Buzzing snubbing caps on B+ supply

[fwt1712]

I have finished a cathode follower based preamp and the 0.01uf 1kv ceramic snubbing caps on the HV bridge rectifier diodes are buzzing (TDK CK45-E3AD103ZYNNA, sourced from Mouser so will be legit. but can be heard at 10 ft). Touching the caps (with something insulated of course) reduces the buzz a bit so I think it is them that are noisy. However, the audio circuit is dead quiet ie no hum and no buzz and sounds very very good.

It appears that I can use film caps instead of ceramics as snubbers ie bypass each of the diodes in the fw bridge.

I am using a 230ct transformer and getting 172 rectified dc across each diode (our ac runs a little hot ie around 245v so this is as expected). I am also getting a 40 volt drop on the 3900ohm B+ dropping/smoothing resistors (part of an LC circuit) which indicates I have approximately 10mA in the HV circuit (feeding 6dj8s).

My questions is, what voltage film cap should I use, specifically do I need 1kv which is what the ceramics are specced at or will something > 172Vdc, which I measure, suffice. I don't think I will get a film cap near 1kv to fit and I don't understand if there are serious voltage spike or other issues that occur in these circuits.

If I do need this high voltage rating I suppose I could drop uF a little to fit the space. I also assume I could do away with the snubbers completely if I get completely sick of the buzz. One final thought, would a smaller ceramic be less likely to buzz - the TDKs are nearly 1/2" across (11mm) or should I do something else?

thanks to all,
Nigel.

 

[nigelwright7557]

The problem is your diodes, they suffer from switching noise.
Try HER158G R0.

 

[rayma]

Ceramics are bad for AC, so use film types rated at 500V or more (in this case).

 

[fwt1712]

The diodes used are HER108 what is the difference between them and 158s? A quick scan says more current capacity - is this what helps??

 

[nigelwright7557]

The diodes used are HER108 what is the difference between them and 158s?

I am just passing on my experience with HER158 in that I didn't need snubber's with them in my application.

I found adding snubbers actually made the noise worse as I was getting some AC on the DC.

 

[6A3sUMMER]

Take a 230V to center tap winding, that is more like 245V.
245Vrms x Root(2) = 346.5V peak.
The peak inverse voltage across the diode is 346.5V x 2 = 693V peak.
That is the peak reverse AC volts from the winding to center tap which is across the diode and cap, plus the voltage across the filter cap which is now charged to 346.5V.
PIV is 693V across the diode and across the cap.

That is why the original caps were rated at 1kV.
Just imagine 693V across the cap, and then a power mains transient occurs, the ROOM lights go bright for a moment, and then the caps short.

And have you ever heard of Ceramic Phono Cartridges, and Ceramic Microphones.
Those Ceramic effects are reversible, which makes Ceramic caps into a 'loudspeaker'. Buzz . . .

 

[wg_ski]

I’ve even heard FILM caps exhibit piezoelectric sounds. The output coupling cap in a switching supply very often does - it’s just too high in frequency to hear. Try building one with a 15kHz switching frequency, and it will drive you from the room. Ceramic caps can be far worse with mechanical noise, especially when operated near their voltage rating. Why? Because the dielectric properties change with voltage - and the higher you run it the closer to the top end of the curve it gets. At rated V, you probably have only 10% of it’s rated capacitance “left” so it may not be doing much. All class 2 ceramics do this, which is why they are terrible in the audio path. Get a 3, 5 or 6kV ant it will quiet down substantially. Also, snubbers need a RESISTOR in series with them to dissipate the stored energy in the LC tank, in order to actually damp the ringing. Also to act as a FUSE in case the cap shorts. This reduces the peak current, and with it, the mechanical noise.

 

[jackinnj]

Do you really need snubbers? If you have a scope look at the rectifier output and see if there is ringing present -- ringing is a function of the diode junction capacitance and leakage inductance of the transformer. Take a look at the "Quasimodo" thread for more info.

HER108 junction capacitance is only a few tens of pF's at those voltages. Do you really need snubber caps?

Just out of curiousity are you using a toroid or EI-core transformer?

I’ve even heard FILM caps exhibit piezoelectric sounds. The output coupling cap in a switching supply very often does - it’s just too high in frequency to hear. Try building one with a 15kHz switching frequency, and it will drive you from the room. .

Menno van der Veen wrote a paper for the Journal ofthe Audio Engineering Society on this topic: "Non Linear Distortions in Capacitors", AES 124th Convention working paper, May 2008.

 

[wg_ski]

If you push the ringing frequency up high enough it may not bother anything. But it just as easily could - if it mixes with other stay HF/UHF from DAC clocks, PWM carrier from switching supply or class D or some CPU running. Just the amp itself could sound quite clean, but go to $ with digital sources. If you do have something weird going on that you can’t otherwise explain, it’s time to scope the power supply with something that can read up to a GHz. An old 20 MHz scope will often show absolutely nothing.

Ringing amplitudes across the diode bridge still need to be kept in check, if nothing else but to keep peak stresses low. If it’s only a few volts or so it won’t blow diodes. But in something like a switching supply it can easily hit 2X without one, which is why they are always used.

 

[kodabmx]

I've never needed "snubbers". If you think you need them, use X2 caps because they are cheap and work perfectly.

I also use them for coupling the audio stages in the amp.

20PCS X2 Safety Capacitance 275v 0.01UF 103k 10NF, Foot distance: 10MM | eBay

 

[Mark Tillotson]

I’ve even heard FILM caps exhibit piezoelectric sounds.

They may make sounds, but its technically not piezo electric, its due to simple electrostatic forces across the dielectric. In a piezo material (aka ferroelectric), differently charged atoms move in a concerted fashion within the lattice coupled to the electric field, distorting the lattice from within. This is often much more distortion than the electro-static forces could produce.

 

[fwt1712]

I am using a triad EI transformer (vps230-190), which has 0.190A capacity so should be pretty under-stressed driving 2 x 6dj8 tubes.

I don't have access to a scope to look at any noise artifacts but may just try and see what happens if I just remove the snubbers first.

 

[Brice]

Yeah I had this problem before.
Caps around a diode is not sufficient. And Ceramic caps buzz like mad.
You need a RC network, not just a C.
You can buy a real snubber component, which as a RC build in, or do it externally.
I've just done a PS for another project and I used across the bridge a RC network to kill the bridge noise. I used 100R / 2.2nF. My PS is dead silent and clean on the scope.

 

[trobbins]

That ceramic capacitor datasheet only appears to provide a DCV spec of 1kV - there is no ACV spec, so you appear to have poorly chosen the type of cap used.

I suggest you and many others have seen caps across rectifier diodes and are just naively following on a tradition that is not technically substantiated. In days of yore, with large industrial diodes, there can be a need for RC snubbering due to associated transformer reactances, large diode forward currents and long reverse recovery times. Early days of meeting EMC requirements also seem to have been some incentive for using caps in some equipment. Nowadays you have available fast diodes like the ubiquitous UF4007 that imho meets nearly all valve amp needs.

If you constrain the loop area of the rectifier current pulses, and make appropriate connections to filter caps, and then out to circuitry, you should have no need for any additional circuitry. Your power supply requirements appear to be benign as you don't have high load currents, and hopefully aren't using humungous levels of filter capacitance.

You may want to sketch out your power supply connections and post, as its not uncommon to see ill-advised grounding connections and schemes.

 

[baudouin0]

Ceramics are piezo electric that is a voltage can cause them to distort slightly. The high frequencies generated by the switching diodes can cause them to generate noise. You could try a different type/lower value or use film. I put a couple in a mic amp with 48v across them, You could hear people speak.

 

[fwt1712]

Snubbing caps have been removed and all extraneous mechanical noise is gone - diodes etc. are completely silent. There doesn't seem to be any difference in the quality of audio output and it remains absolutely free of hum or any other 'nasties'. Thanks everyone for your help and advice.

 

[baudouin0]

They will stop mains buzz on long wave radio but that's about it.