Tube amp mechanical noises - ceramic caps in rectifier, etc

[pula58]

I just tried a new solid state rectifier setup.

Instead of the 6 1n4007's like you'd find in a Fender Twin reverb (for example) I tried 4 UF4007's with 10nF ceramic caps across each one (to encourage equal reverse-bias voltage sharing).
Man, I can hear those 10nF ceramic caps buzzing away at 60Hz! I think they are phsically vibrating!

I have also noticed that if you drive the grid of a preamp tube or output tube into conduction that you can hear that too (not coming out of the speaker, but rather, as a mechanical noise/vibration).

Anyhow, would a different capacitor type (10nF, 100V) that would not have this audible 60Hz vibration, cause this is a noise I don't want my amps to have!

[VacuumVoodoo]

Snubber caps on UFxxxx diodes are a no-no - get rid of them. They can cause worse buzz than any 1Nxxxx diode.
BTW that buzz is 120Hz, not 60.

[pula58]

Snubber caps on UFxxxx diodes are a no-no - get rid of them. They can cause worse buzz than any 1Nxxxx diode.
BTW that buzz is 120Hz, not 60.

Are you talking about buzz that I'd hear at the output, through the speaker, caused by the caps. Or, the mechanical type of noise I am describing?

Would two 1000V diodes be safe in a power supply with center tapped HV, full wave recitifed, 485V HV while in standby.

Why do you suppose Fender used three diodes, the 1N4007's are 100V rated.

[VacuumVoodoo]

1N4007 are rated 1000V. In a full wave rectifier the max AC voltage from PT HV that you can safely use with a single 1N4007 per leg is 350-0-350V, because when polarized in reverse the diode will see full rectified voltage 350x1.4 on anode and on cathode the negative peak voltage on it's feeding winding another 350x1.4. It's just under 1000V. So to play safe they used 2 diodes in series to get 2000V voltage withstand. Now, one of those diodes can get zapped and short out, this in turn would bring the 2nd diode down with a hard wired short on one half of HV winding. Puff goes the PT. Put a 3rd in diode in there. If one goes bad you still have 2 giving you 2000V combined voltage withstand.

Yes, ceramic caps can produce mechanic buzz as well as screw up UF diodes reverse recovery and generate transient voltage spikes in 100s of Volts at 120Hz. These leak all over the circuit and get amplified. Watch for wasps and bees swarming and trying to get inside your house.
If you have a ceramic disc cap in tone stacks treble filter try scratching it gently with a small screwdriver or such. It will come out through the speaker.

[Structo]

Alek, are you saying that whenever you use Ultra Fast diodes in a power supply that you should not bypass each diode with a .01uF ceramics?

I used these thinking they would be quieter.

Hmmmmm.....

Mine has a bit of buzz which I always thought was the PT vibrating.....

Dang, just when you think you know somethin about somethin.

http://www.mouser.com/Search/ProductDet ... 11-STTH110

http://www.st.com/internet/com/TECHNICA ... 003261.pdf

[pula58]

Dang, just when you think you know somethin about somethin.

Yah, me too!

[VacuumVoodoo]

Alek, are you saying that whenever you use Ultra Fast diodes in a power supply that you should not bypass each diode with a .01uF ceramics?

I never found any use for them in linear power supplies for audio. I used ultrafast diodes where they are needed i.e. SMPS running at hundreds of kHz switching frequency.
50/60Hz is too slow to benefit from the fast switching while switching with inductive source feeding it is an invitation to buzz salsa. Put a snubber cap across the diode and you have an LC oscillator. Even parasitic capacitance of wiring can be enough.
I generally use full bridge rectifiers, no CT with SS rectification. It gives me advantage of not having to worry about 1000+ volt reverse voltage across diodes.
Even with standard 1N4007 diodes a simple capacitor snubber can do more harm than good. A couple kOhm resistor in series with each snubber cap is often an efficient wasp extermination procedure. Resistor value usually has to be determined empirically. An oscilloscope is a must.

[Firestorm]

Maybe look at the Fender 135 power supply: fwb with diodes and caps, but the diodes are essentially 1N4007s, not UFs. These amps are all ultralinear so no choke and thus have the added complication of a resistive voltage stabiliser, which you can ignore if you want a choke or even a B+ to Bscr resistor. I've always found this power supply to be dead quiet.

[Structo]

Is that the Twin Reverb 135?

I notice it has .002uF snubbers.

I may have to revisit my power supply.

[Cygnus X1]

Are 5408's any safer to use?

[Structo]

The 5408 is a standard recovery diode.

But, it is a 3A 1000V diode!

I don't know what the equivalent diode is for the ones I used but I thought
we had discussed the UF diodes a while back and the consensus was
that they were good to use in guitar amps.

Those 1N4007's are so small I never really trusted them.