Diode bypass caps

[JC]

You may wish to do some search-engine research on the concept of "conduction angle" of diodes in the power supply in question.  You may find that the diodes are only in a forward-biased condition for a relatively small portion of the AC waveform supplied by the transformer.  I'm not sure that small caps across the diodes could effect that significantly.

That being said, it certainly seems like a worthwhile experiment if it is not too physically challenging to tack them into place.

[Paul Joppa]

It's a fascinating subject alright. I did a little googling, hoping to find a straightforward description of real-world diodes and the kinds of capacitive and/or resistive snubbers that might be employed in parallel or series with them. What I found was mostly confusing, contradictory, and/or irrelevant. I have not found a good simple model with which to explore the details.

Certainly slamming on and off the current through the transformer windings will produce rapid changes in the magnetic forces, which will produce mechanical force on the iron parts. It seems reasonable that an impedance element - capacitor , resistor, and/or inductor - could slow the transitions. If the impedance is small enough to not have much effect on the performance, then its effect on transient energy is probably confined to radio frequencies, or at the lowest the transformer resonant ringing which is typically in the 50kHz-100kHz range. But of course that's just speculation on my part.

For what it's worth, the intrinsic capacitance of a UF4007 diode (as used in the Paramour) is on the order of 10pF, so I would not expect an additional 10pF to make a substantial difference. Snubbers are usually more like 10nF (0.010uF) I think.

Here's an off-the-wall point: In some cases, the power line voltage is asymmetrical and may either have DC riding on it, or a DC current can be generated in the power supply because one diode does more of the work than the other. Any DC current through the transformer can lead to a substantial increase in vibration. Various schemes have been proposed to resolve this problem if it exists, but - again - Google leads to a confusing mass of unsubstantiated speculation along with a number of interesting things to try.

[JC]

I have only been able to observe power line asymmetry once, and, interestingly, it was in a 3-Phase commercial power situation.  There were also a number of potentially very reactive large loads in the form of magnetic "ballasts" for large HMI lighting instruments.  By "large", I mean in the 4 kW - 12 kW range, per instrument.  I was never able to observe that particular power system after those loads were removed, so I wasn't able to form a theory, nor did I think to check for DC at the time.

I have certainly seen many examples of caps across SS diodes over the years, but I don't recall ever seeing a good explanation about how they improved the power supply; only that they reduced the RRS.  It almost seems as if one must take it as an article of faith, and just do it, since it can be demonstrated to work.  I have sat around speculating many a time about why it works, but every hypothesis seems to always wind up having some serious objections that can never seem to be overcome.

I have decided that no one knows!