Parallel connecting of big capacitors with smaller capacitors for optimising the sound??

[Newmarket]

Different capacitor types and values in parallel make sense for DC blocking and RF stuff. Sometimes i have seen this principle in passive speaker crossovers where the load (tweeter) has quite a low impedance...

How does that work for DC blocking ? As opposed to simply having the largest practicable capacitance and so the minimum voltage across the capacitor.

 

[rock soderstrom]

1

Thank you very much! More grades please, I will give my grade later, I don't want to influence anyone...

 

[Philbo King]

That seems a rather sweeping statement tbh. RF obviously had its particular challenges, techniques and solutions. But audio has some amazing specs compared to other fields eg SNR, Dynamic Range, THD, clock jitter etc.

Opinions do vary....

 

[Latie]

I think most people have already seen this or have done it themselves. You connect a smaller, high-quality capacitor in parallel with a larger, usually electrolytic capacitor to hopefully improve the sound characteristics.

You can see this in many places, so to narrow down this discussion and to make it more concrete, I would like to discuss it using the example of capacitor in parallel to a tube cathode resistor . Let's assume a V1 position in a head amp or microphone preamp, no overall negative feedback. Something like this for instance:
View attachment 121182

What do you think, does it make sense at this point to complement a bigger capacitor (like c9) with a smaller film capacitor (let's say a MKP 0.22uF)?

What do you expect from this? Is there a big difference?

Or is it better to just to use one single good quality electrolytic capacitor and do it without the second capacitor?

Would a version without a capacitor be the best option, according to the theorem that the best capacitor is no capacitor? (Diodes instead of a cathode resistor?)

What do you think?

Just speculating:
In some sense, two capacitors in paralell may be "more" like "no-capacitor" than one of the two capacitors alone.
(just because of that "what do you think")

Saludos
Marcelo

 

[JohnRoberts]

ChatGPT sounds more like a clueless politician every day.

Cheers

Ian

It is a product of it's creators, just like we get the politicians we support.....

JR

 

[thor.zmt]

I think most people have already seen this or have done it themselves. You connect a smaller, high-quality capacitor in parallel with a larger, usually electrolytic capacitor to hopefully improve the sound characteristics.

That is both a complex and simple problem.

Look at the involved impedance's (R/L/C) and see at what frequencies any effects can happen.

That said, there are secondary and tertiary effects not completely captured by the basic RLC equivalent model.

You can see this in many places, so to narrow down this discussion and to make it more concrete, I would like to discuss it using the example of capacitor in parallel to a tube cathode resistor . Let's assume a V1 position in a head amp or microphone preamp, no overall negative feedback. Something like this for instance:
View attachment 121182

Who says there is no feedback? C4 is feedback element.

As the input is a variable capacitor, you have a feedback loop. Not much loop gain, but that is another story. And then there is the argument of the triode's build in feedback (durchgriff).

What do you think, does it make sense at this point to complement a bigger capacitor (like c9) with a smaller film capacitor (let's say a MKP 0.22uF)?

No. But what makes sense is to consider the relevant current loops.

By connecting the output transformer "ground" lead to the cathode of Roe1 C9 no longer forms part of the output current loop. It may even be possible to dispense with it completely

What do you expect from this? Is there a big difference?

Not with 220nF vs 250000nF.

Or is it better to just to use one single good quality electrolytic capacitor and do it without the second capacitor?

Why not use other means to just eliminate the capacitor completely? Or use a compact size Film Cap. I used to use some 22uF/35V DC Film Cap's from Rubycon (IIRC) for tube cathodes. They were quite small mechanically. Not sure if are still being made though.

Alternatives would be to ground the cathode and go for "contact bias" or restructure the circuit so that the cathode moves into the negative supply line or use fixed bias. At 1.5V bias, you can get tiny coin cells to connect in series with R5, negative towards gate.

Would a version without a capacitor be the best option, according to the theorem that the best capacitor is no capacitor?

Yes.

Diodes instead of a cathode resistor?

No. Diodes are relatively non-linear. There are many ways to re-arrange the circuit so that cathode RC can be avoided.

Thor

 

[thor.zmt]

I feel dumber after reading that....

JR

#me2

ChatGTP is worse than useless for electronics.

Thor

 

[Peter Simonsen]

I think most people have already seen this or have done it themselves. You connect a smaller, high-quality capacitor in parallel with a larger, usually electrolytic capacitor to hopefully improve the sound characteristics.

You can see this in many places, so to narrow down this discussion and to make it more concrete, I would like to discuss it using the example of capacitor in parallel to a tube cathode resistor . Let's assume a V1 position in a head amp or microphone preamp, no overall negative feedback. Something like this for instance:
View attachment 121182

What do you think, does it make sense at this point to complement a bigger capacitor (like c9) with a smaller film capacitor (let's say a MKP 0.22uF)?

What do you expect from this? Is there a big difference?

Or is it better to just to use one single good quality electrolytic capacitor and do it without the second capacitor?

Would a version without a capacitor be the best option, according to the theorem that the best capacitor is no capacitor? (Diodes instead of a cathode resistor?)

What do you think?

Me I strongly belives, that 1 good quality capacitor would be more than enough.

I never in any of my designs heard a quality improvement by adding a smaler cap on the cathode (C9)..As always Ymmv

Kind regards

/Peter

 

[RCAguy]

Paralleling a distortion-prone electrolytic with a low distortion poly makes sense IF there is any significant voltage drop at audio frequencies. For power decoupling or input, interstage, or output coupling (DC blocking) with a cutoff frequency on the order of 1.6Hz, even a ‘lytic cannot generate objectionable distortion from 16Hz up. No drop, no distortion! However a timing (filter, EQ) circuit in the audible range calls for a low distortion capacitor, or paralleling with a poly of at least a few percent capacitance. [Cf. Douglas Self]

 

[Cqwet Dbdfte]

Every power supply, except perhaps batteries, brings in unwanted signal, ripple, HF noise, rectifier transitions, load dependent feedback that needs to shunted to ground.
Every capacitor has an impedance vs. frequency curve, with a big dip at its resonance point. Below the resonance the impedance is capacitive, above inductive. The closer to resonance the better the bypassing.
The caps internal inductance determines the resonance frequency. Paralleling several cap values and types has been done for a very long time also in audio amps to shunt unwanted signal to ground. A plate loaded gain stage will include roughly half of the power supply noise in the output signal. Constant current loaded gain stages are more resistive to supply noise, but will couple noise capacitively into the load, so a smaller number is better.
The resulting bypassing effort can of course be checked with a scope.
Signal coupling caps MAY benefit from paralleling, but this is not often done.
Much is written on benefits of various coupling caps, some prefer one type or another. Dielectric absorption is s factor, so film caps with better numbers for this would be a first choice.
Low ESR of caps are related to dielectric properties. ESL more to the construction technique. Film and foil may have higher inductance than metallized dielectric.
An extreme example of a capacitors are those used in radar circuits where thousands of amps is discharged every second, and metal foils brought out in a very low inductance fashion, thus caps also have a dv/dt limit as well.
Selecting bypass caps for low noise performance requirements is a balance between cost, size, longevity, temperature, and noise involved.
"Noise" not necessarily gaussian white noise but any unwanted signal.
The "best" cap may be a 100uF 600V NPO/C0G until you find out the cost of it.
Polypropylene caps are reasonable, and available, Teflon, polystyrene are considered better. Mylar are OK but not premium signal coupling caps, but good for bypassing.
Some swear by the use of paper-in-oil caps yet these have not so stellar dielectric absorption.
Tantalum caps have a bad rep since long ago, but some may like them. Besides expensive they also can fail spectacularly, and is a conflict material.
Niobium and organic electrolytes I have not looked into, perhaps OK for low voltage.
Electrolytics are indispensable but have a finite lifespan, which have to be considered in servicing equipment. Temp- and ripple current ratings have to be checked against the use case. Their ripple filtering capability may be more important than their energy storage value, and selected for lowest ESR values.
This is a pretty big subject, and it is hard to avoid using capacitors.
In the early days of electronics when caps were unobtanium transformer coupling was used. The Western Electric 47 mike used no coupling cap but ran plate current thru the output transformer which of course also provided galvanic isolation and voltage stepping.
Transformers were also used in RF circuits to keep kilovolts away from antennas

 

[thor.zmt]

Im rethinking and reviewing my own habits from time to time.

Good idea.

My own experiences are very mixed and inconsistent. The issue also touches the controversial topic of "capacitor sound" which is not only highly subjective but also difficult to measure.

Yes and no. I would argue a lot of this sound is easy to measure, you just need to know WHAT to measure.

In a world where the great unwashed from the "objectivist" group will come with Pitchforks and Torches to the door of the Heretic who states that not everything sound the same and that not all differences are imaginary and on the other side the "subjectivists" will insist that a holographic sticker treated with "electro cream" will improve the sound, there seems to be no middle ground from which to investigate the subject with sufficient science.

I believe that the component (here capacitors) quality can play a role in terms of sound, but the circuit environment also plays an important role here.

I once compared numerous different capacitors of all qualities and prices with a dedicated apparatus and the results were disillusioning.

I wanted to believe in it, but the differences were negligible in a double-blind situation and, above all, not reproducible. In essence, it didn't matter whether it was an old used electrolytic capacitor from a cheap Italian organ or a super expensive "high end" MKP cap.

Well, did you actually "test the test"?

Let me give you an example. My digital scope has around 80dB dynamic range in FFT mode. If I am looking for the 10th harmonic at -90dB which I postulate is the cause of an audible difference, will my test using this tool reveal it?

And will the failure of my test to reveal it mean that it doesn't exist?

Many years ago we used to have a "component tester" which was an Audio Note Japan / Kondo M7 line copy. Every component could be selected by rotary switch from up to 6 Options. We tested all sorts. Naturally blind (but not ABX), but literally blind rto even what was being tested. And yes, there are differences. For example, in my experiences much of "capacitor sound" in high voltage circuits is down to mechanical behaviour, not anything an HD or RLC evaluation would reveal and which only exists in a listening room where the components are excited by soundwaves from the speakers.

So it doesn't matter which capacitor you use?

It depends A LOT on the circuit.

Electrolytic Capacitors are offensive enough to have distortion levels that disturb Mr Audio Precision. Especially Polar ones. Cyril Bateman did a massive article series in Wireless world on the topic.

Mylar and Polycarbonate (and Acrylic) can also show objective problems.

High K ceramics (that large value in small size) are plain awful. Of course, this can be used intentionally to acoustic advantage. If you have a small speaker with modest LF extension using an "awful high K" coupling capacitor in the highpass (perhaps in a bypass with 1/3rd film) will generate harmonics of the blocked fundamental, leading to the "missing fundamental" acoustic illusion.

But USUALLY it is best to design a circuit so that problems arising from capacitors are minimised.

Like, which of these following two circuits will perform better, objectively and subjectively in your view?

Interesting, I also follow the min. 10% rule of thumb.

This is a very bad rule for RF, switching and arguably for audio.

In truth it is necessary to analyse the complex circuit, including the impact of the load.

For example, for digital (CMOS based) audio systems (10's of MHz clock) I now religiously use tightly coupled ground and power planes with usually only multiple 0603 or smaller physical size X7R 1uF capacitors. Compare the complex impedance of 1uF & 100nF in 0603 to just using two 2 X 1uF. The 2 X 1uF wins at ALL frequencies with lower impedance and less resonnances. Electrolytic Capacitors I typically select for a moderate ESR (not super low, though that depends on the overall circuit) to act more as RC snubber, rather than actual decoupling capacitor.

Yes, electrolytic capacitors have become much better in recent decades.

Not really. Modern commodity types maximise C/V vs size.

Compare that older "smooth anode" types and compare how the foil for "Audio Grade" types is etched compared to minimum size/cost types. They are larger and cost more but can be shown to be objectively better.

I don't think I have the equipment or the knowledge for reliable measurements ATM.

AP2 or even Quantasylum QA400 is fine.

Thor

 

[soapfoot]

#me2

ChatGTP is worse than useless for electronics.

Thor

The problem with training an AI to effectively summarize the internet is that it will repeat/spread all sorts of misinformation

 

[Cqwet Dbdfte]

Like, which of these following two circuits will perform better, objectively and subjectively in your view?

View attachment 121344

I think the bias is not missing between "out_smrt" and "out_stpd" net names , but maybe some voltage bias is needed on signal chain electrolytics, or use two, or bipolar electrolytics? Pin 4 to a negative supply?
A high impedance load on those coupling caps should reduce whatever effect from parasitic properties.
Otherwise there is no objective difference between the circuits.

 

[thor.zmt]

I think the bias is not missing between "out_smrt" and "out_stpd" net names

but maybe some voltage bias is needed on signal chain electrolytics, or use two,

Does not work to lower distortion induced, see Bateman.

or bipolar electrolytics?

That would help, if you can get them and fit physically and budgetwise.

Pin 4 to a negative supply?

Power supply is omitted from drawings, but would be (say) +/-12V.

A high impedance load on those coupling caps should reduce whatever effect from parasitic properties.

But for the Feedback Loop DC block the impedance cannot be high, output we do not control.

Otherwise there is no objective difference between the circuits.

I disagree.

The second circuit uses non-polar film capacitors on input and on the feedback loop which if selected correctly are free from any distortion we worry about (unlike electrolytic capacitors) which in turn terminate into a very high termination impedance which is possible as they have zero relevant DC leakage. MKP film, or other low distortion film types should be used.

The output coupling capacitor is enclosed in the AC feedback loop and thus it's distortion is corrected by the looped feedback.

Once we establish operating conditions where we can measure the distortion of the Capacitors in Circuit 1 (see D. Self), we will observe no added distortion in Circuit 2, over that of the active components.

Unless you happen to have the magical "distortionless" electrolytic capacitors at hand, #2 is objectively superior (and likely cheaper).

Thor

 

[ruffrecords]

It reminds me more of some of my former students in exams. Somehow they learned the subject matter but didn't really understand it and to cover it up they say way too much.

How would you grade this answer?

(Marks 1 to 6, where 1 is very good and 6 is really bad).

See me after.

Cheers

Ian

 

[Cqwet Dbdfte]

Including the polar output cap in the feedback loop would of course negate, or rather move the cap issues into the feedback system.
The same currents will still pass thru the cap which hopefully feedback reduces to oblivion.
100dB open loop gain(?) to give 10dB?
I would prefer no feedback from intrinsically more linear single devices with more gain, even a transformer if space, budget, a loading allows it.
The output cap may be eliminated with a DC-servo in some cases, or like in solid state power amps not used at all.
Not sure exactly the application here.
I have not looked for any specific cap IM or harmonic distorsion data, nor measured any.
Probably a good idea.

 

[thor.zmt]

Including the polar output cap in the feedback loop would of course negate, or rather move the cap issues into the feedback system.

Which we need to note is present inherently in the circuit.

I would prefer no feedback from intrinsically more linear single devices with more gain, even a transformer if space, budget, a loading allows it.

With unlimited space, weight and budget, of course.

What I illustrated is very commonly circuit design and how to make small adjustments that don't fundamentally change the circuit, but smartly avoids the problems of real capacitors.

The output cap may be eliminated with a DC-servo in some cases,

This then implies adding DC protection and Relays. And of course the servo. Which needs care how it is tied into the signal circuit.

or like in solid state power amps not used at all.

Most solid state amp's use mechanical high current relays in DC protection circuitry, which often cause more distortion than capacitors.

Not sure exactly the application here.

No specific application, just an illustration of how to minimise non-ideal capacitor problems in a real circuit with minimal effort.

I have not looked for any specific cap IM or harmonic distorsion data, nor measured any.
Probably a good idea.

To repeat, Cyril Bateman's Article Series in Wireless World in the last millennium is a great start. It ran across many issues.

Generally, good coverage of components nonideal properties is found in the pages of Linear Audio.

Thor

 

[Newmarket]

Opinions do vary....

Well tbh it's not really about "opinions". There are hard facts in play here. The question is what parameters are relevant in any specific case.

 

[living sounds]

Yes, use only (quality) bipolar electrolytics in the signal path, even (or especially) in places with significant DC bias present. Measurements are clear that a bipolar beats a polar cap, usually by big margins, when it comes to distortion.

BTW, it's usually OK for audio decoupling of common (not high speed) op amps to use normal (not low-ESR) electrolytic bypass caps at the regulator and the smaller x7r ceramics at or near the op amp.

 

[Cqwet Dbdfte]

Thor, thanks for the tip about Batemans work.
https://linearaudio.net/cyril-batemans-capacitor-sound-articlesHe gave up early on measuring tantalum's as their results were too ridiculous!