Debunking Linus's Audio Grade Caps

[blueskull]

I know there's a thread on this matter. Just so you know.

I'd like to run an experiment on this matter using some typical configurations (caps as HPF, caps as LPF, caps as power decoupling, etc.).

I just shipped some of my stuff to China as I'm moving back to China, so I won't be able to test them now.

But I will make the final move by the end of November, so in a month I can do the test.

In the meantime, post your ideas and caps you'd like to test. I'll test them in the near future.

I'd appreciate if someone has some really niche ones to test (like some NOS or fancy ones), as long as you can ship them to me before I leave. PM for me address.

BTW, if you have an amp or something that you'd like to test, I'll be happy to test.

Gear: Keysight M9260A with new, binned DAC, -122dB THD on best channel combination.

[Black Phoenix]

This is a big list:

https://www.hificollective.co.uk/components/capacitors.html

But the main ones audiophilies say that they can ear differences in sound is the  Elna Silk, WIMA MKP/MKS, Panasonic EZP-E and Nichicon Muse.

[Ysjoelfir]

You should definitely test some of the "good old" Paper in Oil Caps, like those nice russian military style ones you get over ebay for relatively cheap. I may have some of those, I can't exactly remember where they went. Also, vintage caps in generell could be interesting. I definitely have some of those and potentially also some "audiophile grade" caps, pre aged of course.
Why not try some vaccum caps? Could be interesting

I'm not sure if shipping the stuff from germany to you now would be cheaper than shipping it to china. Would have to look that up...

[rsjsouza]

This is a big list:

https://www.hificollective.co.uk/components/capacitors.html

But the main ones audiophilies say that they can ear differences in sound is the  Elna Silk, WIMA MKP/MKS, Panasonic EZP-E and Nichicon Muse.

Holey moley! I had absolutely no idea how many manufacturers and/or OEMs of capacitors existed. I have several orange dipped and used to have several "Mullard" and polystyrene capacitors - could have made some dough there (£0.78 for a single orange capacitor?)

[exe]

I'd test these two scenarios:

1. "electrolytic cannot be used for decoupling or anywhere in signal path".
2. Same about MLCC

These are just for curiosity only, I don't think audiophiles accept numbers as proof of anything .

PS did I mention that I tested "audo-grade" panasonic aluminum caps? They seem to have a slightly lower ESR, and less inductance comparing to normal.

[nctnico]

2. Same about MLCC

MLCC capacitors with cheap dielectrics (for example: X... Y....) are known for distorting audio because their capacitance varies with the voltage across them.

[Gyro]

You should definitely test some of the "good old" Paper in Oil Caps, like those nice russian military style ones you get over ebay for relatively cheap. I may have some of those, I can't exactly remember where they went. Also, vintage caps in generell could be interesting. I definitely have some of those and potentially also some "audiophile grade" caps, pre aged of course.
Why not try some vaccum caps? Could be interesting

I'm not sure if shipping the stuff from germany to you now would be cheaper than shipping it to china. Would have to look that up...

I have to admit ( ) that Soviet military PIOs sound much better in Speaker crossovers than MKP/MKS film types. There's something very clean about the midrange and treble (no 'tizz' or sibilance, just clear transparency on cymbals etc.) it's night and day. They're very cheap too - especially compared to the outrageously expensive leaky non-hermetic Audiophool type crap.

I'm no believer in 'magical properties' in straightforward audio signal level coupling applications where they are just taking out a DC offset, but in speaker crossover filters, where there is significant current involved, PIOs do seem to make a positive difference. Maybe it's as simple as internal mechanical damping or lack of piezoelectric effects (film caps can sing very audibly in some applications). I haven't bothered to go into any detailed analysis though.

[TimFox]

2. Same about MLCC

MLCC capacitors with cheap dielectrics (for example: X... Y....) are known for distorting audio because their capacitance varies with the voltage across them.

The good mlcc caps are C0G/NP0.

[hamster_nz]

2. Same about MLCC

MLCC capacitors with cheap dielectrics (for example: X... Y....) are known for distorting audio because their capacitance varies with the voltage across them.

Agreed,, but how much does the voltage over a coupling cap vary anyway?

At worse, with a lot of deep bass, it will be maybe a fraction of a volt (AC), but it could of course be a constant few volts DC at the same time.

[nctnico]

2. Same about MLCC

MLCC capacitors with cheap dielectrics (for example: X... Y....) are known for distorting audio because their capacitance varies with the voltage across them.

Agreed,, but how much does the voltage over a coupling cap vary anyway?

At worse, with a lot of deep bass, it will be maybe a fraction of a volt (AC), but it could of course be a constant few volts DC at the same time.

Well it could be a couple of volt AC. Also think about capacitors in equalisers and bass/treble circuits.

The good mlcc caps are C0G/NP0.

But these don't come in large capacitances for a reasonable price. I used to use a lot of polyester film capacitors in audio circuits.

[mzzj]

At worse, with a lot of deep bass, it will be maybe a fraction of a volt (AC), but it could of course be a constant few volts DC at the same time.

I wouldn't call it coupling capacitor anymore in that case, more like high pass.
Filter circuits need reasonably good caps or the corner frequency has to be so far from audio band that its practically just a coupling cap.
Al-electrolytic makes lousy dc blocking capacitor if the -3db is at 20hz but if you set the LF cut-off to something like 0.2hz the distortion is minuscule.

[exe]

2. Same about MLCC

MLCC capacitors with cheap dielectrics (for example: X... Y....) are known for distorting audio because their capacitance varies with the voltage across them.

Interesting, I'd like to see numbers. My understanding is that change in capacitance doesn't happen that fast. It's more about DC bias. I wouldn't use Y-caps anyway, they have too much variance. But X... caps are not born equal. Those that have larger size and higher voltage are better than small ones. So, I'd like to see a decent X7R in a signal path.

Al-electrolytic makes lousy dc blocking capacitor if the -3db is at 20hz but if you set the LF cut-off to something like 0.2hz the distortion is minuscule.

Why so?

[nctnico]

2. Same about MLCC

MLCC capacitors with cheap dielectrics (for example: X... Y....) are known for distorting audio because their capacitance varies with the voltage across them.

Interesting, I'd like to see numbers. My understanding is that change in capacitance doesn't happen that fast. It's more about DC bias. I wouldn't use Y-caps anyway, they have too much variance. But X... caps are not born equal. Those that have larger size and higher voltage are better than small ones. So, I'd like to see a decent X7R in a signal path.

Some Google-fu
https://www.edn.com/design/analog/4416466/Signal-distortion-from-high-K-ceramic-capacitors

[tooki]

This is a big list:

https://www.hificollective.co.uk/components/capacitors.html

But the main ones audiophilies say that they can ear differences in sound is the  Elna Silk, WIMA MKP/MKS, Panasonic EZP-E and Nichicon Muse.

Holey moley! I had absolutely no idea how many manufacturers and/or OEMs of capacitors existed. I have several orange dipped and used to have several "Mullard" and polystyrene capacitors - could have made some dough there (£0.78 for a single orange capacitor?)

Well, some of those are defunct brands, so just selling old stock of brands that have since been subsumed into other brands. And I’m sure many of the “audiophile” brands are just rebadges of other brands.

[ConKbot]

Interesting, I'd like to see numbers. My understanding is that change in capacitance doesn't happen that fast.

Depends, do you call about -25% at 4V bias fast?https://product.tdk.com/en/search/capacitor/ceramic/mlcc/info?part_no=C1608X7R1A225K080AC
And that's not even cherrypicking the worst of the worst. Some 0402 and smaller caps are at -80 percent by 6V or so.

[rsjsouza]

This is a big list:

https://www.hificollective.co.uk/components/capacitors.html

But the main ones audiophilies say that they can ear differences in sound is the  Elna Silk, WIMA MKP/MKS, Panasonic EZP-E and Nichicon Muse.

Holey moley! I had absolutely no idea how many manufacturers and/or OEMs of capacitors existed. I have several orange dipped and used to have several "Mullard" and polystyrene capacitors - could have made some dough there (£0.78 for a single orange capacitor?)

Well, some of those are defunct brands, so just selling old stock of brands that have since been subsumed into other brands. And I’m sure many of the “audiophile” brands are just rebadges of other brands.

Hehehehe... Most of mine are 1980's Siemens and Philips capacitors... Quite vintage!

[dzseki]

2. Same about MLCC

MLCC capacitors with cheap dielectrics (for example: X... Y....) are known for distorting audio because their capacitance varies with the voltage across them.

The good mlcc caps are C0G/NP0.

While true, you quickly find yourself that decoupling caps on amplifier inputs are often in the range of several uF, then an "audiophile" capacitor can come out cheaper than stacking 1uF C0G capacitors, how's that? 

[exe]

Depends, do you call about -25% at 4V bias fast?https://product.tdk.com/en/search/capacitor/ceramic/mlcc/info?part_no=C1608X7R1A225K080AC
And that's not even cherrypicking the worst of the worst. Some 0402 and smaller caps are at -80 percent by 6V or so.

I meant it takes time for dielectric to polarize. So, I'd expect that peak AC value doesn't contribute much to distortion, but, I can be wrong. That's why I wanted to see some data.

The second thing is, when people want to show how bad DC-bias is for caps, they use a crappy cap and show data. That's not how I design my circuits. I start with what my supplier has in stock, then pick the best according to simulators provided by kemet, murata, etc. That's why I would like to see a decent X7R cap of reasonable size (0805 and bigger). It all goes down to "volumetric capacitance".

Some Google-fu
https://www.edn.com/design/analog/4416466/Signal-distortion-from-high-K-ceramic-capacitors

Thanks, that's interesting. But I didn't get if they used real caps, or just used simulator. The result is quite unexpected to me. Or, may be not, as, -70dbV is quite a small value after alll, so I can relate that to DC bias. Still, would love to see more data. People in comments there say that setup is sub-optimal. I'd also say 10nF is unusually small value for AC-coupling.

[dmills]

Douglas Self has some stuff on this in "The design of active crossovers", and of course C. Batemans work is still good.

Regards, Dan.

[mzzj]

Al-electrolytic makes lousy dc blocking capacitor if the -3db is at 20hz but if you set the LF cut-off to something like 0.2hz the distortion is minuscule.

Why so?

Al electrolytics are non-linear so they distort the signal if there is considerable voltage swing over them. Used strictly as a bypass capacitor or high-pass with corner frequency way below audio frequencies ensures that they don't have much any voltage swing/variation over them and  don't cause unwanted distortion.
(distortion might well be still way lower what is considered audible but something like 100x worse than otherwise good amplifier specification)
 

Douglas Self's "Audio Power Amplifier Design Handbook " is one of the best and covers/debunks lots of audiophoolery:
https://www.amazon.com/gp/product/B01FIWBNA0/ref=dbs_a_def_rwt_hsch_vapi_taft_p1_i3

[mzzj]

Some of the material is also available on Self's website:

http://www.douglas-self.com/ampins/dipa/dipa.htm#2

chapter 5.8 specially