Another definitive advice on capacitor distortion.

[JohnRoberts]

This is a mature topic, years ago forum member Sam Groner did his own research into distortion in electrolytic capacitors.
====

Quadrature Bridge Measures Harmonic Distortion in Capacitors​

This article considers the measurement of harmonic distortion in capacitors using a bridge circuit. Each arm of the bridge consist of one resistor and one capacitor. The bridge is driven in quadrature, such that both the differential and common-mode output voltage are nominally zero. The article introduces a procedure to construct a capacitor with very low distortion contribution, which is used in the reference arm of the bridge. The resolution of the bridge is estimated to exceed −150 dB. Measurement results for various capacitor types are presented. These show that parts with ceramic C0G dielectric have more consistent, and generally speaking also lower, distortion than Polyester, Polypropylene or Polystyrene film capacitors.

Linear Audio, Volume 12, September 2016

This article is co-authored by Scott Wurcer.
====
article may no longer be free...

http://www.nanovolt.ch/publications/index.html

JR

 

[vintageampfixer]

This is a mature topic, years ago forum member Sam Groner did his own research into distortion in electrolytic capacitors.
====

Quadrature Bridge Measures Harmonic Distortion in Capacitors​

This article considers the measurement of harmonic distortion in capacitors using a bridge circuit. Each arm of the bridge consist of one resistor and one capacitor. The bridge is driven in quadrature, such that both the differential and common-mode output voltage are nominally zero. The article introduces a procedure to construct a capacitor with very low distortion contribution, which is used in the reference arm of the bridge. The resolution of the bridge is estimated to exceed −150 dB. Measurement results for various capacitor types are presented. These show that parts with ceramic C0G dielectric have more consistent, and generally speaking also lower, distortion than Polyester, Polypropylene or Polystyrene film capacitors.

Linear Audio, Volume 12, September 2016

This article is co-authored by Scott Wurcer.
====
article may no longer be free...

http://www.nanovolt.ch/publications/index.html

JR

It's not free ... what's their rough synopsis, John?

 

[JohnRoberts]

It's not free ... what's their rough synopsis, John?

IDK, I surely don't remember what they thought.

I do seem to recall that Sam had to craft an unusually low distortion sine wave generator and test rig just to measure it...

JR

 

[Voyager10]

From all this, my conclusion remains that THD from electrolytic coupling capacitors is a non-issue, unless someone can enlighten me and tell me where my reasoning goes wrong.

We do have to worry a little bit about IMD, so a LF signal could generate sum-and-difference frequencies in a more sensitive range. (OTOH auditory masking effects will make it hard to distinguish close-together frequencies, so perhaps not actually more audible).

 

[abbey road d enfer]

The cut-off frequencies of all stages are 0.33Hz or below, so voltage across the coupling caps will be very low, even at 20Hz. At a frequency of 20 Hz and measuring at 10mV, 100mV, 1V and 5.8V (=max), THD never exceeded -90dB. So it's still in the weeds.

I use the same approach as you, very low cut-off frequency, and never had an issue.
OTOH, I'm a subscriber of a group that concern is distortion in IEM, where they try to quantify THD below -130dB. Just about any capacitor ruins their measuremnts.
I'm always amused because I don't think they hear the difference. It's just a digital version of a dick contest.

 

[MidiMania]

I have read the article and the thread. Being a scientist myself I really find this papper annoying. If the author had only concentrated on the distortion characterstics of capacitors it could have become a very interesting study. Aiming to make a point about audibility of distortion and the entire listening test methodology of how to examine that in such a slim paper is close to rediculous.

Someone else here has written this already: In most audio devices there would be many electrolytic caps in series. If they all add together they way it is shown here the outcome would be very bad. Circuit designers therefore aim to design for lowest distortion per component in order to achieve a low distortion for the entire circuitry. So even the conclusion made in the paper is in best case irrelevant.

Worst part is the one about the 'non-linear' components such as op-amps. It is claimed that they add more to the distortion - simply because they are not linear. Thats a claim, not at all an empiric statement. And stated in such a flat way it is simply wrong.

 

[RCAguy]

Key issue: “Due to the increasing sensitivity of the ear to the low-mid harmonics, the 4th and 5th harmonics would probably be the first ones to be audible.” Precisely why lesser quality capacitors distortion is important, as harmonic artifacts once above the threshold are perceived as a higher % relative to the fundamental, that may even still be inaudible. Also applies to subwoofer distortion, where audible harmonic artifacts are post crossover and lead to tearing the omave from main speakers. The science is in “Subwoofer Camp” at issuu.com/filmakertech

 

[jp8]

Though the Subwoofer Camp article clearly explains the effect, I'm sorry to say that I fail to see why "lesser quality capacitors distortion is important" as they are not likely to exceed the allegedly acceptable 1% LF THD from professional subwoofers. This 1% is still 67 dB above the worst case HD that I measured for the 5th harmonic when choosing the cut-off frequency way too high. Admittedly, these were high quality Nichicon audio caps, but a while ago I also measured a piss-poor elcap, which did not differ much from a quality brand.

With the above I don't mean to say that you can just drop in any fancy brand with nicely coloured, gold printed sleeves. You can expect the better brands, especially the Japanese, to have a longer lifetime and closer matching of capacitance and ESR within one batch. The latter could be significant for input coupling caps in mic preamps, as any impedance difference affects CMRR.

Talking about lifetime: I long assumed elcaps drying out when used under normal temperatures was kind of a fairy tale, until ALL 16 or so coupling caps on the FX unit of my Mackie ProFX12 failed and measured less than 10% of the rated value and no signal was passing... All cheap Chinese crap capacitors. And they call it "Pro"...??

Jan

 

[Newmarket]

Talking about lifetime: I long assumed elcaps drying out when used under normal temperatures was kind of a fairy tale, until ALL 16 or so coupling caps on the FX unit of my Mackie ProFX12 failed and measured less than 10% of the rated value and no signal was passing... All cheap Chinese crap capacitors. And they call it "Pro"...??

What was the age of your ProFX12 ? Because tbf all wet electrolytics decay from the off depending greatly on voltage and, critically, temperature. Of course construction and QC plays a significant part. And there are "events" like the "capacitor rash" issue of some years past that affected a lot of PC motherboards.

 

[jp8]

What was the age of your ProFX12 ? Because tbf all wet electrolytics decay from the off depending greatly on voltage and, critically, temperature. Of course construction and QC plays a significant part. And there are "events" like the "capacitor rash" issue of some years past that affected a lot of PC motherboards.

I'm not sure about its age. Perhaps the last digits of the serial refer to manufacturing date, which would lead me to believe it's from August 2006. Would that make sense? See attached typeplate picture.

I bought it as "for spares or repair" with broken PSU, which btw was not caused by elcaps. Replaced that with a DIY PSU. After using it for perhaps a year or so, the FX unit stopped working. The coupling caps on the FX unit are not in any way stressed with high voltages, AC currents or temperatures, yet they were all dried out. AFAIK, the "capacitor rash" you were referring to was causing bulging of the top due to boiling electrolyte under heavy AC load. This was caused by wrong chemistry of the electrolyte. I'm not sure this was the same effect plagueing the elcaps on the FX unit. I assume in my case it was more likely caused by poor quality seals on the bottom side as there are no high AC currents involved.

Jan

 

[Newmarket]

Thanks for the details. 15 or so years might be considered lifetime for elecs - defined by a halving of capacitance (lesson = go BIG). Not so sure on temp stress tbh - older fx modules can dissipate significant heat imo. OTOH I'm sitting next to a A&H GL2 from the 90s and (without measuring) I don't suspect any cap failures. Of course much depends on usage.
Similarly with a Soundcraft Spirit desk I use for input channels and my Soundscape I/O system from the last century (DSP card updated c.2007).
wrt Capacitor Rash - simply cited as a possible exception. But from experience, and also discussion with hardcore PSU designers, thermal stress is a dominant parameter in most cases. Alongside Ripple Current/ ESR in power applications.

 

[jp8]

Assuming the previous owner did not have the ProFX12 powered on all the time, 15 years lifetime would be really short IMHO. FX unit does not run hot and all caps were dried out, not just near the DSP chip. Lifetime of the product should not be limited by some elcaps that are not stressed in any way.

Jan

 

[abbey road d enfer]

. Lifetime of the product should not be limited by some elcaps that are not stressed in any way.

Well, actually, in many products, electrolytic caps are the main life-time reducing factor.
Just consider MTBF.

 

[jp8]

Mind the words "that are not stressed in any way". These should just not fail within the economic lifetime of a mixer.

Elcaps in e.g. PSUs often degrade more quickly, but even there, when applying proper derating, component choice and placement away from hot components, PSU lifetime and MTBF should be such that it outlives electromechanical parts of a mixer in my opinion. Or the mixer should become functionally obsolete before the PSU dies.

AFAIK, the MTBF of (good) elcaps is measured in millions of hours (> 100 years). I'd expect most PSU failures to be caused by wear, not random failures. But as said, wear/lifetime is to a great extent a design choice. But as long as PSUs are considered a closing item on the budget, or designers are making stupid design choices, things will not change to the better.

Jan

 

[CalavoBob]

I've been working on several 1950s era Telefunken v72 mic pres lately and although the blogs about them warn about common failures of the electrolytics, the electrolytics in well over 75% of the units are testing just fine. Those guys knew how to make a cap.

 

[soapfoot]

I've been working on several 1950s era Telefunken v72 mic pres lately and although the blogs about them warn about common failures of the electrolytics, the electrolytics in well over 75% of the units are testing just fine. Those guys knew how to make a cap.

Or perhaps it’s more that they didn’t know quite so much, and so had to be conservative at every level to ensure there weren’t too many failures!

I feel like engineering teams can now target service life and other granular factors with increasing degrees of precision, allowing them to optimize (make components smaller and smaller) while still satisfying minimum longevity requirements (though sometimes just barely!)

I tend to derate massively where I can—when servicing my older gear I generally choose the highest-temp, highest-voltage, longest-life part that will physically fit

Most likely those parts will be better (and last longer) than the originals due to the massive derating

 

[vintagelove]

I've been working on several 1950s era Telefunken v72 mic pres lately and although the blogs about them warn about common failures of the electrolytics, the electrolytics in well over 75% of the units are testing just fine. Those guys knew how to make a cap.

I have a completely stock Telefunken v72, with what testing I had available to me, it still functioned like it was new, and was actually quieter than the pream on the interface I tested it with.

I'm torn because I plan on selling it to fund building a small tube mixer and was considering servicing it prior to selling, but... there's nothing to service.


Sounds great too.

 

[Newmarket]

At risk of being pedantic...we are really considering MTTF rather than MTBF wrt capacitors given that they aren't going to get repaired* so there is no B(etween).
*ignoring any 'reforming' that may be attempted.
Also, orientation may be a factor in self healing of wet electrolytes.

 

[Newmarket]

Had an interesting discussion about reliability with a TDK Lambda bod some years ago. They were moving away from communicating in terms of MTBF since customers often didn't understand it's meaning. eg a PSU might have a quoted MTBF of 25 years (they always seem to be around that !). Customers would then often expect it to go 25 years before a failure. Whereas, statistically, there is a 37% probability of a unit reaching or exceeding the MTBF.

 

[Cqwet Dbdfte]

Had an interesting discussion about reliability with a TDK Lambda bod some years ago. They were moving away from communicating in terms of MTBF since customers often didn't understand it's meaning. eg a PSU might have a quoted MTBF of 25 years (they always seem to be around that !). Customers would then often expect it to go 25 years before a failure. Whereas, statistically, there is a 37% probability of a unit reaching or exceeding the MTBF.

Yes, MTBF is generally misunderstood by most. Calculators exists for such purpose which piles guesstimates upon guesstimates. Too many assumptions lead to useless guesswork.
Years ago I looked into how electrolytics were made, found som 20 different aluminium foils, and the electrolytes are closely guarded secrets in the good brands.
I recently dug into a 30 year old Jadis preamp and was a bit surprised at the effort in the powersupply. Well regulated plus massive amounts of capacitance, like 10 times more than old school designs.