Help me understand this TNT-audio web page about snubbers

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
No two (different) amps would use the same value parts for "snubbing?"

I've been desperately following this thread w/my curious, but feeble mind, intellect, and limited experience to better understand the technical side of this snubber stuff.

The Hafler XL280 PS has a cap across its (2) RBs. Am I wrong to have believed they serve as a snubbers? After reading this thread especially the part about "it" (the snubber) requires a resistor in series w/the cap. Futher the resistor does most of the work.

I've read & re-read the PS Mod article in the link below. One detail that caught my eye as (I think) it relates to this thread is the author suggests a cap across the RB to suppress spikes the diodes produce as they turn on and turn off.

Solid State Power Amplifier Supply Part 1

I want to mod the PS of another amp w/much of the usual many do. Of particular interest is separate L/R RBs & capacitor banks - right now both channels share them.

If I'm in the wrong pew, though the right church, my apologies.

thnx for reading this far, Tony
 
Member
Joined 2011
Paid Member
A single capacitor across the transformer secondary (the input terminals of the Rectifier Bridge) lowers the ringing frequency of the resonant circuit, but does very little to increase the damping factor zeta. So it is a very poor "snubber" of unwanted oscillatory ringing.

Four capacitors, placed across the four diodes in a Rectifier Bridge, are equally ineffective.

Morgan Jones (author of the much-loved textbook Valve Amplifiers) wrote an article about this in Linear Audio magazine, volume 5, entitled Rectifier Snubbing - Background and Best Practices. He found that capacitors placed across rectifier diodes, are very poor suppressors of ElectroMagnetic Interference. Very poor snubbers. To effectively suppress EMI, you need to introduce a parallel resistance. It is this resistance which damps the resonant circuit and eliminates oscillatory ringing. Capacitors don't do the job. You need resistance to achieve damping.

Unfortunately, the required value of resistance to get excellent snubbing (damping factor zeta = 1) is uncomfortably low; low enough that it would dissipate unacceptable amounts of power - especially in valve circuits whose transformer secondaries carry > 200V AC. Morgan Jones's solution (and everybody else's too) is to put a capacitor Cs in series with the snubber resistor. This capacitor Cs is chosen such that its reactance is high at the AC mains frequency (50 / 60 Hz), but low at the secondary's oscillatory frequency omega_n (0.1 to 5 MHz).

Linear Audio is a print magazine whose articles are not freely available online - you pay money to buy a physical copy.

Jim Hagerman also wrote an article about snubbers, which is available online: (link). He came to the same conclusion: an effective snubber must place a resistance in parallel with the transformer secondary, to damp oscillatory ringing. He also recommends placing a capacitor Cs in series with this resistor, to limit power dissipation at 50/60 Hz. His paper includes all of the mathematical formulae that help you calculate the optimum snubber resistance Rs and the optimum series capacitance Cs, assuming that you know the transformer leakage inductance, transformer capacitance, and rectifier capacitance.

Hagerman also discusses the idea of a CRC snubber circuit (see his Figure 7), containing two capacitors and one resistor. This is also my own preferred snubber, for the four reasons I discuss in detail on pages 5-7 of the Quasimodo design note. (link) For those who find the mathematics in Hagerman's article a bit daunting, the Quasimodo "no-math" approach might be an appealing alternative.
 
Last edited:
A capacitor which lowers resonant frequency will normally automatically reduce Q too, as the transformer resistance will form a greater proportion of the total circuit impedance. A reduction in frequency will also reduce induction into nearby circuits. A resistor (i.e. full snubber) may help, but it is not always necessary to ensure critical damping. The capacitor acts by reducing the voltage spike caused by a sudden change in current. Without this you are relying on just stray capacitance to absorb stored magnetic energy.

Caps across rectifier diodes are doing a different job. They are preventing a sudden change of current caused by charge storage in the junction. They may also prevent hum modulation in any circuit which includes an RF oscillator. A snubber will not do such a good job.

As in all engineering, you need to decide exactly what problem you are solving: RF radiation from diode charge storage or ultrasonic ringing in transformers?
 
As in all engineering, you need to decide exactly what problem you are solving: RF radiation from diode charge storage or ultrasonic ringing in transformers?

How to solve both ? In the Morgan Jones article transistormarkj mentioned,
caps across the diodes actually made things worse if I rember correctly.
Would relying on the properties of soft recovery diodes (without caps across
them) and a RC snubber across transformer secondaries be a good aproach or
are there better alternatives ?
 
From a newbie: presumably this is a challenge in power supplies for most audio equipment and not just amplifiers?

Thanks for the summary and the links - while I don't follow all of the detail yet I'm reading the lot. I had asked elsewhere whether I needed to consider snubbers in my own little project now described here http://www.diyaudio.com/forums/powe...e-steadily-build-linear-12v-power-supply.html but did not get a conclusive response.

I guess that even the "no math" approach requires investment that probably makes little sense in a one-off project. :-(
 
How to solve both ? In the Morgan Jones article transistormarkj mentioned,
caps across the diodes actually made things worse if I rember correctly.
Would relying on the properties of soft recovery diodes (without caps across
them) and a RC snubber across transformer secondaries be a good aproach or
are there better alternatives ?

That has been my experience. Rather than trying to clean up RF energy, prevent it from being generated. Use soft recovery (not necessarily fast recovery, though a good soft recovery diode will be reasonably fast) rectifiers. Then there is little use in C's and R's to improve a clean output.

Transformers always always always could use a good snubbing. Whether they be small single VA pcb transformers or large B+ units, anything that is followed by a choke or rectifier needs to be snubbed.

I have further found Hagerman's article, for as often as it continues to be recommended, to be of no help whatsoever, and at best can only get you in the ballpark. Haven't tried the quasimodo jig, but I don't understand why we make things that are so simple so complex. Snubbing a transformer requires an assortment of high WVDC polypropylene capacitors, a 5k potentiometer, and a scope (I prefer digital for this application). I get the most smooth and transient-free waveforms with this method, no matter what size transformer I am dealing with. Snubbing is simple. If you don't have a scope but want to tweak your DIY equipment, my sympathies are with you.
 
Member
Joined 2011
Paid Member
Rather than trying to clean up RF energy, prevent it from being generated. Use soft recovery (not necessarily fast recovery, though a good soft recovery diode will be reasonably fast) rectifiers. Then there is little use in C's and R's to improve a clean output.
Why not do both? It's DIY.

Slay your enemy with a broadaxe, then thoroughly spray him with your napalm flamethrower too, just to make sure he's good and dead. You want him dead, dead, dead, t f dead. Your enemy is ringing. Kill him twice.

Why not
...employ a belt-and-suspenders approach: prevent transformer secondary ringing, two different ways. First, snub the secondary so it cannot possibly ring, even if stimulated. Second, use soft recovery rectifiers so the secondary cannot possibly be stimulated. Especially in a DIY piece of equipment where the cost of a snubber is completely negligible, I think it makes no sense at all to ever omit snubbers.
(Quasimodo design note p.14)
 
Last edited:
You did not understand me. All I do is snub the transformer; not the rectifiers. This is in line with quasimodo's recommendation- he is not advocating snubbing the rectifiers; just the transformer. Where I take exception to the jig is the complexity is unnecessary if you have simple tools at your disposal. Equations are not needed, and a very simple cut and try method produces excellent results.
 
Member
Joined 2011
Paid Member
Actually we agree. We both snub the transformer; we don't snub the rectifiers. I merely add a second sentence: NEXT, when it comes time to select the rectifiers, go ahead and choose soft recovery parts. Belt= snubbed transformer; suspenders= soft recovery rectifiers.

The "complexity" of Quasimodo is one transistor and two ICs; it can be slapped together on a proto-board in two hours, picture below. It worked quite acceptably. Later I built another Quasimodo, this time on a PCBoard, so I could reclaim my proto-board for other experiments, projects, and foolings-around.

A pleasant safety feature is that you can connect your transformer to Quasimodo, and connect Quasimodo to a 9 volt transistor radio battery: (example). You don't need any connection to the AC mains {or a variac} to optimize a snubber. Nor do you need equations; it's a "no-math" empirical approach that uses measurements only.

Connecting an oscilloscope to a circuit powered from a 9V battery, is considerably less dangerous (and less frightening to newcomers), than connecting a scope to a circuit powered from the AC mains. The lethal AC mains. Especially so if the transformer's secondary voltage is higher than the mains voltage, as occurs in vacuum tube circuitry.
 

Attachments

  • proto_board_QM.jpg
    proto_board_QM.jpg
    203.2 KB · Views: 563
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.