Rectification in aluminium electrolytic capacitors

I've just been having a discussion elsewhere about the performance of aluminium electrolytic caps for audio coupling in a zero bias situation. This is a method very widely used nowadays.

I've observed, indeed I first heard of it decades back, that provided the signal

*across* the capacitor doesn't exceed about 100mV ac there is no measureable non-linearity. I've confirmed this using Audio Preciosn test gear which measures down to 0.0008% THD (-102dB). Hence if the value of the cap is such that 100mV is never exceeded, there will be no distortion contribution from the cap. I've also confirmed that if the 100mV is exceeded, you do indeed get distortion.

The mechanism seems to be electrolytic rectification AIUI. I presume that the absence of any effect below ~ 100mV is due to the 'forward' voltage of this rectifier, and current only flows when it's exceeded.

Any more thought on this ?

I did find this page about electrolytic rectification btw.

formatting link

The oscillograms are rather interesting.

Graham

Reply to
Eeyore
Loading thread data ...

Eeyore wrote in news: snipped-for-privacy@hotmail.com:

Was this tested with non-polarized electrolytic capacitors as well? Any of the more esoteric capacitors like Black Gate (graphite mixed with the electrolyte) or polymer-based caps?

It would seem to confirm that electrolytic caps should ideally be biased.

--Damon

Reply to
Damon Hill

I did test one example of a non-polarised type. Interestingly, once that 100mV threshold had been exceeded it produced twice as much distortion (at low signal levels) as a polarised type. It would seem that's simply because it is internally 2 such caps back to back. I've therefore concluded that for 'low level' signal coupling the NP types offer no advantage and in fact degrade the signal more.

No, I haven't tested any more esoteric types. The last I heard, Sanyo ? were stopping manufacture of the Black Gates.

Ideally. I have used them 'back to back' with the junction polarised to rail via a very high value R. It's cumbersome though.

The zero bias method works well provided you don't exceed ~ 100mV across the cap. A value of 100uF deals with this in most real world audio circuits.

Graham

Reply to
Eeyore

Eeyore wrote in news: snipped-for-privacy@hotmail.com:

Rubycon, actually. They manufacture Black Gates for a third party, and a new agreement was negiotiated recently. For better or worse, that brand will continue for a while. I use them, sparingly, for input stage bypassing and feedback circuits.

The chemistry of electrolytics has always seemed like a black art and the Black Gate web page itself doesn't help things a bit. I can't blame professional engineers for being skeptical, though even some major audio manufacturers use them.

formatting link

...if you'd like to take on their bafflegab.

--Damon

Reply to
Damon Hill

aluminium

method

signal

measureable

measures

100mV

I've

distortion.

the

this

Hey Graham

I just want to clarify a thing or 2 if its OK.

By no signal across the cap, you are referring to AC, correct? If that's the case, aluminum electros would be the cap of choice for a mic pre where the 48v from phantom power would need to be blocked, right? But only Aluminum electros, or do other electros also exhibit this trait?

Thanks

Reply to
tempus fugit

signal

measures

I'm no electrochemist, but perhaps it takes something in excess of 100 mV or so of reverse polarity to begin the reduction reaction that damages the oxide layer.

If the peak voltage is kept low, no oxidation or reduction occurs and the cap is linear.

Did I guess right?

--
Paul Hovnanian     mailto:Paul@Hovnanian.com
------------------------------------------------------------------
 Click to see the full signature
Reply to
Paul Hovnanian P.E.

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.