered ultra-fast soft recovery?

[Gingertube]

No - UF4007 is the Ultra Fast Soft Recovery version.
Cheers,
Ian

[mozwell]

Often the caps are added so the complete unit can pass EMC compliance certification. In Au its mandatory for electrical equipment, and often with linear amps, all you need to do to get compliance is to add these caps across the bridge rectifier diodes.

[Chuck H]

Supression of diode switching noise spikes.
Caps not required if using Ultrafast Soft Recovery diodes or Schottky diodes for the bridge.
Cheers,
Ian

Ian, you may be the guy to ask about this.?. What about WRT clipping diodes like a Zener? Is there switching artifact noise present in the audio for something like a clipping circuit that could be mitigated by small value bypass caps?

[Gingertube]

Chuck,
The power diode switching noise spikes come from a splat of reverse current to sweep out the junction capacitance charge before the diode actually really switches off. This happens as the voltage across the diode reverses. The size of the current spalt is directlty proportional to the junction capacitance. UF diodes typically have 1/5th the junction capacitance of a standard diode (so the switching noise spike sould be 1/5th too). Schottky Diodes have 1/10th or less the junction capacitance of a standard diode.

Clipping diodes are something I'm having trouble visualizing - thinking guidance below.

1st they are normally wired as a back to back pair.
2nd they are normally small signal diodes with very low junction capacitances.
So I think that for standard diode clip circuit there will be minimal if any noise spike problem.
You also sometimes see RF bipolar transistors used (wired as diodes) as the clip devices because (I guess) they have exceptionally low junction capacitance.

Zeners are another "kettle of fish".
They are a reverse bias thingy - they conduct in avalanche mode when you exceed their reverse voltage spec (their zener voltage). With forward bias they act just like a standard diode with the same 0.7V voltage drop. They also have high junction capacitance. They also tend to have some avalanche noise all the time (more as the voltage approaches their zener voltage). That is why you see a cap across them when used as voltage references in voltage regulator circuits and the like. Offsetting this is that they are normally used at a circuit point where the signal voltage is much higher so any switching noise is less significant.

My thinking at the moment is that the back to back diodes you see in clipping circuits will NOT prevent switching noise. The "companion diode" of the clipping pair will be forward biased with respect to that reverse current turn off spike.

I'm also thinking that choosing the circuit point to apply clipping such that the signal levels are suitable for clipping by standard small signal diodes (or bipolar RF transistors wired as diodes) or best of all Schottky diodes would be preferable to zener clippers.

The other offsetting consideration for all the above clip circuits is that they work with a series resistance (in series with the back to back diodes) so you get soft clipping. That series resistance is going to limit the reverse current splat to probably insignificant levels which limits the noise spike.

Sorry I can't give a definitive answer but above is my best "educated" guesses.

Is there a Solid State physics guru in the house?

Cheers,
Ian

[Chuck H]

That's actually a lot of info for my purposes. You may know from my history here that I use Zeners for the "Paul Ruby mod" on one of my designs. I've always noticed a small, buzzy artifact that follows the duty cycle shift of the PI under clipping conditions and I've suspected the Zeners because there's no visible crossover (which typically causes that sort of thing). The next time I get one of these on the bench I think I'll try some small value caps across the Zeners and see what happens No great shakes if it doesn't do the trick. The noise is tiny enough that I can (almost) let it go even with my OCD

[loudthud]

Here's an old thread on the subject of rectifier noise:

http://music-electronics-forum.com/t34483/

[SoulFetish]

That's actually a lot of info for my purposes. You may know from my history here that I use Zeners for the "Paul Ruby mod" on one of my designs. I've always noticed a small, buzzy artifact that follows the duty cycle shift of the PI under clipping conditions and I've suspected the Zeners because there's no visible crossover (which typically causes that sort of thing). The next time I get one of these on the bench I think I'll try some small value caps across the Zeners and see what happens No great shakes if it doesn't do the trick. The noise is tiny enough that I can (almost) let it go even with my OCD

Chuck, you don’t elevate your heater supply voltage by chance, do you?

[Chuck H]

Good call, but not on this one.

[SoulFetish]

Good call, but not on this one.

just checkin'

[Merlinb]

Is there switching artifact noise present in the audio for something like a clipping circuit that could be mitigated by small value bypass caps?

Nor really. The spikes in a power supply are principally due to transformer leakage inductance, which you don't have in a clipping circuit.

[Chuck H]

Thank you Merlin, and Ian and LT. Some more scoping has revealed almost certainly that the artifact I'm hearing is ringing due to a minor oscillation issue. I may or may not trouble to hunt it down since this amp has a certain "something" (mojo?) that later examples with a different layout do not.