Capacitors across rectifiers;why?

[DaveP]

I guess the formulae for snubbers  applies just as well to damping the ringing on input transformers.

However its a lot of maths when you can simply use a few hundred pF and a 50k pot and watch a square wave on the scope.

Or just use the right terminating resistor. (snubber?)

DaveP

 

[trobbins]

I am not sure why  you should stop at critical damping. Why not reduce the resistance further and make it over damped?

The energy in the transformer winding leakage inductance needs to change from one level to another due to the diode commutation.  The more of that energy bypassed by the snubber, compared to allowing it to transfer to other circuit paths (eg. by parasitic capacitance and circuit dV/dt) is obviously of benefit.  However, there is a mains frequency (plus harmonics) current through the snubber.  So for a certain C, the snubber R needs to compromise both those issues.

 

[ruffrecords]

I am not sure why  you should stop at critical damping. Why not reduce the resistance further and make it over damped?

The energy in the transformer winding leakage inductance needs to change from one level to another due to the diode commutation.  The more of that energy bypassed by the snubber, compared to allowing it to transfer to other circuit paths (eg. by parasitic capacitance and circuit dV/dt) is obviously of benefit.  However, there is a mains frequency (plus harmonics) current through the snubber.  So for a certain C, the snubber R needs to compromise both those issues.

OK, I can understand that, but why should the compromise be at the critical damping point?

Cheers

Ian

 

[trobbins]

Ahh well, to each their own compromise.  From a practical perspective, its easy to tune to 'about' optimum.  From a time perspective, the energy is dissipated in the shortest time.  And some just fixate on what they perceive is an optimum because the maths has focussed on that point.

 

[ruffrecords]

Ahh well, to each their own compromise.  From a practical perspective, its easy to tune to 'about' optimum.  From a time perspective, the energy is dissipated in the shortest time.  And some just fixate on what they perceive is an optimum because the maths has focussed on that point.

OK, so basically there is no real reason to aim at critical damping, just so long as it is not under damped. Given how hard it is to determine the correct value. Also, given the certainly no-zero secondary winding resistance for HT supplies, you will probably not be too far away simply forgetting the series resistor.

Cheers

ian

 

[Whoops]

Hello everyone,
I've been following this thread and I've read  the Snubber article posted.

I'm trying to understand this, but be patient because I'm not as knowledgeable as any of you.

What are the practical implications of not using a snubber?
How do you know if you need one or not?

To damp HF noise from diodes turning on/off and perhaps from mains.
JR

The reason I use it is that in the early days I did not have one and I got this very short 100Hz spike sitting on the top of my HT supply. The capacitor suppresses the small spike you get whenever the diodes turn off.
Ian

In the late 70 Rectifiers became much faster and the AC Line harmonics became a problem that needed to be controlled. Today the IN4xxx diodes and similar are very fast devices.  As Ian is saying now all the rectifiers need some type of snubber.
Duke

So how do this HF noise, Spike on HT supply, Line Harmonics manifest itself in an audio circuit?
Imagine I build a mic preamp or and equalizer how would the audio circuit be affected ?

thanks

 

[trobbins]

  given the certainly no-zero secondary winding resistance for HT supplies, you will probably not be too far away simply forgetting the series resistor.

Not just the dc resistance, but also the distributed capacitance.  Yes a few have commented that HT supplies for valves are 'no issue', especially when a valve diode is used, or a  UF4007 instead of a 1N4007, and filter capacitance is kept to  a normal rather than insane level.

 

[rackmonkey]

Depending on placement of the caps and the values, you don't create a short but rather a voltage quadrupler.  Do you have a drawing?

BT

 

[ruffrecords]

Correct. A cap by itself only shifts the ringing to a lower frequency (where it may be less of a problem). A suitable snubbing network is usually a 10nF cap in series with a 1k to 5k resistor. You can of course tune the exact resistor value if you have the equipment and patience.

To be honest I have never quite understood this. The articles on RC snubbers I have seen all seem to be aiming for a critically damped response. I have no idea why; what you want if anything is a grossly overdamped response. You just want to knock that spike on the head hard.

Cheers

Ian

 

[benb]

To be honest I have never quite understood this. The articles on RC snubbers I have seen all seem to be aiming for a critically damped response. I have no idea why; what you want if anything is a grossly overdamped response. You just want to knock that spike on the head hard.

Cheers

Ian

Thinking through this, the more you cut down on the voltage spike, the bigger the current spike through the snubber circuit will be. This current spike can radiate much like the voltage spike.

You also want to keep the diode bridge circuit area (and the snubbers connected to it) small, and keep the wires between the transformer and the bridge close together (better, twisted together) so they don't radiate as much. This helps reduce the radiation of both any left-over RF and 60Hz.

 

[trobbins]

Imho, it is best to snub any secondary winding leakage inductance using a C-RC snubber directly across the winding - snubbing at the source is always the best method.  Use the bell-ringer style jig to make it easy to tune the C-RC snubber.  Remove any and all caps placed in parallel with each diode - they act a path for transient current to more easily loop through (where you don't want it to flow).

 

[ruffrecords]

Thinking through this, the more you cut down on the voltage spike, the bigger the current spike through the snubber circuit will be. This current spike can radiate much like the voltage spike.

I believe this in not correct. Here is a link to the classic rectifier snubber paper.

http://www.hagtech.com/pdf/snubber.pdf

This makes it clear that the resonance is caused by the leakage inductance and interwinding capacitance of the mains transformer (fifth page). Snubbing tries to reduce this resonance by adding a resistor in parallel with the transformer winding and hence the LC circuit in order to dampen it. You cannot use a small enough resistor directly across the winding for power dissipation reasons so you put it in series with a capacitor. This capacitor needs to be large enough to be a very low impedance  at the resonance. The value of the resistor is chosen to give a damping factor of 0.5. The article says any higher and the resonance goes on for too long and any smaller and the resistor dissipation is too high.

My problem with this is you still get at least a whole cycle of ringing at full amplitude. If instead, you set the resistor value to zero, you really dampen the spike - the resistance value is now just the resistance of the wiring to the transformer plus its secondary DCR.

As the author says on the last page:

Of course all this theory may be hog wash. The real test is how these snubbers perform in real circuits.

All I can say is I have found that a 100nF 275VAC capacitor right across the input to the bridge rectifier knocks these spikes stone dead.  Attached is a picture I took of the spike when I first saw it on an HT supply back in 2007. Since adding caps directly across the bridge input I have never seen one again.

Cheers

Ian

Edit: For some strange reason the jpg image fails the security checks??? But the same image saved as a .png passes???

 

[Tubetec]

On a couple of vintage marshalls I have , the center tapped full wave rectified ht winding originally came with two .1uf 1000 volt dc caps from both phases of the winding to ground , they seemed to be prone to failing short though. I had always assumed it was to do with reducing any incoming harmonics from the mains.  They seem to work fine without them in any case , most tube amps Ive come across dont seem to use them . Ive also seen some people recomending  bypassing each diode in a bridge configuration with its own cap , what would be the pro's and cons of this setup compared to  the single cap method described by Ian earlier in this post ?

 

[trobbins]

the center tapped full wave rectified ht winding originally came with two .1uf 1000 volt dc caps from both phases of the winding to ground , they seemed to be prone to failing short though.

Not all caps are created with equal specs.  Placing a cap across fairly high continuous VAC requires it to be rated for that application - so inspection of the cap datasheet is necessary, and a 1000VDC marking on the cap may not provide sufficient evidence.  Also, it would be wise to choose an X rated cap, as that is then designed for such continuous VAC application.

The hassle with Ian's method is that it is suck-it-and-see, and not all power transformers are used for the same application or have the same secondary side windings.  Many of us would check hum level in an amp, and turn a hum pot if there was one for minimum hum, or tube roll if they could detect a difference.  Not everyone wants to pursue noise emission levels, but some do,  and so a bell-ringer is of assistance.  Transformer secondary windings used for tube B+ typically have much larger distributed C and series R, which typically self-absorbs the disturbance from leakage inductance, especially where the rectifier diode has a relatively soft turn-off characteristic, such as with a valve diode.

 

[Tubetec]

Thanks Trobs,
As you would expect with amps this old(40 years+) ,new filter caps all along the ht rail and replacement bias smoothing caps were the first step in bringing spurious noise under control ,pulling the grid wires up and away from chassis also reduced the induced noise a good bit.

 

[trobbins]

pulling the grid wires up and away from chassis also reduced the induced noise a good bit.

That's an interesting comment, as usually running grid wires  next to chassis allows them to be further away from other signal sources (and so reducing coupling from stray capacitance).  So that suggests either there was some stray coupling that was beneficial (ie. neutralising hum by adding a selective hum signal back in), or the chassis grounding scheme was not so good (typically due to how a power transformer secondary CT connection is taken direct to chassis, rather than direct to the input filter capacitor negative terminal).
Ciao, Tim

 

[CJ]

RC networks are serious bizness on commercial HV supplies,

here are a couple of stacks, spiral networks are big $$$ but good to maybe 100 KV if you submerge them in an oil tank,

top one uses Glastic (red stuff), a combo of fiberglass and plastic, and Phenolic spacers over the bolts (black)
1/8"  Glastic good for 30 KV  IIRC.

cutting Glastic strips on the table saw was not fun. scratch and itch, luckily our lungs were coated with so much weed resin that it didn't matter if we wore a mask or not,  ;D

bottom stack probably worth about $1000, , solder job has to be precise, no sharp edges allowed,

 

[CJ]

close up>

 

[Tubetec]

Cheers Tim,
I think the grid wires passed close to the heater chain ,changing them so they came down from the socket vertically  and across to the tag board ,puts a couple of inches  distance between them . I do get what your saying about how sometimes an induced noise can subtract from,or cancel a hum induced earlier in the circuit though.

Hey Cj,
High voltage rock n roll eh, funny how health and safety in the workplace has gone from non existant to so strict that certain types of electronic work  simply arent covered by insurance anymore here,and we have to rely on far away places with very poor working conditions for workers to get stuff done.

Nice to see the change  in relation to the beneficial herbs sweeping across the States, were still a distance behind on that score over in the emerald isle, medicinal  is on the way ,but recreational might be a while off yet.
I see Willy's getting in on the act now too ,
https://williesreserve.com/products/
Mmmmmmm budulicious 

 

[trobbins]

Great examples of low current series stacks of diodes.  Modern products would use specialist avalanche rated diodes.  Many such diode stacks have to work in applications that switch rapidly from forward conduction to reverse withstand, and so the reverse recovery effect and the dynamic induced voltage across diode capacitance becomes significant, and is managed by swamping inherent part junction capacitance be an external capacitor.

For us mere mortals trying to determine when the 1N4007 is getting beyond an acceptable operating PIV, the use of 2-3 series 1N4007 is normally quite acceptable to do without adding balancing R's and Cs, but should really be done with diodes from the same batch, and with the diode bodies connected by a short lead wire (to keep Tj the same).  I normally transfer to two series 1N4007/UF4007 when above 280-0-280Vac secondaries in common full-wave rectifier.  The rate of change of V and I across the diodes in that mains frequency application is at a snails pace compared to when diodes are exposed to forced commutation/smps type applications.