[Heathkit] SB-610 Selenium Diodes - caps across silicon diodes.

[Kenneth G. Gordon]

Garey Barrell wrote:

> The only instance where I have seen improvement by using caps to bypass 
> individual diodes is in a VERY high gain audio amplifier power supply 
> (such as in a "direct conversion" receiver).   They help with what was 
> called "tunable hum", but required more on the order of 0.1 uF across 
> each diode.

I was not talking about the "tunable hum" type of noise, but an RF "hash" 
type of noise, of almost the same type as motor commutator noise, or leaky 
power-line type noise. 

>  I use 1N4007 diodes for everything except transmitter power 
> supplies where I use 1N5408s.   The 1N4007 is a PIN structure, possibly 
> that accounts for the difference in noise generation?

That very well could be. As I said, once I determined that the source of MY 
noise was the diodes used in the power supply, I have, ever since, routinely 
bypassed each diode in every power supply I have built. Now, the first 
experiences I had of this type of noise was in the 1960s, and, as Garey 
mentions, diode development has progressed by leaps and bounds since 
then. Perhaps I need to try building a power supply WITHOUT my 
bypasses and measure the resulting noise.

> 
> Fast recovery diodes are preferred for DC/DC converters, but they are so 
> noisy they are normally well shielded and bypassed enough to take care 
> of any RF noise anyway.

True. I used to build DC-DC converters for seismic arrays too, and those 
were PARTICULARY noisy. Since we were using very low power VHF 
telemetry to pass the information to our central receiving point, elimination 
of noise was pretty important, especially since the distances involved were 
at the limits of normal range for the equipment available.

> 
> None of my design references for power supplies recommends bypassing of 
> individual diodes, and only a few suggest small (0.01 uF) caps across 
> electrolytic capacitors.

And none of MY design references mentions "switch-off transients" in HV 
power supplies either, but that doesn't mean they aren't sometimes a 
severe problem. I also discovered THOSE on my own too, subsequently 
confirmed by somewhat obscure references.

>   Most designs for RF purposes have 0.01 - 0.1 
> uF caps across output terminals and of course on transformer primaries 
> with "AC Line rated" bypasses.

Correct.

>  Small caps directly across transformer 
> secondaries are typically discouraged since a failure will usually 
> destroy the transformer.

Well, if it is properly protected by fusing or breakers, it SHOULDN'T be. But 
in any case you are right. One has to properly size any capacitors one uses 
for bypassing, both for capacitance AND for break-down voltage.

I think 110 VAC line bypass capacitors are especially rated for that service, 
and IIRC, that is around 1600 volts peak, which should give us SOME idea 
of the peak voltages occuring on power lines due to transients.

Ken W7EKB