Worthwhile CDP tweak-SNUBBERS on diodes

Dr.H

Member
2002-04-08 6:41 pm
Hi All,
This is a worthwhile but oft forgotten tweak: Bypass your power supply diodes with snubber circuits, which consist of a resistor (100 ohms) in series with a ceramic capacitor (330nF-470nF). Note that each of the four diodes must be bypassed and that the orientation of the snubber is not important.

The overall impact is too lower the noise floor, thuis allowing more detail, greater perception of subtle nuances, and more emotional involvement with the music.

If you would like to set-up an experiment to see teh results for yourself using an oscilloscope, do teh following:
1. Buiild a simple AC-DC four diode rectifier and add a post-rectification smoothing cap.
2. Set the scope to measure AC and measure between the ground leg of the cap and AC input before teh diodes.
3. Note how distorted the displayed waveform (which should be sinusoidal) is
4. Add a snubber across teh first diode. The waveform should imporve a bit
5. Adding the second snubber goves a much better waveform, 95% pure sinewave
6. Add the next snubber distorts teh waveform a bit, while the last snubber gives the cleanest 95% sinusoidal waveform

Happy listening
Ryan
 
Dr.H said:
Yes, thanks guys, I used both sources for the tweak. Pete Goudreau is the best! I used normal non-SMT components. I know that Pete used SMT's. Still the benfit is 90-95% there as judged by teh waveform on the scope.

How did you use the R and C formulas in the PDF?

The formula for R is square root of (L/C). the problem here is L is unknown. Maybe there's a way to calculate it. Also C is another unknown but can be calculated. The only problem is the formula uses R.

thanks for the help.

Jayel
 
An article in Audio Amateur (now AudioXpress) discussed this very topic in great detail. The article was exhaustive, including theory and a decent amount of math, and was intended for power amps. I guess I can dig it up given a few days.

The article started with measuring the most significan resonance which is a function of the magnetics used, the diodes themselves as well as the load (and possibly a bunch of other things).

Now, the first thing they did was to add parallell capacitors to reduce the frequency of the resonance. Only then did they target the resonances themselves using custom snubbers optimized for that particular frequency. So in short, we measure the problem, reduce it and then go after it.

I believe Pete Goudreau's application is most likely optimal in execution (with SMT's, form factor etc.) and for low power devices, but that for power amps, the suggested method mey be even more effective. My guess is that the method may even find use in Pete's application.

Petter