Why can you use electrolytic caps as coupling caps?

[Yuu]

I've been reading about headphone amps and I see circuits like this by people who know much more than me


and I wonder why this is okay? Everyone knows electrolytics only like to be biased one way at DC.
Like, looking at the input cap for example, you will have input bias current on the op amp and that will give some DC offset. Maybe it's negative or positive. Depends on direction of I_B. Maybe it's normal for it to go into the op amp and so the right side will be negative. That doesn't tell you anything about the voltage on the input side though. Being a newbie, I have no clue what DC offset might be there coming from DAC. So this is something I'm unsure about and would love to hear what the people here think and learn why this is okay.

[TimFox]

An electrolytic capacitor can be considered as a capacitor in parallel with a "crappy" diode, so one should pay attention to the polarity of the voltage across it.
A better solution than shown in your circuit is to used a "non-polarized" electrolytic, which is basically two capacitors in series (with opposite polarity).
See:  https://techatronic.com/non-polarized-capacitor-detailed-information-what-is-a-non-polarized-capacitor/
However, if the voltage across the polarized capacitor itself is very small, the unit will still work as a capacitor, so long as the diode current is negligible.
Put a large reverse voltage across the polarized capacitor and it may well explode (literally).
In my audio work, however, I avoided using a normal electrolytic capacitor without a substantial DC voltage across it.

[Yuu]

I've seen this same thing in audio power amplifiers for speakers and stuff. One day I'd like to make one and probe the voltages. Anyway, I think you're right. I think they use these electrolytics because they can count on very small DC voltages across them. I need to get around to reading more.

[magic]

IIRC common wisdom is that elcos can take maybe up to 0.5V reverse bias without major trouble.
I presume it takes some non-zero voltage to trigger the electrochemical reactions involved in DC current flow.

The input side would be OK unless there is horrible DC offset from the upstream gear.
Input current of NE5532 flowing through 10kΩ to ground is just a few mV - not a problem.

But for output coupling? That looks a little dodgy
It certainly robs you of one of the benefits of AC coupling, which is protecting the load from output stage failing short.
Not even sure what's the point of this capacitor at all - amp output is zero-centered without it, with just a few mV offset.

If you are concerned about A1 offset amplified by closed loop gain, put a cap between the stages like in Objective2.

[TimFox]

That voltage seems reasonable for the parasitic diode in the capacitor.
Before electrolytic construction was used for capacitors, it was used in "electrolytic detectors" (Fessenden's invention for an RF detector diode) and "electrolytic rectifiers" (for power applications).
On the latter, see  https://www.nature.com/articles/114462d0  for a 1924 technical article.

[CaptDon]

I saw a couple of mistakes in the verbage at techatronic. Wonder if they proof-read any of it?

[TimFox]

Nobody proofreads in this millennium.

[Yuu]

If you are concerned about A1 offset amplified by closed loop gain, put a cap between the stages like in Objective2.

Yeah I'm designing and laying out something similar to the O2 for personal use right now. Thank you for your post. I had a couple shots of bourbon to offset the 400mg of caffeine I had while designing that circuit today >_> so I may be slightly intoxicated.
These forums are nice. I'm thankful for all the discussion you all provide.

Edit: the circuit in the picture is a Douglas Self circuit from his small signal book

[Benta]

Edit: the circuit in the picture is a Douglas Self circuit from his small signal book

Throw that book away.
There's no reason for a big output cap in an amplifier with ground-referenced output. And the input cap can be small enough that it doesn't need to be a polarized type.

[Yuu]

Throw that book away.
There's no reason for a big output cap in an amplifier with ground-referenced output. And the input cap can be small enough that it doesn't need to be a polarized type.

Interesting. Some people seem to like Self so I'm interested what other people think on that. Self, from his writing, seems like a decently smart guy and presents a lot of data so I like his work.

I am inclined to agree about the caps though. I wonder if the output cap that's there does anything for transients during turn-on or something? I'm not sure.

[Benta]

I'd never heard of him myself, so I had to use web search.
Seems impressive enough, but at over 100 Euro for that book, I'd expect better proof-reading.

No, the output cap has nothing to do with transients at turn-on.
Apart from that: where would you buy BC184/BC214 transistors today? I doubt they were even available in 2014 (date of the book).

[TimFox]

Now that TH semiconductors are being phased out, vacuum-tube vendors often have the parts.
https://www.tubesandmore.com/products/transistor-bc184-general-purpose-92-case-npn
They have BC184B and BC184C for $0.30, selling them to the guitar amplifier enthusiasts.

[Benta]

I think you misunderstood my point.
That you can get BC184 today from somewhere is fine.

I was talking about the book, and that an author would use oddball transistors already phased out instead of choosing, eg, BC327/BC337 instead made me wonder about the rest of the book, in addition to the strange capacitors in his circuit. And then charging over 100 Euro for said book.
When I look at the other books on my shelves, I haven't found anything similar. It's shoddy.

[MikeK]

Isn't that output cap backwards?...Since R13 references that side to ground, isn't the other side the positive side?

[Yuu]

When I look at the other books on my shelves, I haven't found anything similar. It's shoddy.

Are there any books you recommend on audio circuits Benta? A lot of the analog books I've found online seem to just be on integrated analog circuits and I already have one on that.
Any general analog design books you like?

Isn't that output cap backwards?...Since R13 references that side to ground, isn't the other side the positive side?

Hehe that's kind of why I asked the question. Honestly both sides should be at about 0 V DC so to me there wasn't any clear orientation of the input cap or the output cap.

If you guys are curious, here are a couple more pics from that short chapter on headphone amps. Some of Self's plots are old from 1999 so I feel like that's part of why some of the transistor and IC choices seem outdated. Like the TL072 below, I believe that's not something you would want to use anymore.

[RFDx]

I've been reading about headphone amps and I see circuits like this by people who know much more than me and I wonder why this is okay?

This HP-amplifier made it into quite a few of the more premium commercial preamps from Cambridge Audio. There is no output capacitor and none is needed. The current driver uses BC327/BC337 complementary transistors. The op-amps are NE5532 types.

[Tation]

On the same book Self gives some clues about why such capacitors are much larger than initially thought. It is related to avoiding/minimizing distortion caused by the caps. See chapter 2.

[Tation]

I was talking about the book, and that an author would use oddball transistors already phased out instead of choosing, eg, BC327/BC337 instead made me wonder about the rest of the book

Again, Self has studied distortion mechanisms in amplifiers in detail on other books/docs (maybe also on this, didn't look). Not on this circuit, but on common-emitter stages he warns about the distortion generated by the non-linearity of the base-colector capacitance, thus preferring in such cases transistors with lower Cob. BC184 has lower Cob than BC337 in a 1:3 ratio. Maybe he is using BC184 as a "default small-signal amplifier transistor with low distortion".

I would use now BC546/547/550 instead of BC337, much more similar to BC184, but still with slightly worse Cob and noise figure.

In some way these Self books are focused on HiFi (real, on-Earth, HiFi, not audiophoolery) so do not expect many of his circuits/proposals being identical to those in more general-purpose books.

[TimFox]

I own D Self's "Small Signal Audio Design, 2nd Edition" (2015), a voluminous collection of basic information (where he specifically excludes what is pilloried on this site as "audiophool" considerations) and measurements on actual circuits.
Looking at the date codes on his Audio Precision scans, much of the data and circuits presented date to 2009 and 2011, and the built circuits used normal transistors available at that date.
He includes a good summary of the op amps available in 2015, and includes discrete circuits as well as ones using ICs.
The author has been publishing practical stuff like this in places like Wireless World since the 1970s.
I see no reason to discard this book:  feel free to substitute newer transistors for those he used in 2009.

[schmitt trigger]

In the last pair of decades of of the XX century, Doug Self was the go-to authority for audio circuitry.

He would thoroughly explain his choices of components and circuit topology. You could or could not agree with his particular choices, but he strived hard to demonstrate that those choices weren’t arbitrary.

[Terry Bites]

I built a few Doug Self (and JLL Hood) things back in the mid 70's. The transistors were not so oddball then. Mail order suppliers tended to carry all of his favourites.
Onsemi (Ex Motorola) and generic suppliers like Multicomp still offer a lot of MPSAxxx parts.

It's a mystery to me what he didn't go more Euro on his selections. Thank god he didn't go all Alan Sugar NKTxxx on us.

You can pick up loads of D Self articles in WirlessWorld magazine's archives. www.americanradiohistory.com/Wireless_World_Magazine.htm
Best EE mag ever.

[Vovk_Z]

I understand that one of the basic design ideas of Self is how to make things cheap and simple. That's why cheap polar electrolytes instead of non-polar or instead of a film types. They work fine too if you just want to make things done and if you are fine with 0.00x % distortion instead of 0.000x %.

[nigelwright7557]

Electrolytic caps will work backwards up to about a volt or 1.5 volts.
After that it becomes a short and if enough current flows will explode.

[Benta]

On the same book Self gives some clues about why such capacitors are much larger than initially thought. It is related to avoiding/minimizing distortion caused by the caps. See chapter 2.

I don't have the book. but minimizing distortion by using electrolytic output caps without DC bias seems way out.

[TimFox]

On the same book Self gives some clues about why such capacitors are much larger than initially thought. It is related to avoiding/minimizing distortion caused by the caps. See chapter 2.

I don't have the book. but minimizing distortion by using electrolytic output caps without DC bias seems way out.

As Tation posted, on page 72 of the 2nd Edition, Self starts a detailed analysis of electrolytic capacitor distortion, based on measurements.
Some points of his, specifically for unbiased electrolytics:
1.  Do not use electrolytics to set time constants, as in equalization networks.
2.  As Tation stated, he recommends using larger capacitors than required for high-pass filtering, to reduce the voltage across the capacitor itself and thereby reduce its distortion.
3.  Rule of thumb: keep the AC voltage (within the amplifier's bandwidth) below 80 mV (again, across the capacitor).
4.  If no substantial DC blocking is required, a lower-voltage capacitor can be used (carefully) for a higher capacitance in a given space.
5.  Assuming these small AC voltages across the capacitor, DC bias is not required.
He demonstrates these points with data taken with his trusty Audio Precision analyzer, connecting the generator through an unbiased polarized capacitor to a reasonable load resistor at the analyzer input.
He refers to a series of articles on "Capacitor Sound" in Wireless World From July 2002 to March 2003, specifically part 4 in November 2002 on the subject of DC bias.