Snubbing capacitors

Jump to Latest
G

goodguys

Member
Joined 2011
Hi Thanks for reading. I am having a problem with psu noise and am wanting to use snubbing caps to try out and see if it helps reduce all the buzzing noise i am getting.


Are there any packaged bridge rectifiers, or types of diodes, that are less vunerable to this noise, what would you recommend.

What size caps and resistors ( ohms and power handling ) would you recommend for snubbing.


I have a few packaged bridge rectifiers, how should the caps be soldered on to the lug terminals, is a diagram possible.

Thanks
 
W

wg_ski

Member
Joined 2007
If diode “noise” is occurring, it is high frequency oscillation bursts occurring at 120 (or 100) Hz rates. It results in a “buzz” sound that just gets into everything and almost impossible to filter because electrolytic caps suck at those frequencies and everything becomes a radiator. The oscillation is caused by a resonance between the transformer leakage reactance and the effective capacitance of the diode. This effective capacitance is dominated by the diode’s reverse recovery time. The longer the time constant, the lower the frequency and the stronger the oscillation (ie, less damping). This effective capacitance does depend on the load current. More I results in more C. The idea is to damp it out by introducing high frequency loss to the circuit. The use of fast recovery diodes pushes the frequency UP, where it is more easily damped by natural losses in the circuit which go up with frequency. It may go away entirely. Putting a series RC network in parallel with the transformer secondary will also add the necessary loss, if the capacitor is much larger than the diode’s capacitance so the lossy path dominates. The bigger the rectifier, the bigger the cap you need. You rarely need over 0.1 uF even for a 50 amp standard recovery bridge. A couple ohms or tens of ohms of resistance in series will be enough damping to make it inaudible. Suitable values can be calculated (you want critical damping, Q=0.5 or lower) or can just as easily be found by experiment. If the cap is “too big” it won’t hurt anything other than wasting power in the resistor due to undesired 60 Hz current in it. Too small and you can’t introduce enough damping to eliminate the oscillation (the resulting Q is too high). A capacitor with a lossy dielectric (class 2 ceramic) will often have enough ESR at these ultrasonic frequencies not to need a separate resistor. You will often just see a .1 or .01 uF Z5U ceramic or similar in vintage equipment. If you've ever wondered what that’s for this is it. Sometimes they put the snubbers across each individual diode in a bridge. And no, you never need or want some expensive “audio grade” capacitor.

Not all transformer/rectifier/load combinations will result in this oscillation and resulting buzzing noise. Layout also makes a difference. I’ve build plenty of big power amplifiers that required no snubber at all. And several pieces where it couldn’t be cured any other way. If you know what to look for and the fix it’s just no big deal, apply where needed.
 
peufeu

peufeu

Member
Joined 2001
audio.peufeu.com
If you want to know if you have a rectifier ringing problem, make a highpass filter with 10nF and 1kOhm, and use that to observe the voltage on the secondary of the transformer with a scope. The highpass will remove 50Hz AC voltage, leaving only the HF ringing visible.

There should be a load on the power supply, for example the amp itself.

Use "AC line" trigger to view the waveform, then if the HF spike and ringing are present you can set the trigger on it, then adjust the snubber to damp it.
 
Last edited:
G

goodguys

Member
Joined 2011
wg_ski said:
If diode “noise” is occurring, it is high frequency oscillation bursts occurring at 120 (or 100) Hz rates. It results in a “buzz” sound that just gets into everything and almost impossible to filter because electrolytic caps suck at those frequencies and everything becomes a radiator. The oscillation is caused by a resonance between the transformer leakage reactance and the effective capacitance of the diode. This effective capacitance is dominated by the diode’s reverse recovery time. The longer the time constant, the lower the frequency and the stronger the oscillation (ie, less damping). This effective capacitance does depend on the load current. More I results in more C. The idea is to damp it out by introducing high frequency loss to the circuit. The use of fast recovery diodes pushes the frequency UP, where it is more easily damped by natural losses in the circuit which go up with frequency. It may go away entirely. Putting a series RC network in parallel with the transformer secondary will also add the necessary loss, if the capacitor is much larger than the diode’s capacitance so the lossy path dominates. The bigger the rectifier, the bigger the cap you need. You rarely need over 0.1 uF even for a 50 amp standard recovery bridge. A couple ohms or tens of ohms of resistance in series will be enough damping to make it inaudible. Suitable values can be calculated (you want critical damping, Q=0.5 or lower) or can just as easily be found by experiment. If the cap is “too big” it won’t hurt anything other than wasting power in the resistor due to undesired 60 Hz current in it. Too small and you can’t introduce enough damping to eliminate the oscillation (the resulting Q is too high). A capacitor with a lossy dielectric (class 2 ceramic) will often have enough ESR at these ultrasonic frequencies not to need a separate resistor. You will often just see a .1 or .01 uF Z5U ceramic or similar in vintage equipment. If you've ever wondered what that’s for this is it. Sometimes they put the snubbers across each individual diode in a bridge. And no, you never need or want some expensive “audio grade” capacitor.

Not all transformer/rectifier/load combinations will result in this oscillation and resulting buzzing noise. Layout also makes a difference. I’ve build plenty of big power amplifiers that required no snubber at all. And several pieces where it couldn’t be cured any other way. If you know what to look for and the fix it’s just no big deal, apply where needed.
Click to expand...


Which fast recovery diodes do you recommend to try and reduce this 120hz noise. The transformer is 17 amp 9volts, so diodes should be higher than 17 amp.


Z5U ceramics, what value do you recommend for these.


Thanks
 
P

phase

Member
Joined 2004
The worst diode bridge with a simple snubber will perform much better then the best diode bridge on it’s own.
If you follow the Quasimodo write-up by Mark here, you will find the perfect values.

Myself, I have had great results on every piece of equipment by using a .01uf film, along with another .1uf film in series with an 18 ohm resistor. As close to the bridge as possible, across the transformer output is the most effective.

Great bang for the buck upgrade!
 
G

goodguys

Member
Joined 2011
Not sure where the buzz is coming from, am doing a process of elimination one at a time, thought i would start with the rectifier if that does not work i will move on to snub the transformer secondary later on.

Have been thinking about shottky after reading quasimodo but can't find them in higher than 17 amp rating needed. Fast recovery rectifiers like 1N4933 appear to be a good compromise.
Maybe the simple Z5U ceramic wg-ski mentioned in #3 may be a quick solution.

Have you tried any rectifier packages or diodes and got good results.

Thanks
 
Last edited:
Mark Johnson

Mark Johnson

Member
Joined 2011
Paid Member
DigiKey lists 238 different Schottky diodes which are rated 17.5A or higher, 40V or higher, in thru-hole packages: LINK-1. Why "40 volts"? Because that's comfortably above the maximum voltage from a (17 amp, 9 volt) power transformer. If you read their datasheets I think you'll discover that most high current, low reverse voltage Schottkys are designed to be used in arrays of solar panels, where efficiency is paramount.

However there is a global shortage of semiconductors at present, and huge numbers of component part-numbers are out of stock & backordered.

Of those 238 Schottky diodes, only 62 happen to be in-stock today at DigiKey: LINK-2.

Still, 62 possibilities is better than nothing.
 
G

goodguys

Member
Joined 2011
Found this over here in britain: ON Semi 600V 30A, Silicon Junction Diode, 2-Pin TO-220 FFP30S60STU | RS Components


Has soft recovery characteristics which may help, maximum forward voltage drop does seem a little high.


On farnell: https://uk.farnell.com/ixys-semiconductor/dss60-0045b/diode-schottky-60a/dp/1080069?st=schottky
Not sure if this one is soft recovery.


https://uk.farnell.com/ixys-semiconductor/dss25-0045a/diode-schottky-25a/dp/1080068?st=schottky
This may be soft recovery.


https://uk.farnell.com/on-semicondu...st-40a-250v-to-220ac-2/dp/2317431?st=schottky
May be soft recovery


https://uk.farnell.com/ixys-semicon...-rectifier-150v-30a-to/dp/3438337?st=schottky
May be soft recovery.


Which do you think will give the best performance and reduce the noise the most.


Thanks.
 
Last edited:
S

SemperFi

Member
Joined 2002
goodguys said:
Which fast recovery diodes do you recommend to try and reduce this 120hz noise. The transformer is 17 amp 9volts, so diodes should be higher than 17 amp.


Z5U ceramics, what value do you recommend for these.


Thanks
Click to expand...

Why on earth a Z5U ceramic? Use a pulse capable and more stable type. Use a snubber cap.

Just because the transformer can supply 17A doesnt mean you use a >17A diose. You're circuit's load determines the minimum current rating. The smaller the diode the smaller the parasitics and the lower the ringing you need to snub. Never over size diodes. The opposite is better.
 
peufeu

peufeu

Member
Joined 2001
audio.peufeu.com
Do you even know this is the source of the problem?

peufeu said:
If you want to know if you have a rectifier ringing problem, make a highpass filter with 10nF and 1kOhm, and use that to observe the voltage on the secondary of the transformer with a scope. The highpass will remove 50Hz AC voltage, leaving only the HF ringing visible.

There should be a load on the power supply, for example the amp itself.

Use "AC line" trigger to view the waveform, then if the HF spike and ringing are present you can set the trigger on it, then adjust the snubber to damp it.
Click to expand...

Maybe it's just bad PSRR in the amp. You got to investigate.
 
You must log in or register to reply here.
Top Bottom