The No.1 Website for Pro Audio
AES paper on capacitors for audio
Old 10th September 2020
  #1
Here for the gear
 
🎧 5 years
AES paper on capacitors for audio

I've been fishing the web recently for info about capacitors for overhauling a forty year old tape machine and came across this on Reddit. Looks interesting. Specifically electrolytics used as coupling caps. Thanks to jaymz168 on Reddit.

Summary of Publication:
This paper provides a number of comparative, quantitative evaluations of 10 different makes and models of electrolytic capacitors. Models range from expensive parts specified for use in audio circuits to low-cost general-purpose parts. The datasets comprise out-of-circuit electronic measurements, total harmonic distortion (THD) fast Fourier transform (FFT) sweeps, and cumulative distortion products resulting from 31-tone stimulus performed on the components in a circuit designed to emulate a typical line-level audio recording and mixing console. Results are examined in an effort to identify any measurable properties that may distinguish "audio capacitors" as outliers from their general-purpose counterparts.

PDF Download: http://www.aes.org/e-lib/download.cf...1.pdf?ID=20891
Permalink: http://www.aes.org/e-lib/browse.cfm?elib=20891
Affiliations: Kent State University at Stark, North Canton, Ohio
Authors: Anderson, Ian Z.
Publication Date: 2020-09-01
Introduced at: JAES Volume 68 Issue 7/8 pp. 559-567; July 2020
Old 10th September 2020
  #2
Gear Guru
Unfortunately, a common film cap will exceed the performance of even the best el cap. The discontinued Rubycon "Black Gate" caps were different and measured better than the rest.

Any el cap up to 10 uf is replaced here with a 5 mm WIMA MKS-2 film cap. The extra benefit is you never need to replace them again.
Old 10th September 2020
  #3
Gear Guru
 
tINY's Avatar
 
1 Review written
🎧 15 years
http://www.reliablecapacitors.com/ol...m/pickcap.html

This is a pretty good read... If you have the time.



-tINY

Old 10th September 2020
  #4
Lives for gear
 
lame pseudonym's Avatar
 
🎧 15 years
I don't see dilectric absorbtion in the list of factors that they measured. That is usually my first concern. Second concern, I should say, I won't pay $50 for a capacitor.
Old 11th September 2020 | Show parent
  #5
Gear Guru
 
tINY's Avatar
 
1 Review written
🎧 15 years
Quote:
Originally Posted by lame pseudonym ➑️
I don't see dilectric absorbtion in the list of factors that they measured. That is usually my first concern. Second concern, I should say, I won't pay $50 for a capacitor.

Well, when your subwoofer can reproduce 5Hz, it probably makes a difference....



-tINY

Old 11th September 2020 | Show parent
  #6
Lives for gear
 
JohnRoberts's Avatar
 
🎧 10 years
Quote:
Originally Posted by tINY ➑️
http://www.reliablecapacitors.com/ol...m/pickcap.html

This is a pretty good read... If you have the time.



-tINY

I had problems with some of that stuff when the ink was wet... (80s).

JR
Old 11th September 2020 | Show parent
  #7
Lives for gear
 
JohnRoberts's Avatar
 
🎧 10 years
Quote:
Originally Posted by lame pseudonym ➑️
I don't see dilectric absorbtion in the list of factors that they measured. That is usually my first concern. Second concern, I should say, I won't pay $50 for a capacitor.
While I share your disdain for $50 capacitors I only considered DA in capacitor selection exactly one time back in the early 80s.

DA (dielectric absorption) can be modeled as smaller RCs in parallel with a large C, this is relatively innocuous in lightly loaded DC blocking applications, but can make a difference if cap is charged and discharged by different impedances. An obvious problem for sample and hold circuits, less obvious but a source of errors in side chain time constant circuits that are routinely charged and discharged differently to deliver different attack and release times.

My one design application where DA mattered was a CX** decoder kit (CX was a below threshold 2:1 companding NR system). From the CBS licensee package, including pro forma decoder designs, I noticed that the CX encoder, designed by Urie, used a tantalum cap in the side chain attack/release circuit. Since tantalum caps are notorious for DA, used in a circuit where it could matter, I used a tantalum cap in my decoder side chain to mirror those subtle timing errors. My effort for precision was misplaced. I found a math error in the proforma design time constant circuit, so I notified CBS of the error in their documentation and made it correct in my kit version.

I found out from Urie at the Fall AES show that year (just before my kit article published), that two CBS licensees had copied the incorrect time constant circuit verbatim, and already shipped tens of thousands of decoders with the wrong time constant. CBS decided to change the encoder to agree with all the wrong decoders, making me the oddball because I did it correctly to meet their published spec. I rushed the changes into my kit article so now we were all wrong the same way.

This was about the only time I ever designed in a tantalum capacitor.

JR



** Back in the 80s CBS awarded me a free license to their ill fated CX vinyl noise reduction system, in return for a kit cover article on Popular Electronics magazine. That seemed like a layup so I cranked out a kit article in weeks.
Old 11th September 2020 | Show parent
  #8
Gear Guru
 
tINY's Avatar
 
1 Review written
🎧 15 years
Quote:
Originally Posted by JohnRoberts ➑️
I had problems with some of that stuff when the ink was wet... (80s).

JR

I wouldn't take this article as gospel... But it does a good job at getting you thinking about a marginal problem and what you might do about it.

The biggest condition that people don't account for: How big is your signal compared to the constant DC bias voltage across your cap?




-tINY

Old 11th September 2020 | Show parent
  #9
Lives for gear
 
JohnRoberts's Avatar
 
🎧 10 years
Quote:
Originally Posted by tINY ➑️

I wouldn't take this article as gospel... But it does a good job at getting you thinking about a marginal problem and what you might do about it.

The biggest condition that people don't account for: How big is your signal compared to the constant DC bias voltage across your cap?




-tINY

I think I already talked about this recently... I am waiting on PCB for a bump box I designed to add voltage gain between my cheap 5.1 decoder (1V output) and hypex amp modules (need 2.5V for full output). For convenience I am running the bump box from single 5V supply..

So to answer your question 1.25VAC max, with 2.5V DC bias. I purchased some cheap electrolytic caps from Digikey and don't even remember what brand.. I'll report back if they hurt my ears, but don't hold your breath.

JR
Old 11th September 2020 | Show parent
  #10
Lives for gear
 
lame pseudonym's Avatar
 
🎧 15 years
Quote:
Originally Posted by tINY ➑️

Well, when your subwoofer can reproduce 5Hz, it probably makes a difference....



-tINY


Well, I guess you can go argue with Walter Jung and Richard March:


"It is quite illuminating to consider what effect a phenomenon such as DA will have on an a.c. signal consisting largely of transients (such as audio) might have. For example, when an a.c. voltage is applied, there is a tendency for the dielectric absorption phenomenon to oppose this change in polarity.

When music is the a.c. signal, the sonic degradation is one of compression or a restriction of the dynamic range. Also, a loss of detail results, and the sharpness is noticeably dulled. With dielectric types which have high DA, there is a definite "grundge" or hashy distortion added to the signal.

It is quite important to describe the sonic thumbprint that DA contributes to subjective audio. The effects of DF and DA can be perceived differently. DF is primarily a contributor to phase and amplitude modulation DA reduces or compresses dynamic range. This it does by not returning the energy applied all at once.

With signal applied to a capacitor with DA present, the amplitude is reduced by the percent DA. When this energy does get returned (later), it is unrelated to the music and sounds like noise or "garbage" being added; the noise floor is also raised. High-frequency and/or transient signals are audibly compressed the most. "


http://reliablecapacitors.com/oldRC/...m/pickcap.html


This is what I'm hearing. Your mileage may vary, and I wish you well.

.
Old 11th September 2020 | Show parent
  #11
Gear Guru
 
tINY's Avatar
 
1 Review written
🎧 15 years
Quote:
Originally Posted by JohnRoberts ➑️
I think I already talked about this recently... I am waiting on PCB for a bump box I designed to add voltage gain between my cheap 5.1 decoder (1V output) and hypex amp modules (need 2.5V for full output). For convenience I am running the bump box from single 5V supply..

So to answer your question 1.25VAC max, with 2.5V DC bias. I purchased some cheap electrolytic caps from Digikey and don't even remember what brand.. I'll report back if they hurt my ears, but don't hold your breath.

JR

That's a good test case - leaves you about 0.75v headroom....

It may not be an issue at the impedances involved.....



-tINY

Old 11th September 2020 | Show parent
  #12
Gear Guru
 
tINY's Avatar
 
1 Review written
🎧 15 years
Quote:
Originally Posted by lame pseudonym ➑️
Well, I guess you can go argue with Walter Jung and Richard March:
...

This is what I'm hearing. Your mileage may vary, and I wish you well.

.

This was the weakest part of the article... why would the energy stored in Dielectric Absorption be returned to the conductor all at once? Wouldn't it be returned over a long time due to the resistive elements?

Have you seen any measurements?



-tINY

Old 12th September 2020 | Show parent
  #13
Lives for gear
 
JohnRoberts's Avatar
 
🎧 10 years
Quote:
Originally Posted by lame pseudonym ➑️
Well, I guess you can go argue with Walter Jung and Richard March:
Do you mean Marsh? I kind of already did back decades ago.

JR
Quote:

"It is quite illuminating to consider what effect a phenomenon such as DA will have on an a.c. signal consisting largely of transients (such as audio) might have. For example, when an a.c. voltage is applied, there is a tendency for the dielectric absorption phenomenon to oppose this change in polarity.

When music is the a.c. signal, the sonic degradation is one of compression or a restriction of the dynamic range. Also, a loss of detail results, and the sharpness is noticeably dulled. With dielectric types which have high DA, there is a definite "grundge" or hashy distortion added to the signal.

It is quite important to describe the sonic thumbprint that DA contributes to subjective audio. The effects of DF and DA can be perceived differently. DF is primarily a contributor to phase and amplitude modulation DA reduces or compresses dynamic range. This it does by not returning the energy applied all at once.

With signal applied to a capacitor with DA present, the amplitude is reduced by the percent DA. When this energy does get returned (later), it is unrelated to the music and sounds like noise or "garbage" being added; the noise floor is also raised. High-frequency and/or transient signals are audibly compressed the most. "


http://reliablecapacitors.com/oldRC/...m/pickcap.html


This is what I'm hearing. Your mileage may vary, and I wish you well.

.
Old 12th September 2020 | Show parent
  #14
Lives for gear
 
lame pseudonym's Avatar
 
🎧 15 years
Quote:
Originally Posted by tINY ➑️

This was the weakest part of the article... why would the energy stored in Dielectric Absorption be returned to the conductor all at once? Wouldn't it be returned over a long time due to the resistive elements?

Have you seen any measurements?



-tINY


The release time would be at least comparable to the acquisition time, wouldn't it?

Anyway, I'm not the kind of guy who determines what I'm hearing by looking at my slide rule. What they're hearing is kinda like what I'm hearing, and that's good enough for me.



.
Old 12th September 2020 | Show parent
  #15
Lives for gear
 
JohnRoberts's Avatar
 
🎧 10 years
Quote:
Originally Posted by tINY ➑️

This was the weakest part of the article... why would the energy stored in Dielectric Absorption be returned to the conductor all at once? Wouldn't it be returned over a long time due to the resistive elements?

Have you seen any measurements?



-tINY

DA is a well measured and understood mechanism in sample and hold design. For linear signal paths with symmetrical charge /discharge loading the effects of DA are more subtle if at all.

IIRC back in the day I wrote a LTE regarding one similar discussion in TAA suggesting that they use null testing to detect effects of DA in a typical audio path application, if at all.

===

DA is a difficult mechanism to characterize. The popular model shows a bunch of parallel smaller series RCs in parallel with the large C. In reality there are no actual resistors inside the capacitor. The DA phenomenon is related to how the electrostatic charge is distributed and moves within the capacitor media. A RC is the best electrical model for the observed behavior, just not literally what is going on.

===
Quote:
Originally Posted by tINY ➑️

That's a good test case - leaves you about 0.75v headroom....

It may not be an issue at the impedances involved.....
Headroom? I do not follow.

In a lightly loaded DC blocking application the bias voltage is constant and floating on top of the changing AC. There are millions of polar electrolytic capacitors operated successfully with almost zero bias voltage. My 2.5V of bias is an accident of the single supply design I used to KISS.

JR
Old 12th September 2020
  #16
Gear Guru
It's very common to find an el cap mounted backwards. The common rule is + lead towards the output device. In opamp circuits that is done and they should be checked. They still pass audio with reverse bias but the THD probably isn't very good.

Soundcraft 6000 and Delta consoles use a negative 2 volt DC bias in their hybrid sum stage. The output el cap is wired with the + pin seeing the negative DC. That causes issues but most never complain.
Old 12th September 2020 | Show parent
  #17
Gear Guru
 
tINY's Avatar
 
1 Review written
🎧 15 years
Quote:
Originally Posted by JohnRoberts ➑️
===

Headroom? I do not follow.

In a lightly loaded DC blocking application the bias voltage is constant and floating on top of the changing AC. There are millions of polar electrolytic capacitors operated successfully with almost zero bias voltage. My 2.5V of bias is an accident of the single supply design I used to KISS.

JR

With the signal level you specified, there will always be at least 0.75v across the cap at any instance. ...assuming a sine wave signal at 1.25Vrms

1.25v * 1.4 = 1.75Vp.

2.5v-1.75v = 0.75v



-tINY

Old 12th September 2020 | Show parent
  #18
Lives for gear
 
JohnRoberts's Avatar
 
🎧 10 years
Quote:
Originally Posted by tINY ➑️

With the signal level you specified, there will always be at least 0.75v across the cap at any instance.
Maybe I wasn't clear... this is a DC blocking capacitor, passing an audio AC waveform. The capacitor will pretty much always have the full 2.5V DC bias across it, except perhaps at extreme LF.

For 0V audio AC initial conditions the top of the cap will be sitting at 2.5V DC and the bottom at 0V DC. At full scale AC (2.5VAC output total or 1.25VAC on each leg) at the audio signal peak excursion the top of the positive leg cap will rise to 2.5V+1.75V=3.25V. I set my overload indicator to fire at 3.3V. The bottom of the cap will be 2.5V DC lower so +1.75V delivering full scale signal. The capacitor has no idea what the PS rail is but the op amps do.
Quote:


...assuming a sine wave signal at 1.25Vrms
yes for 2.5VAC full scale 2x 1.25VAC (opposite polarity legs feeding + and - amplifier inputs).
Quote:

1.25v * 1.4 = 1.75Vp.
yes peak DC voltage excursion
Quote:

2.5v-1.75v = 0.75v
this is describing the negative polarity leg.

The inverted opposite polarity negative leg at signal peak will swing the top of the capacitor in series with - output to +0.75V DC, but the bottom of that cap will be 2.5V lower for the desired -1.75V.
Quote:


-tINY

Sorry if I am making this more confusing than it needs to be.

JR

PS; The buffer is actually two inverting stages in series to make the two differential output legs. 3.5V p-p from a 5V op amp is cutting it close but the modern CMOS op amps can handle that.
Old 13th September 2020
  #19
Gear Guru
 
tINY's Avatar
 
1 Review written
🎧 15 years

There was always a gospel (never tried it on the bench) that polarized caps should have positive voltage applied at all times. For tantalum caps, even a little reverse voltage can damage them very quickly. For the electrolytic types, I thought there was a change in ESR as the DC voltage got close to 0v. I do think they tolerate some reverse voltage.

I am assuming that the source of the signal is grounded to the same ground as the divider....




-tINY

Old 13th September 2020
  #20
Lives for gear
 
lame pseudonym's Avatar
 
🎧 15 years
So I finally got around to reading the whole article all the way through, and the only parameter that the authors measured was harmonic distortion.

Harmonic distortion.

Harmonic distortion.

A parameter that hasn't been relevant in audio for how many decades now?
Old 13th September 2020 | Show parent
  #21
Lives for gear
 
lame pseudonym's Avatar
 
🎧 15 years
Quote:
Originally Posted by tINY ➑️

There was always a gospel (never tried it on the bench) that polarized caps should have positive voltage applied at all times.




-tINY

My first CD player, a Phillips, had the signal-pass electrolytics dc-biased with resistors.

I saw another company that took care to have an equal number of signal-path electrolytics with half signal entering the positive side and half signal entering the negative side. Whether that made a difference who knows but it was a pretty cool thing to do.
Old 13th September 2020 | Show parent
  #22
Lives for gear
 
JohnRoberts's Avatar
 
🎧 10 years
Quote:
Originally Posted by lame pseudonym ➑️
So I finally got around to reading the whole article all the way through, and the only parameter that the authors measured was harmonic distortion.

Harmonic distortion.

Harmonic distortion.

A parameter that hasn't been relevant in audio for how many decades now?
Perhaps because inexpensive op amps have delivered vanishingly low distortion (high linearity) since the mid/late 70s.

Capacitors properly specified and applied don't matter either, IMO. YMMV

JR
Old 13th September 2020 | Show parent
  #23
Lives for gear
 
JohnRoberts's Avatar
 
🎧 10 years
Quote:
Originally Posted by tINY ➑️

There was always a gospel (never tried it on the bench) that polarized caps should have positive voltage applied at all times. For tantalum caps, even a little reverse voltage can damage them very quickly. For the electrolytic types, I thought there was a change in ESR as the DC voltage got close to 0v. I do think they tolerate some reverse voltage.

I am assuming that the source of the signal is grounded to the same ground as the divider....




-tINY

It appears we are talking about completely different circuits.

The DC blocking capacitors in my circuit are relatively lightly loaded so steady state bias of 2.5V will not deteriorate significantly within the audio passband.

JR
Old 13th September 2020 | Show parent
  #24
Lives for gear
 
JohnRoberts's Avatar
 
🎧 10 years
Quote:
Originally Posted by lame pseudonym ➑️
My first CD player, a Phillips, had the signal-pass electrolytics dc-biased with resistors.

I saw another company that took care to have an equal number of signal-path electrolytics with half signal entering the positive side and half signal entering the negative side. Whether that made a difference who knows but it was a pretty cool thing to do.
Coincidentally I had one electrolytic DC blocking cap in my early Phillips (Magnavox) CD player fail open circuit (late 1980s). I replaced it with one I had laying around and it worked for years after that.

There was only one capacitor per channel, AFAIK all such circuits are single ended.

JR
Old 13th September 2020
  #25
Gear Guru
SSL has circuits with DC polarized feed resistors to el caps. I've found them to be completely removable with modern low offset opamps.
Old 17th September 2020 | Show parent
  #26
Gear Addict
 
YashN's Avatar
 
🎧 5 years
Quote:
Originally Posted by Jim Williams ➑️
Soundcraft 6000 and Delta consoles use a negative 2 volt DC bias in their hybrid sum stage. The output el cap is wired with the + pin seeing the negative DC. That causes issues but most never complain.
This is mine. Still don't know whether it counts as a 200 Delta (upper section Mixing module) or something like that or a Delta (rightmost end of hand-rest) 8...

I suppose I should check the internals of the Mixing module?

That paper is a very nice read.
Attached Thumbnails
AES paper on capacitors for audio-sc.jpg  
Old 17th September 2020
  #27
Lives for gear
 
kludgeaudio's Avatar
 
The paper is a reasonable one, but you should read two other important things. First of all, read Doug Self's discussion of capacitor linearity in the Audio Power Amplifier Design Handbook. He describes how using oversized electrolytics can dramatically reduce distortion that is caused by nonlinear series resistance, with some good measurements.

Another solution to the same problem which may be more convenient than oversizing capacitors is to use bypass capacitors. A good discussion of this can be found in Paul Stamler's article in the 5/05 issue of AudioXpress. Same basic concept; trying to reduce ripple across the capacitor at high frequencies.

There are nonlinearities inherent in electrolytics, we know about them, we can deal with them, it's not voodoo and it's not really even difficult to deal with.

And yeah... electrolytics should never come within half a volt or so of the zero crossing and certainly should NEVER have reverse voltage across them at any time.
--scott
Old 18th September 2020 | Show parent
  #28
Gear Addict
 
YashN's Avatar
 
🎧 5 years
Quote:
Originally Posted by kludgeaudio ➑️
Audio Power Amplifier Design Handbook. He describes how using oversized electrolytics can dramatically reduce distortion that is caused by nonlinear series resistance, with some good measurements.
Thought I saw a similar thing in his small signal book. Yes, was very interesting to see. What I saw showed the lowest dist. with 1,000uF (it was the max he measured) in the location of the DC blocking cap IIRC.

In my gear (several pieces of equipment, and not just the mixer, many of the instruments), I have noticed a slew of these annoying small BP/NP caps which bug me.

Actually, those old signal path implementations I am looking at bug me a lot...

There are several potential enhancements one could implement nowadays.
Old 18th September 2020
  #29
Gear Addict
 
YashN's Avatar
 
🎧 5 years
I think it was 2 weeks ago or so when I measured DC offset around an op amp in one of the instruments.

Looked like most of the offset was positive, but it would sometimes drop to a tiny negative value on the DVM.

What do I do in this case?
Old 18th September 2020 | Show parent
  #30
Lives for gear
 
kludgeaudio's Avatar
 
Quote:
Originally Posted by YashN ➑️
I think it was 2 weeks ago or so when I measured DC offset around an op amp in one of the instruments.

Looked like most of the offset was positive, but it would sometimes drop to a tiny negative value on the DVM.

What do I do in this case?
It doesn't hurt anything on that stage; the opamp behaves the same way at the zero crossing as anywhere else because crossover distortion is mostly a non-issue today. It might be an issue with the stage afterward if you have a lot of gain. If it is, use a DC blocking cap that can handle the negative value, like a film cap or a bipolar electrolytic bypassed with a film cap if the value needed is too large to find a film cap off the shelf.
--scott
πŸ“ Reply
Topic:
Post Reply

Welcome to the Gearspace Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…

Forum Jump
Forum Jump