Which Capacitors for Audio?

[Jeff5]

I suspect availability.

A current small scale (expected 10's of units per year) commercial consultancy project of mine needs a range of high quality capacitors.

The smallest values are 100pF - 1,000pF and the best quality available is polystyrene in 50V DC Rating (Xicon 23PS).

Larger values I selected WIMA FKP2 in 63V DC, but I also need some 1uF, these ended up WIMA FKP4 in 400V, simply based on what Mouser stocks and pricing. I'd have liked physically smaller 63V types, but they just don't have availability.

Thor

That makes sense. Thanks

 

[ct.waveform]

I 'think' this question is a good fit for this thread. Let me know if it's not. Thanks.

Can anyone shed light on why some people use filter caps with larger voltage ratings for lower frequencies and smaller voltage ratings for higher frequencies?

I'm building a Pultec and I'm noticing this trend in many builds.

Not wishing to disagree with Thor, who is probably right in the more sane cases, but I think it's part of audiophile idiocy. It has got completely out of control in some hi fi equipment over the last couple of decades with an obsession over dielectric absorbtion and electrolytics being considered Satan's spawn. Forgetting that capacitors are wound components, they have gone for larger and larger caps (with larger and larger inductances) without considering how much current they will be carrying or any consideration of noise; picked up from electric and magnetic fields and often helped along by casework. In short they think it's going to sound better, which has some truth in it, but they have gone too far. In the setting of a loudspeaker crossover (one of these easiest place to test what sounds good, but not necessarily representative of the feedback loop of an op amp) you can get enough current to move the plates and modulate the signal so the further apart they are the less this will happen. Also in this setting there are mechanical resonances inside the audio band which benefit from the damping properties of polypropylene. I have tortured various capacitors (and driving amplifiers) in the past and it's surprising quite how loud they can "play". (Since microphony is the corollary of this you might wonder why they then put the crossovers inside the cabinets where the sound intensity is astronomical?). They often transport their listening results directly into amplifiers and other equipment.

So I would say it's misguided fashion and the additional expense is often worth it for having something to write about in their marketing blurb and to give some "audiophile" cred. Capacitors unquestionably sound different but putting big ones in places that are close to any amplifier input (which is most of the places) is a terrible idea.

 

[ct.waveform]

I had the same question when I got into all this stuff. I have discovered that the type of cap offers more variety than the brand of a cap. Such as putting a polypropylene cap inplace of a ceramic can give slight tonal differences. Good old Panasonic FC's and FM's are cheap very good electrolytics. Nichicons are also well regarded. No need for black gates!

I don't know the answer to the poly film question. I just use whatever. I have WIMA's all over one channel of my LA 4 (MKS I think), but then in my other channel I have some panasonic polyester films because I had them on hand. Don't notice any difference.

What I want to know is resistors. Does it really matter if I use the bags of 10 for $0.50 from the local store? Do Vishay dale/Xicon resistors make that big of a difference? I know that tolerances come into play here, but what about noise?

Yup, high quality resistors make a huge difference in audio, particularly in respect the TCR (temperature coefficient of resistance) and power rating chosen. It's not hard to see why, either. Take a standard feedback network with a gain of 30 and look at the power dissipated in each resisitor. It's 30x as high in one resistor as it is in the other. So 300ppm/degC is not going to cut it. You can see the slopes of dissipation in any of the datasheets for the high quality Vishay/Dale resistors. If you can't even keep the gain of your amplifier constant, what hope do you have of it sounding good?

In other applications, say a CR arm compensating for the impedance of a midbass loudspeaker, you can imagine the R is doing a fair bit of work. (In fact you can often see scorched PCBs under these resistors.) Not only is the change in temp going to modulate the voltage across it and the loudspeaker but this is going to be frequency dependent because of the higher amplitudes at lower frequencies. Moreover, the change in load is going to affect the shape and -3dB points of the preceding circuitry. You can't do anything about the TCR of the copper in the coil, which is huge, but you'd be surprised by how much having a constant R in parallel with a variable one linearises things at the frequencies of interest. Current driving a loudspeaker reduces distortion by about 20dB so there are real gains to be picked up, not something marginal.

 

[ct.waveform]

I have a lot of time for Rod but he does occasionally make silly blunders in his explanations. In his first paragraph what he means to say is:

"if the cap is large enough (compared to frequency and circuit resistance), it will never have any appreciable AC voltage across it. "

I added the AC. He then goes on to say "With no voltage, there is no stored energy." which, if the capacitor is blocking dc is incorrect. Other than that it is a very good article.

Cheers

Ian

Me too. I like Rod and I think he's done us all a huge service with his website. I couldn't get that link to open but if it's Rod then I've probably read it already. I also seem to remember there was something else I disagreed with him about, if it's the one I'm thinking of. It was something to do with the big capacitor to Gnd in the f/b network and there was an element he'd completely overlooked. I can't now bring to mind what it was exactly, but I do remember that Kendall Castor-Perry had proposed a solution (or partial solution) to it in a discussion of a Linear Audio article on the topic.

 

[ct.waveform]

Tau (time-constant)=1second, -3dB point at 0.16 Hz and -0.00027dB at 20Hz. A little on the overkill side, but nothing wrong with it.

I'd say 1/f noise and bias stability might be good starting points for what's wrong with it. Sorry.

 

[Jeff5]

Not wishing to disagree with Thor, who is probably right in the more sane cases, but I think it's part of audiophile idiocy. It has got completely out of control in some hi fi equipment over the last couple of decades with an obsession over dielectric absorbtion and electrolytics being considered Satan's spawn. Forgetting that capacitors are wound components, they have gone for larger and larger caps (with larger and larger inductances) without considering how much current they will be carrying or any consideration of noise; picked up from electric and magnetic fields and often helped along by casework. In short they think it's going to sound better, which has some truth in it, but they have gone too far. In the setting of a loudspeaker crossover (one of these easiest place to test what sounds good, but not necessarily representative of the feedback loop of an op amp) you can get enough current to move the plates and modulate the signal so the further apart they are the less this will happen. Also in this setting there are mechanical resonances inside the audio band which benefit from the damping properties of polypropylene. I have tortured various capacitors (and driving amplifiers) in the past and it's surprising quite how loud they can "play". (Since microphony is the corollary of this you might wonder why they then put the crossovers inside the cabinets where the sound intensity is astronomical?). They often transport their listening results directly into amplifiers and other equipment.

So I would say it's misguided fashion and the additional expense is often worth it for having something to write about in their marketing blurb and to give some "audiophile" cred. Capacitors unquestionably sound different but putting big ones in places that are close to any amplifier input (which is most of the places) is a terrible idea.

I'm glad you point this out, because I was feeling influenced by these pictures of huge caps and thinking I should use bigger ones too. I was wondering about noise issues also. Maybe there's a reason that some of the classic pieces of gear have so few parts inside, with many components mounted outside the case. I'll be happy with normal caps that fit comfortably in the alloted space now. Thanks!

 

[thor.zmt]

Not wishing to disagree with Thor, who is probably right in the more sane cases, but I think it's part of audiophile idiocy.

I think not.

It has got completely out of control in some hi fi equipment over the last couple of decades with an obsession over dielectric absorbtion and electrolytics being considered Satan's spawn.

Based on relative distortion data, it's not an Obsession and electrochemical capacitors are the worst possible capacitor to use for almost any application.

I find that using bipolar types for signal positions and enclosing the capacitors in the AC feedback loop and/or refactoring the circuit to only use film or high quality ceramic types gives measurable and audible benefits.

Forgetting that capacitors are wound components, they have gone for larger and larger caps (with larger and larger inductances) without considering how much current they will be carrying or any consideration of noise; picked up from electric and magnetic fields and often helped along by casework.

Simply untrue. And if your casework has AC magnetic fields IN THE CASEWORK (as opposed to screening them) you have problems on a level we not need to talk capacitors.

In the setting of a loudspeaker crossover (one of these easiest place to test what sounds good, but not necessarily representative of the feedback loop of an op amp) you can get enough current to move the plates and modulate the signal so the further apart they are the less this will happen.

FWIW, I had a design where polystyrene RIAA EQ capacitors were notably microphonic but silver mica were not.

A lot of capacitor problems are real. Ranting that audiophiles are stupid for dealing with them is not constructive.

Also in this setting there are mechanical resonances inside the audio band which benefit from the damping properties of polypropylene.

Polypropylene is not particularly well damped. Hence my preferrence for stacked film typed in plastic cases, e.g. WIMA FKPxx & MKPxx.

I have tortured various capacitors (and driving amplifiers) in the past and it's surprising quite how loud they can "play". (Since microphony is the corollary of this you might wonder why they then put the crossovers inside the cabinets where the sound intensity is astronomical?). They often transport their listening results directly into amplifiers and other equipment.

Yup. And this is a strong factor in "capacitor sound" often overlooked.

So I would say it's misguided fashion and the additional expense is often worth it for having something to write about in their marketing blurb and to give some "audiophile" cred.

Unlike in recording and pro audio, the systems used by audiophiles have few pieces of equipment (to wit, digital transport, DAC, Preamp and Poweramp plus speakers and such systems are often quite static but see long term small strip "Tuning" to the owners preference.

"Rolling" tubes, Op-Amps and capacitors or swapping cables (and yes they make differences) are accessible to relative laymen in ways that serious circuit optimisation or restructuring grounding are not.

Capacitors unquestionably sound different but putting big ones in places that are close to any amplifier input (which is most of the places) is a terrible idea.

Audiophiles gear tends to use direct coupling and servo's as a result, not that this really helps the way it's usually done, IMNSHO. Amplifier input circuits have sufficient loop area and sensitivity, that is a (say) 27mm long capacitor instead of a 10mm one causes problems, you have to fix the bad layout etc first.

Thor

 

[thor.zmt]

I'm glad you point this out, because I was feeling influenced by these pictures of huge caps and thinking I should use bigger ones too.

When we (for example) take electrolytic capacitors, old types often used smooth aluminum foil.

The metal foil most likely is also thicker due to process limitations in making commercial foil back 70 years ago and the isolating paper or film was likely also ticker.

This makes for a capacitor that is comparably large physically for a given rating when we look and compare modern parts.

It also makes for a part with less microphonics and distortion than a modern part that is much smaller.

It is one of the reason that modern replicas of old gear do not sound the same. (Transformers have the same issue, it is literally impossible to make exact replicas).

The same capacitor value and Voltage rating do not reliably produce the same result in more subtle aspects of sound quality.

At the same time just making capacitors physically big is not an Automatic solution.

Yet, using for example electrolytic capacitors with higher than required ripple current ratings (makes them bigger) and higher than required voltage and temperature rating can result in capacitors that are closer in mechanical design to classic parts.

Equally, modern film capacitors have much thiner film than classic examples at a given voltage rating. Using higher voltage makes party physically bigger but uses thickef foil and thus again sound quality is closer to older original parts.

However given that in many cases true equivalents do not exist, it is questionable if going this route is valid. At best it will still not sound like the original.

I was wondering about noise issues also. Maybe there's a reason that some of the classic pieces of gear have so few parts inside, with many components mounted outside the case.

Serviceability among other factors. Limitations in construction. There was no 3D printing and metal fabrication was commonly done in house with simple gear to fold metal and punches to make holes. Printed circuit boards were primitive and often more of a reliability issue than they are now, so many items are hardwired.

History of electronics is interesting, I grew up with old tube and early transistor gear and still remember much of this first hand.

Thor

 

[abbey road d enfer]

I'd say 1/f noise and bias stability might be good starting points for what's wrong with it. Sorry.

Care to enlighten me?

 

[terry setter]

I wanted to test C0G/NP0 caps on the output of my tube mics.

Thor said: Hmm, not a good use, as the values and voltages available are extremely limited.

Finally, I ordered three .18mF/200V C0Gs and paralleled them just in front of a Cinemag NiCo output transformer. When I scoped them, I couldn't believe my eyes. They were TERRIBLE!

Thor asked: You ordered from Mouser? Or another reliable supplier of genuine parts? I am not even aware of such values/voltages being commercially available. Did you verify that you got C0G and not Y5U?

Reply: I ordered from Mouser. 3ea. 018.mF/200V C0G caps (Mouser number: 80-C350C184K2G5TA). They were expensive, about $2.75 each, and they only had about 40 VDC on them. I'd like to think that they are the real McKoy, but, like I said in my earlier posts, I found that they did not work for me as the final blocking cap in my tube mic circuit, despite the fact that all the other types I tried were acceptable or even excellent.

Has anyone else tried C0G caps in such a configuration?

 

[thor.zmt]

Reply: I ordered from Mouser. 3ea. 018.mF/200V C0G caps (Mouser number: 80-C350C184K2G5TA). They were expensive, about $2.75 each, and they only had about 40 VDC on them.

Ok, that should work and mouser stock SHOULD BE ok.

Has anyone else tried C0G caps in such a configuration?

Not specifically in a Mic. But I routinely use TDK 0.1uF/50V/1206 SMD in coupling cap positions with as much as 2/3 of DC across them (in most cases much less).

When they became very difficult to source due to worldwide shortages we tried many alternative options. Testing was objective (Audio Precision) and subjective.

The finaly adopted solution was to use Panasonic stacked film 0.22uF/16V ECP-U (or equivalent) in parallel with 10nF/50V/C0G/0805 from Samsung who had an OEM factory around the corner from us in Dongguan.

For higher blocking voltage we went to through hole WIMA MKP with C0G bypass.

The new combo was sonically about equal, cost less and reduced sourcing headaches by over 60dB.

The original coupling capacitor value is 0.47uF?

Thor

 

[abbey road d enfer]

The finaly adopted solution was to use Panasonic stacked film 0.22uF/16V ECP-U (or equivalent) in parallel with 10nF/50V/C0G/0805 from Samsung who had an OEM factory around the corner from us in Dongguan.

Can you elaborate on that? What advantages result from paralleling devices? What is the load and signal level?

 

[thor.zmt]

Can you elaborate on that? What advantages result from paralleling devices? What is the load and signal level?

The paralleling is the result of listening tests (single blind preference using visually identical devices, PCBs inside modified).

Load usually is several megaohm plus J-Fets or MOSFET, some times tubes.

Signal Level, broadly line, often attenuated.

Typically these capacitors are the sole coupling capacitors outside feedback loops in low noise, low distortion circuitry.

Thor

 

[abbey road d enfer]

The paralleling is the result of listening tests (single blind preference using visually identical devices,

OK, but you don't have a theory about the mechanisms at work?

 

[thor.zmt]

OK, but you don't have a theory about the mechanisms at work?

Nothing I'd care to share in public. But the capacitors have different behaviour both electrically and mechanically.

Once the the supply issue resolved itself production went back to using 0.1uF C0G, the 10n were left in place as the PCB's had been redesigned to include them.

Thor

 

[abbey road d enfer]

Nothing I'd care to share in public. But the capacitors have different behaviour both electrically and mechanically.

Sure they do.
Do you think it's a mechanism of compensation (like a PTC would compensate a NTC) or just cumulative effects (like paralleling an electrolytic with a ceramic, which cumulates high capacitance and low inductance)?

 

[living sounds]

Yup, high quality resistors make a huge difference in audio, particularly in respect the TCR (temperature coefficient of resistance) and power rating chosen. It's not hard to see why, either. Take a standard feedback network with a gain of 30 and look at the power dissipated in each resisitor. It's 30x as high in one resistor as it is in the other. So 300ppm/degC is not going to cut it. You can see the slopes of dissipation in any of the datasheets for the high quality Vishay/Dale resistors. If you can't even keep the gain of your amplifier constant, what hope do you have of it sounding good?

Is this relevant for the currents in small signal solid state circuitry utilizing IC op amps, say a mixing console?

 

[thor.zmt]

Do you think it's a mechanism of compensation (like a PTC would compensate a NTC) or just cumulative effects (like paralleling an electrolytic with a ceramic, which cumulates high capacitance and low inductance)?

No, electrically in this situation we can illustrate that first of all small capacitance changes are immaterial and that the effect of the capacitors are minimised due to operation from a low source impedance and terminating into a very high impedance.

So any audible effects must relate to other factors. These could be such items as microphonics, e.g. the parallel "monolithic" capacitor attenuates microphonic noise from the larger capacitor.

Other putative noise related effects may play a role (e.g. MDI). Any effects from dielectric absorption might be reduced too, especially at higher frequencies.

I also find that adding small value C0G SMD Capacitors on PSU line bypassing gives objective improvements in excess of what basic lumped element modeling suggests.

It is something that would probably worth researching, but I'm not a tenured professor at a research focused University. I just need results.

Thor

 

[thor.zmt]

Is this relevant for the currents in small signal solid state circuitry utilizing IC op amps, say a mixing console?

It will depend, it CAN.

Past this effect, there are other effects that cause increased harmonic distortion with certain types of resistors.

Thor

 

[living sounds]

It will depend, it CAN.

Past this effect, there are other effects that cause increased harmonic distortion with certain types of resistors.

Thor

It's my understanding that you can get to a point of increased distortion in these types of circuits with certain types of very small surface mount resistors. But can this be an issue with the usual 0.6w through hole metal film resistors?