Coupling caps

[wyn palmer]

Yeah, except I've never heard anyone who liked odd-order harmonics.

Transistor harmonics are odd, tubes are even. We likely do not mind even order because the ear canal generates even-order harmonic distortion, as do natural environments like forests, and the brain readily removes even order, so it's a more natural sound, whatever way one interprets that. Even-order sounds warm and relaxed, like a whiskey aged in a burnt-oak barrel forgotten for a century in a broken down barn and allowed to slow age in the seasons. Odd-order is that grating sound of nails on a blackboard that makes one want to emulate Godzilla in Tokyo and smash the amplifier into rubble.

The ideal amplifier, as Harman-Kardon put in one of its ads, is a straight wire with gain. That is, it does not alter the input signal other than to make it larger.

Well, how about the third harmonic component generated by saturation in analog tape recorders? Driving biased tape more into saturation is reputed to add a punchiness to low frequency information in a way that is certainly preferred by some. In this case it adds a tone that is an octave and a half above the fundamental and that is considered a good thing in certain instances and contributes to the "tape sound".
Incidentally, I am well aware of the psycho acoustic issues that you describe, but they are hardly a complete set.
In fact, plug ins and hardware exist to add controlled second and third order harmonics (and even higher order ones) to recordings in order to duplicate the "analog tape sound" and other effects in both digital and analog environments.
By the way, unless I missed something, I'm still awaiting a reference for the rectification-effect generated distortion in electrolytic caps- restricting it to bipolar isn't necessary, but I do wish for it to be restricted to ALoxide types. I can find references for rectification in monomolecular films involving Al oxide, and that makes sense from a quantum mechanical point of view, but I cannot find a reference that involves bulk oxide layers, nor can I find any references that describe any changes in non linearity associated with changes in the polarity of the voltage across the device as long as the reverse polarity is restricted to less than 1.5v or so, and at that point the devices are approaching failure.
I would welcome the enlightenment as clearly I am harboring an illusion.
If you have already provided this information, I apologize.

 

[Pio1980]

Well, how about the third harmonic component generated by saturation in analog tape recorders? Driving biased tape more into saturation is reputed to add a punchiness to low frequency information in a way that is certainly preferred by some. In this case it adds a tone that is an octave and a half above the fundamental and that is considered a good thing in certain instances and contributes to the "tape sound".
Incidentally, I am well aware of the psycho acoustic issues that you describe, but they are hardly a complete set.
In fact, plug ins and hardware exist to add controlled second and third order harmonics (and even higher order ones) to recordings in order to duplicate the "analog tape sound" and other effects in both digital and analog environments.
By the way, unless I missed something, I'm still awaiting a reference for the rectification-effect generated distortion in electrolytic caps- restricting it to bipolar isn't necessary, but I do wish for it to be restricted to ALoxide types. I can find references for rectification in monomolecular films involving Al oxide, and that makes sense from a quantum mechanical point of view, but I cannot find a reference that involves bulk oxide layers, nor can I find any references that describe any changes in non linearity or other associated with changes in the polarity of the voltage across the device as long as the reverse polarity is restricted to less than 1.5v or so, and at that point the devices are approaching failure.
I would welcome the enlightenment as clearly I am harboring an illusion.
If you have already provided this information, I apologize.

Something else that has brought me to think that straight clean digital audio technology may be the most trustworthy for true fidelity. I'm in debt to Ken Kantor for bringing me to this notion.

 

[wyn palmer]

Yeah, except I've never heard anyone who liked odd-order harmonics.

Transistor harmonics are odd, tubes are even. We likely do not mind even order because the ear canal generates even-order harmonic distortion, as do natural environments like forests, and the brain readily removes even order, so it's a more natural sound, whatever way one interprets that. Even-order sounds warm and relaxed, like a whiskey aged in a burnt-oak barrel forgotten for a century in a broken down barn and allowed to slow age in the seasons. Odd-order is that grating sound of nails on a blackboard that makes one want to emulate Godzilla in Tokyo and smash the amplifier into rubble.

The ideal amplifier, as Harman-Kardon put in one of its ads, is a straight wire with gain. That is, it does not alter the input signal other than to make it larger.

Oh, and also by the way. Have you ever heard of "the translinear principle"? Essentially it employs the exponential nature of the p-n junction to do things like performing extremely low distortion amplification in the current mode in which bipolars work best or direct computation of functions like RMS AC to DC conversion.
You also can generate second order distortion in a bipolar differential pair by creating an asymmetry, for instance by modifying the action of the tail current source to add an additional inverse tanh component to the transfer function, or simply by mismatching the input devices.
So, bipolars don't just create odd order harmonics, which is certainly the case for an ideal, perfectly matched, perfectly biased differential pair, but can also be persuaded to do all sorts of surprising things...

 

[wyn palmer]

Something else that has brought me to think that straight clean digital audio technology may be the most trustworthy for true fidelity. I'm in debt to Ken Kantor for bringing me to this notion.

True, but the idea of perfect fidelity to the source disappeared behind a veil of reality a long time ago. There is no perfect transducer, either for recording of non-electronic music or for playback. There is also no perfect room, and recordings of live music generally don't sound real because the recording (generally near field, odd aspect etc.) and listening (far field, fixed aspect etc.) environments and conditions are disparate.
Much of what we hear- such as the 3 dimensional sound field- in my experience does not exist in the real life listening experience and is essentially illusory.
Basically, we're producing a musical instrument here, tuned to our individual tastes and desires, but with at least a few concessions offered to non-subjective reality...

 

[Retrovert]

Well, how about the third harmonic component generated by saturation in analog tape recorders? Driving biased tape more into saturation is reputed to add a punchiness to low frequency information in a way that is certainly preferred by some..

These and other recent questions posed to me are way off-topic for the discussion as to the merits of various coupling capacitors, so I'm not going to here respond. Others may, of course, do as they like. I suggest you start another thread, probably best to do that in Tape, where all the tapeheads congregate. But, again, that's only a suggestion so that others interested in the topic may contribute.

 

[wyn palmer]

These and other recent questions are way off-topic for the discussion as to the merits of various coupling capacitors, so I'm not going to here respond.

I suggest you start another thread, probably best to do that in General Audio.

Sure, do you want to participate? Continuing the badinage could be fun...

 

[Retrovert]

Oh, given the fidelity differences compared to digital audio I am very much done with tape, except for nostalgia. No more 3HD (ambiguously written as THD) for me, and I am glad I no longer need to store music on deteriorating and disintegrating cassettes, or use dbx or Dolby filters, for that matter.

I have given away a fair bit of my old (and fancy) cassette gear to my punk and metal friends, including my fancy dubbing and high-speed duplication deck, because it was just taking up space, and they'll actually play tapes on it and maybe make new ones. I don't need to be buried with that gear, including like the Collyer Brothers.

But that's a topic for another discussion, and I'm sure the tapeheads will happily debate this issue with you. I would suggest the tape forum as the best place.

 

[gslikker]

Playing a lot using my huntron tracker which I got very cheap, I was looking at a lot of capacitors.
There was one type and value of drop shape tantalum capacitors showing some resistive behavior and apparently by ABUSING them I could find one actually behaving quite odd. As said, using a 10 volts t/t AC signal, 50 ohms series resistance, this can be called abuse.
Note: this odd behavior appeared at ONE type of tantalum at ONE value only. They came from some NOS stuff I bought and this very brand/type/value probably was from a bad batch. I took the worst example.

See my tinkering (BTW I am not an EE so this is hobbyist level not pretending anything else than "tinkering"). As the caps were 10 microfarads, on the huntron this did limit the possibilities for its settings limitations to make something visible. Now, the bad one showed some weird behavior which needs some additional "mechanism" other than just being partly resistive.

http://members.quicknet.nl/gerard.slikker/badtantalum.htm

 

[wyn palmer]

These and other recent questions posed to me are way off-topic for the discussion as to the merits of various coupling capacitors, so I'm not going to here respond. Others may, of course, do as they like. I suggest you start another thread, probably best to do that in Tape, where all the tapeheads congregate. But, again, that's only a suggestion so that others interested in the topic may contribute.

Please don't misunderstand, I really am interested in the answer to the question. I see comments concerning the "rectification" in various locations, including a manufacturers notes, but, for example, the NIchicon Muse ES bipolar has almost identical dissipation factor to the Muse EKZ polarized cap, and the distortion measurements shown here:
https://www.diyaudio.com/forums/par...ipolar-caps-measured-120db-thd-140db-imd.html
Are quite exemplary for AC signals of, I believe, 500mvRMS, and not what I would expect if a true rectification was taking place.
Nowhere does Nichicon say that the oxide or film structure of the bipolar caps is different from that of its normal polarized caps and I can see no references in the literature to support the idea that there is an alternative structure. What am I missing? I can model the series caps with diodes in parallel and it demonstrates the distortion for higher AC voltages. Is this what you mean by "rectification"?
If so, perhaps the answer to the original question is, you can indeed build a non-polar cap out of two series caps, but you'd better keep the AC level at or below about 0.5v RMS!

 

[restorer-john]

Have you seen DocFetcher? GPL and free.

No I haven't seen that program.

I use an older version of Acrobat Pro (9.0) on my machines as it still works from XP all the way through to my Win10 machines, 32 or 64 bit add-ons.

It gives me results like this:



With Adobe now charging $200+ per year, I will be sticking with my old full versions indefinitely.

 

[wyn palmer]

Is it acceptable to use 2 polar caps tied together at negative and again at the positive to create a bi-polar cap for a capacitor coupled application? I would just order a bipolar, but getting them in larger uf seems hard to find?

So, as Retrovert said, two electroytics in series (not parallel) with the connects +-, -+ can be used, but each individual cap looks like a cap with a parallel diode connected with the anode (+) at the negative terminal of the cap and the cathode (-) at the positive terminal of the cap. The diode potential seems to be about and perhaps a bit less than that of silicon- so the diode conducts significantly at voltages greater than 0.5v or so and less so at lower voltages and at about 0.4v across the diode there is essentially no effect.
In normal DC biased operation this diode is reverse biased so it does not conduct appreciably and actually acts like a Zener- hence the voltage breakdown limitation of the cap.
When the AC signal causes the voltage across the caps to be reversed so that the -ve terminal of a cap is now positive nothing changes until the voltage exceeds about 2x the 0.4v level- at which point one of the diodes turns on and the distortion rises abruptly. This translates out to about 0.5v RMS AC.
However, if the composite cap is used as a coupling cap then the situation gets more complicated. At frequencies above the low frequency roll off point the voltage across the cap falls as frequency rises so the distortion will also potentially fall, while at frequencies approaching the roll off point the distortion will rise, then eventually fall again as the frequency moves below the roll off point. Also, when a diode is conducting, the cap it is across is bypassed and so the second cap is the only cap that the AC signal sees and the effective cap seems to increase!
The distortion seems to be odd order (I haven't derived the transfer function and performed the power series expansion or run a distortion analysis simulation) so lower than roll off frequencies could produce tones above roll off which might produce the illusion of enhanced bass, for example.
Premature failure of the caps is not an issue because the diode limits the voltage across the reverse biased cap so that the other cap sustains the remainder of the voltage and hence the 1.5v limit before failure, is not reached.

 

[Pio1980]

So, as Retrovert said, two electroytics in series (not parallel) with the connects +-, -+ can be used, but each individual cap looks like a cap with a parallel diode connected with the anode (+) at the negative terminal of the cap and the cathode (-) at the positive terminal of the cap. The diode potential seems to be about and perhaps a bit less than that of silicon- so the diode conducts significantly at voltages greater than 0.5v or so and less so at lower voltages and at about 0.4v across the diode there is essentially no effect.
In normal DC biased operation this diode is reverse biased so it does not conduct appreciably and actually acts like a Zener- hence the voltage breakdown limitation of the cap.
When the AC signal causes the voltage across the caps to be reversed so that the -ve terminal of a cap is now positive nothing changes until the voltage exceeds about 2x the 0.4v level- at which point the distortion rises abruptly. This translates out to about 0.5v RMS AC.
However, if the composite cap is used as a coupling cap then the situation gets more complicated. At frequencies above the low frequency roll off point the voltage across the cap falls as frequency rises so the distortion will also potentially fall, while at frequencies approaching the roll off point the distortion will rise, then eventually fall again as the frequency moves below the roll off point. Also, when a diode is conducting the cap it is across is bypassed and so the second cap is the only cap that the AC signal sees and the effective cap seems to increase!
The distortion seems to be odd order (I haven't derived the transfer function and performed the power series expansion or run a distortion analysis simulation) so lower than roll off frequencies could produce tones above roll off which might produce the illusion of enhanced bass, for example.
Premature failure of the caps is not an issue because the diode limits the voltage across the reverse biased cap so that the other cap sustains the remainder of the voltage and the 1.5v limit before failure is not reached.

Great analysis of something I've wondered. Thank you for the information!

 

[gslikker]

So it just did the following experiment.
I took the huntron and put a 5kHz signal 10Vtt (7 Volts AC) having a series resistance (according the Huntron Tracker) of 50 Ohms. (so it acts as a tone generator)

I connected to two series 1000 µF caps having minus connected together. NO DC develops over each cap connecting the huntron to the + sides of this "bipolar cap"
I connected a bypass diode over each cap (anode to minuses of the caps) enabling both caps to build up a voltage.
I connected the Huntron for half a minute or so.
Now, both caps go to 0.7 Volts DC. So actually, they build up DC.

Of course, removing the diodes voltage over the each cap stay 0.7 Volts, as they are precharged.
Connecting the source again it stays 0.7 Volts, although the diodes are removed..
Replacing single diodes by two diodes in series, nothing happens. So impedance of the source in series with the two caps apparently is NOT able to let the voltage over about 1.4 Volts on one of them to have the diodes conduct.

Now, as the story suggests rectifying at around a voltage 0.4 Volts if no diode present, one would expect the caps charging up to 0.4 Volts maximum from being empty, which does not happen at all.

I see no rectification effect by some imaginary parallel diode function.

What I am trying to say is, that the very moment internal charging of the cap actually happens, the result will be a DC voltage. This DC voltage acts as a precharge, which stays until leakage of the caps take care of it which could be hours at a 1000 microfarad cap. (edit: of course this applies to both caps, and I initially tried 20 Hz but this lead to uneven charges all the time, for example 0.9 volts on one and 0.5 on the other, apparently result of how the current went at moment of removing a huntron probe. this effect was gone using the maximum being 5 kHz for huntron voltage/ohms setting)

I get confused by the fact that feeding a (for example several Volts) source voltage AC through a cap does not CREATE any voltage OVER the cap in case the cap would be ideal and big enough versus the source and destination impedance, so no rectification should occur.

Still trying to figure out but it makes me confused. No time avail for more experimenting at the moment..

 

[wyn palmer]

So it just did the following experiment.
I took the huntron and put a 5kHz signal 10Vtt (7 Volts AC) having a series resistance (according the Huntron Tracker) of 50 Ohms. (so it acts as a tone generator)

I connected to two series 1000 µF caps having minus connected together. NO DC develops over each cap connecting the huntron to the + sides of this "bipolar cap"
I connected a bypass diode over each cap (anode to minuses of the caps) enabling both caps to build up a voltage.
I connected the Huntron for half a minute or so.
Now, both caps go to 0.7 Volts DC. So actually, they build up DC.

Of course, removing the diodes voltage over the each cap stay 0.7 Volts, as they are precharged.
Connecting the source again it stays 0.7 Volts, although the diodes are removed..
Replacing single diodes by two diodes in series, nothing happens. So impedance of the source in series with the two caps apparently is NOT able to let the voltage over about 1.4 Volts on one of them to have the diodes conduct.

Now, as the story suggests rectifying at around a voltage 0.4 Volts if no diode present, one would expect the caps charging up to 0.4 Volts maximum from being empty, which does not happen at all.

I see no rectification effect by some imaginary parallel diode function.

What I am trying to say is, that the very moment internal charging of the cap actually happens, the result will be a DC voltage. This DC voltage acts as a precharge, which stays until leakage of the caps take care of it which could be hours at a 1000 microfarad cap.

I get confused by the fact that feeding a (for example several Volts) source voltage AC through a cap does not CREATE any voltage OVER the cap in case the cap would be ideal and big enough versus the source and destination impedance, so no rectification should occur.

Still trying to figure out but it makes me confused. No time avail for more experimenting at the moment..

Thank you so much. That's a very interesting test. When you add the diodes it is exactly what the simulations say should happen.
It's hardly surprising that this subject seems to create so much confusion amongst everyone- myself included.
As I said, I am completely unable to find any technical papers directly describing the effect, but there is quite a bit of anecdotal "evidence" and hand waving explanations out there and the distortion measurements that support the hypothesis, or at least something akin to it, are real, at least for the caps that were tested.
Not being able to develop 1.4v dc with two diodes is perhaps not surprising as the leakage through the caps at that point is huge as they are approaching failure, so you can't store charge on the - plate.
The 0.5v rectifier may also be a very "soft" one with a very large series resistance and with a relatively large reverse leakage associated with it when it conducts so that charge is not stored on the common terminals, especially when the cap is excited with a relatively high frequency. It might still increase the distortion appreciably. It's really hard to come up with a suitable model with so little real information.
I'll continue to search. Someone has to have actually analyzed this stuff somewhere.

 

[wyn palmer]

Thank you so much. That's a very interesting test. When you add the diodes it is exactly what the simulations say should happen.
It's hardly surprising that this subject seems to create so much confusion amongst everyone- myself included.
As I said, I am completely unable to find any technical papers directly describing the effect, but there is quite a bit of anecdotal "evidence" and hand waving explanations out there and the distortion measurements that support the hypothesis, or at least something akin to it, are real, at least for the caps that were tested.
Not being able to develop 1.4v dc with two diodes is perhaps not surprising as the leakage through the caps at that point is huge as they are approaching failure, so you can't store charge on the - plate.
The 0.5v rectifier may also be a very "soft" one with a very large series resistance and with a relatively large reverse leakage associated with it when it conducts so that charge is not stored on the common terminals, especially when the cap is excited with a relatively high frequency. It might still increase the distortion appreciably. It's really hard to come up with a suitable model with so little real information.
I'll continue to search. Someone has to have actually analyzed this stuff somewhere.

I pulled up the 1945 book which is seemingly highly regarded wrt the design of electrolytic caps. Whether this is true or not for modern caps remains to be seen, but I don't see why not.
"The film is permeable to free electrons but substantially impermeable to ions, provided the temperature of the cell is not high. When the metal underlying the film is at a negative potential, free electrons are available in this electrode and the current flows through the film of the cell. With the polarity reversed, the electrolyte is subjected to the negative potential, but as there are only ions and no free electrons in the electrolyte the current is blocked"
https://archive.org/details/TheElectrolyticCapacitor/page/n43?q=negative+potential,+free
Page 32 etc.
So, when the cap is reversed biased leakage current will flow which is a non-linear function of the applied negative voltage and also seemingly increases exponentially with temperature, so it sort of behaves like a semiconductor diode, although it actually isn't.
https://electronics.stackexchange.c...tic-capacitor-out-of-two-regular-electrolytic
The above thread has an extensive evaluation of the matter at hand, including citations from manufacturers and a reference to the Georgiev book.
"
MANUFACTURER'S EXAMPLE:

In this document Application Guide, Aluminum Electrolytic Capacitors bY Cornell Dubilier, a competent and respected capacitor manufacturer it says (on age 2.183 & 2.184)

• If two, same-value, aluminum electrolytic capacitors are connected in series, back-to-back with the positive terminals or the negative terminals connected, the resulting single capacitor is a non-polar capacitor with half the capacitance.
  
  The two capacitors rectify the applied voltage and act as if they had been bypassed by diodes.
  
  When voltage is applied, the correct-polarity capacitor gets the full voltage.
  
  In non-polar aluminum electrolytic capacitors and motor-start aluminum electrolytic capacitors a second anode foil substitutes for the cathode foil to achieve a non-polar capacitor in a single case.
  

Of relevance to understanding the overall action is this comment from page 2.183.

• While it may appear that the capacitance is between the two foils, actually the capacitance is between the anode foil and the electrolyte.
  
  The positive plate is the anode foil;
  
  the dielectric is the insulating aluminum oxide on the anode foil; "
  

It does appear that the effect is real and the explanation in post #91 is valid to a first order- it's just that a simple semiconductor diode/cap combo is not an adequate model.
Incidentally a - +,+ - arrangement is as good as a + -, - + one.

 

[gslikker]

I just charged a used 500 uF/25V capacitor to 5 Volts the wrong way around, let it form-wrong-way-around 15 minutes or so having a 100k resistor in series, lower its voltage to 4 volts and disconnected.
I just went upstairs, the negative voltage is down to 2.3 and pace of negative voltage going down is now about ONE millivolts per second.

So, the "internal diode" does not "conduct" that much.....

 

[wyn palmer]

I just charged a used 500 uF/25V capacitor to 5 Volts the wrong way around, let it form-wrong-way-around 15 minutes or so having a 100k resistor in series, lower its voltage to 4 volts and disconnected.
I just went upstairs, the negative voltage is down to 2.3 and pace of negative voltage going down is now about ONE millivolts per second.

So, the "internal diode" does not "conduct" that much.....

Apparently not. Thank you.

 

[gslikker]

I think another day I may do more of this type of part abuse
Just checked again the cap, negative voltage is 1.4 and 1mV down baout 3 seconds.
Took off the meter lead and will see if anything left tomorrow.

 

[Retrovert]

I started to put together a writeup on this a few days ago in response to Wyn's statement, and it started to get out of hand, because nothing is ever simple, particularly with oxides, and I must do some real work this weekend.

Some brief comments, not much edited.

The dielectric is constructed by electrochemistry, essentially anodizing (selectively oxidizing or corroding) the aluminum surface, and the result therefore consists of various oxidation states of aluminum oxide. The main oxide of interest is Al2O3. The semiconductor properties are, more or less, that when Al2O3 has a surplus of aluminum it is an N-type semiconductor and when it has a surplus of oxygen it is a P-type semiconductor. Like other semiconductors, creating a junction of metal oxide and metal creates a rectifier, i.e. unidirectional current flow, often with minor leakage current in the reverse direction, since the metal oxides are not perfect by way of the forming process.

Aluminum oxide rectifiers are suitable for very high voltages, BTW.

The de-forming process (above outlined) to reduce the dielectric to base metal would not be expected to mitigate the diode effect because the dielectric surface is far from homogenous in any dimension, particularly given the etching, and the rectification effect is not dependent upon quantity of oxide, only the maximum voltage will be. So as the reduction begins, it is not uniform because the thinner spot, of necessity, passes more current, and hence preferentially reduces. This eventually creates a thin spot, but it will not necessarily create uniformly greater leakage across the entire surface, and as long as the oxide is thick enough to minimize leakage at the applied voltage the layer still functions. So the capacitor will not explode. It will, however, be damaged and have a dramatically reduce lifespan. This is why capacitors are formed at 125% of rated voltage for general purpose use, and 200% and above for long-lifespan or mission critical use. Thicker oxide layer is more robust for AC signals which gradually damage the oxide layer.

The de-forming therefore only serves to shorten lifespan, not make the rectification effects more pronounced. Or so I speculate.

To find the parallel diode you'll need to do an experiment like applying a low-voltage AC signal in both directions and then subtracting to find the alteration. Voltage must be low enough that significant damage to the dielectric does not occur. Cover the capacitor in case of failure.

Georgiev's work from 1945 is still accurate, but the confusion of that era over semiconductors vs. porosity has, to some extent, been resolved, particularly with respect to soakage.

All of the manufacturers mention rectification effects because this is real and exists.

Barbie says, "Electrochemistry is Hard!"

I must get back to work.

 

[MrLem]

Is it acceptable to use 2 polar caps tied together at negative and again at the positive to create a bi-polar cap for a capacitor coupled application? I would just order a bipolar, but getting them in larger uf seems hard to find?

Mouser have stock of UVP1V102MHD which is a 1000uF 35V bi-polar Nichicon electrolytic for a reasonable(ish) price.