# Unusual filtering scheme question

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.

#### ITPhoenix

I found this decent commentary on solid state amplifier power supplies, but the final 680nF capacitor has a 1 ohm, 17w resistor in series with it.

I understand the liberal reservoir capacity he favors, but I have never seen this before, and why such high wattage resistor when the ripple should be miniscule and only 120Hz.?

Solid State Power Amplifier Supply Part 3

#### jan.didden

Paid Member
I understand the use of a zobel to damp hf noise and junk, but 17W indeed seems overkill.
Also his statement "... to get rid of the capacitor inductance" is incorrect - he merely puts a resistor in series with the capcitor inductance. It would still be advised to use a low-inductance capacitor for best filtering results.

jan

#### ITPhoenix

Thanks jan. Are the "orange drops" considered low inductance--for this purpose?

And yes, I took his comment as this filter section taking the inductance out of the main cap bank.

Last edited:

#### Mark Johnson

Paid Member
Also his statement "... to get rid of the capacitor inductance" is incorrect - he merely puts a resistor in series with the capcitor inductance.

Jan, I suspect that he is referring to the inductance of the massive 20,000 uF capacitors upstream. I suspect the (1 ohm + 680 nF) zobel network is an attempt to build a snubber that forms the "R" part of a damped RLC circuit; where L is the inductance of the 20,000 uF filter caps (and their wiring!), and C is the capacitance of those same 20,000 uF caps. I suspect that he calculates the "characteristic impedance" of the LC circuit Z = sqrt(L/C) and finds this to be around 1 ohm.

#### DF96

He may be trying to damp the parallel resonance he has created by bypassing the big caps. I note that his circuit description is long on words and short on numbers. This can sometimes be a warning sign that no actual design has taken place, merely an assembling of 'rules of thumb' or personal preferences.

#### bicefalo

Hi ITPhoenix

I usually put a zobel/snubber on transformer secondary side too ( before diods bridge )
And yes, TNT values look on the hight side....may 330nf+1R/5w will damp well ringing.

What do you mean for "orange drops" ? tantalum ?

Ciao

Marco

#### ITPhoenix

Sprague polypropelenes are coined "orange drops" due to their color and shape. Prefered my musicians in guitars and amps for their supposedly better tone and lower noise. They are slightly more expensive than average films.

In the diagram it appears there is a transformer primary snubber. 220nF in series with possibly 9R1. Too fuzzy; cannot verify resistor value.

Last edited:

#### jan.didden

Paid Member
Jan, I suspect that he is referring to the inductance of the massive 20,000 uF capacitors upstream. I suspect the (1 ohm + 680 nF) zobel network is an attempt to build a snubber that forms the "R" part of a damped RLC circuit; where L is the inductance of the 20,000 uF filter caps (and their wiring!), and C is the capacitance of those same 20,000 uF caps. I suspect that he calculates the "characteristic impedance" of the LC circuit Z = sqrt(L/C) and finds this to be around 1 ohm.

I suspect not
There's also the non-ideal coupling between the primary and secondary which will make all this not very sensible. You'd have at least the leakage inductance between the prim and sec which probably is much higher that the cap inductance in the first place.

jan

#### jan.didden

Paid Member
He may be trying to damp the parallel resonance he has created by bypassing the big caps. I note that his circuit description is long on words and short on numbers. This can sometimes be a warning sign that no actual design has taken place, merely an assembling of 'rules of thumb' or personal preferences.

... or 'everyone knows' as found online

jan

#### jan.didden

Paid Member
Hi ITPhoenix

I usually put a zobel/snubber on transformer secondary side too ( before diods bridge )

This is very sensible!

jan

#### Elvee

Thanks jan. Are the "orange drops" considered low inductance--for this purpose
The 17W rating of the resistor means it is a wirewound type, making the type of capacitor, its inductivity or absence thereof irrelevant (and to generalize, the simple presence of a cap in this place)

#### ITPhoenix

Well, we were doing good there for a minute. This is for Rod Elliot's P27A, 100w, fully solid state guitar amp. I am using tubes for the pre, but will experiment with his power section. All I wanted to do was take it up a notch in performance. I am using a torroid with two 3,300uF in parallel. His minimum is 47,000 total for each rail.

I like the 100uF idea on the amp section since it may be of considerable distance from the power board. I would use at least 18 gauge wires to connect.

Since there are many unkowns in terms of transformer and capacitor selection specs, is there a usual method of determining the proper value of these components, or trial and error, watching the scope while in operation?

Last edited:

#### DF96

Calculation. See the thread on reservoir size in the power supplies forum.

#### ITPhoenix

Thank you everyone for the answers. This is more involved than I thought. DF96, I see that thread is quite long. You even have a commentary on page 1, which serves as sort of a pointer to the calcs.

#### bicefalo

Ola' ITPhoenix
I think "orange drops" are not low inductance cap (no data in Vishay/Sprague tech sheet)
they are better suited in signal path if you like their sound.
I prefer Epcos MKP , Vishay MKP or Arcotronics Nissei Group R73 (better option).

Elvee
I am agree with you about wirewond resistor, for this purpose I am using thik film type ( like RS#239-2503 ) for better damping.

and a couple of tips for ITPhoenx : don't forget to twist the tranformer cables, two turn for inch for example ( important ) and place the Zobel/snubbers very close to diods ( more important...) to help stability on rails.

ciao
Marco

#### ITPhoenix

Muchas gracias, Senor Bicefalo!

The writer recommends "film type" caps for these Zobel snubbers. It is true the calling for 17w begs wirewounds. I saw the film resistors in Mouser, and also "audio" wirewounds in the Ohmite catalog. http://www.ohmite.com/cat/res_audiogold.pdf

So for an hour or so I read the thread DF96 recommendeded for the reservior calculations. I actually meant calculations for these Zobel networks, but while these are for ripple and noise improvement, it appears total capacity has an effect on the actual sound, according to the myriad comments. My point here being, I am not sure if the Zobel values will apply if the total capacitance changes, and if it matters if toroid or EI effects the proper values, in addition to their inductances.

I am now tempted to go with a dual LM338 or LT1085s off separate windings and rectifiers to yield the rails. Practically no one says regulation of a SS power section benefits much from regulated, but in terms of PSRR and noise, it should work IF, and only IF, designed properly, as I understand.

I wind my own trannys so the outputs are unlimited. Oh, and thanks for the twisted AC wiring tip. I may have forgotten.... There are preamp tubes present.

Last edited:

#### bicefalo

Hi ITPoenix

Ohmite Audio Gold Resistors are not suited in snubber app.
Better option is Caddock MP930 ( very expensive, you have to add some heatsink too ) USA made and stocked by Mouser, or BPC series by BI technologies ( may be too fragiles for a mobile application...but very small ).
For exact value have a look at "Calculating Optimum Snubber_Hagerman Tecnolgies" in goggle, very complex reeding but with a couple pratical tips...

About regulated PSU in a class A/B amp, LM338/LT1085 are short of ampers, not good for a guitar amp in my opinion. With regulation you are going to loose some dinamics..
Better optoin shoud be a CRC configuration if you are worried about ripple reyection and diods noise. Calculate R value in way that 2 volts are lost between caps.

ciao.

#### ITPhoenix

Thanks. I saw another trick with the Aikido preamp power supply. For the HT supply, he adds series resistors to the rectifier diodes, which are fast and oversized. I seems the resistors diminish the switching noise/ripple since there are no bypass caps on the diodes. There is no huge H inductor, just caps and resistors. Well, it's only feeding (4) 12AX7s.

Speaking of series drops between caps, that is the way it's normally done with th HT guitar tube amps, so that by the time you get to the end of the CLC/CRC chain (the first tube stage) there is nothing left but DC, so to speak.

Ok, this is starting to look better. I can compensate for the drops by adding a few turns, and still get 35v out.

Incidentally, the whole reason behind this effort is to replace whatever is lost on the lower and upper guitar frequencies. I cannot give you scope outputs or values, but I know something is missing with the "status quo" amps, even the all-tube versions. We are compensating by adding all sorts of gadgets to reach the end, in addition to various methods of addition gain, and volume boosters. With this approach, even more noise and distortion is added, besides it being "artificial".

I am of the camp that even if you cannot define something on paper, or really see something on the scope, or even hear it with your ears, your brain does. This phenomenon is in the audio texbooks. It is briefly treated in the Handbook for Sound Engineers, as well as testimony in a few forums. I do not buy, "It doesn't matter since your signal is all distorted anyway." Agreed, most people cannot tell the difference between different pickups and amps (at least consciously), or whether the signals are processed through effects loops or not, but often the musician can, and that should not be discounted.

Back to the issue at hand, I realize that every change made to the sytem will affect the entire chain, and some experimentation is unavoidable. Since I have played before, I will rely on my subjectivity more than anything else. I certainly do not want to lose anything like "dynamics" as you say, so I will proceed with unregulated power amp supply. I need to check inrush with all this capacitance and deal with that, and tranny EMI shielding and interwinding capacitance shield which everyone agrees is good.