Question on audio amplifiers for a change...

[Chris Carlen]

Because a differential input stage is a useful way to apply global negative
feedback, which is in turn a good way to improve the response
characteristics of the amplifier.

And because a long-tailed pair permits DC coupling of the input.

But I almost never see DC coupling on the inputs (or outputs) of audio
amps, much to my chagrin, as I sometimes want to adapt them to DC use
without having to hack them.


Good day!

 

[Walter Harley]

But I almost never see DC coupling on the inputs (or outputs) of audio
amps, much to my chagrin, as I sometimes want to adapt them to DC use
without having to hack them.

Inputs, I'm with you. Outputs, it's been a long time since I saw AC-coupled
outputs on a power amp - what kind of amps are you looking at? Seems
everything I see is split-supply, with DC negative feedback to minimize DC
offset error (which is why you need the differential stage to be DC-coupled,
at least on the inverting input), and hopefully with DC detection and
shutdown on the output to protect speakers.

Which of course doesn't help you if you need the amp for something other
than speakers! But it's a bad idea to put much DC into speakers. Can
result in unintended pyrotechnics.

 

[Kevin Aylward]

The Doyen of audio amp design, Douglas Self, disagrees with you, Kev.

And I care?

First, I don't believe that Mr. Self states that the diff pair is
fundamentally used to eliminate a cap. This is complete nonsense. Your
general ignorance of electronics suspects me to believe that you
misunderstood what he said. If he does, he is wrong. Period.

For example, essentially, *all* commercial power amps use input coupling
capacitors. This is to prevent DC from *other* devices putting DC on to
the speaker. It would be suicide to omit the input coupling cap for
general consumer/pro audio use.

Look, dude, just because someone publishes a lot, it don't mean what
he/she says is true, or that he/she knows what he is talking about. For
example, just read the latest Electronics World.

Kevin Aylward
[email protected]
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

 

[Fred Bartoli]

Hi all,

Why does the first stage of all commercial audio amps seem to consist
of a differential (long-tailed) pair? AFAIA, the chief characteristic
of this configuration is just to eliminate 'mush' on an input signal,
like if it's come from a long line input and has a high noise floor,
which clearly isn't the case between your CD/tape-deck/phonograph or
whatever. So why?

If your amp has a correct dynamic range, there must be a lot of room for the
mush.


Thanks,
Fred.

 

[Paul Burridge]

And I care?

First, I don't believe that Mr. Self states that the diff pair is
fundamentally used to eliminate a cap. This is complete nonsense. Your
general ignorance of electronics suspects me to believe that you
misunderstood what he said. If he does, he is wrong. Period.

For example, essentially, *all* commercial power amps use input coupling
capacitors. This is to prevent DC from *other* devices putting DC on to
the speaker. It would be suicide to omit the input coupling cap for
general consumer/pro audio use.

Look, dude, just because someone publishes a lot, it don't mean what
he/she says is true, or that he/she knows what he is talking about. For
example, just read the latest Electronics World.

Sure, Kev. Whatever.

 

[Winfield Hill]

Paul Burridge wrote...

Sure, Kev. Whatever.

I have always enjoyed Douglas Self's audio-engineering articles and
bought a copy of his book on that basis. His book is well written
and includes an extensive discussion of various common amplifier
circuits, along with distortion measurements and Self's opinions,
which makes for entertaining reading. However, he is far too short
on the theory, not because he doesn't understand it, but because he
choose to keep his book simple: "mathematics has been confined to a
few simple equations." Well, he kept his promise, and despite all
the book's detail, he doesn't even have basic Ebers-Moll. Pity.

Now to the controversy at hand. Self agrees with Kevin.

The alternate to a differential-pair input stage is a single-ended
amplifier. Self does not have much to say about the relative merits
of these two choices, but his book is primarily about amplifier
distortion, and on page 62 he discusses the single-ended input-stage
amplifier. Note the heading of his lone paragraph on the subject:

"Singleton input stage versus differential pair
Using a single input transistor (Figure 4.3a) may seem attractive,
where the amplifier is ac coupled or has a separate DC servo: it at
least has promises strict economy. However, the snag is that this
singleton configuration has no way to cancel the second harmonics
generated in copious quantities by the strongly-curved exponential
Vin/Iout characteristic (1). The result is shown in Figure 4.2
curve-B, where the distortion is much higher, though rising at the
slower rate of 12dB/octave."

(1) this is a reference to Gray and Meyer's 1984 book, page 172,
where distortion predicted by the Ebers-Moll formula is discussed.

Self properly condemns the single-ended input stage alternate to a
differential pair for its excessive distortion, and not primarily
for other reasons, as Paul claimed. So the "Doyen of audio amp
design, Douglas Self," does in fact *agree* with Kevin after all.

Kevin wins the argument.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

[Stefan Heinzmann]

The Doyen of audio amp design, Douglas Self, disagrees with you, Kev.

Where does he say that he uses differential input stages in order to
save capacitors? He does say in chapters 3 and 6 that capacitors can be
a source of distortion, but that doesn't prevent him from using them
anyway. May I refer you to the schematics of his 50W amplifier in
chapter 6. This amp, of which he says that it has the potential to be
Blameless (in his terminology), nevertheless uses a differential input
stage *and* a capacitor to block DC at the input.

If you care to read his argument on capacitor distortion, you will find
that a condition for its occurence is when the capacitor is used to
block DC with a significant load behind it. His measurements for
demonstrating the effect were done with a 680 Ohm load. In his amplifier
the load of the DC blocking capacitor is 10k Ohm.

It can't be the cost of the capacitor, either, that is the problem. A
10µF electrolytic is cheap indeed. Omitting it at the amplifier's input
would remove any protection against DC voltages being applied to the
amp. The consequence of this could be a smoked loudspeaker or a smoked
amp or both. You would have to prevent this through other means which
are invariably more expensive than the omitted cap.

Having said that I have to add that I find the usage of just one
polarized cap at the input somewhat sloppy. It will not offer sufficient
protection if significant DC is applied at the wrong polarity, because
an ordinary electrolytic will start to leak appreciably beyond about 2V
reverse bias. This is unlikely enough not to be a real problem, I admit,
but why not do it right and use a nonpolarized cap (or two normal ones
in series)?

 

[Paul Burridge]

Paul Burridge wrote...

I have always enjoyed Douglas Self's audio-engineering articles and
bought a copy of his book on that basis. His book is well written
and includes an extensive discussion of various common amplifier
circuits, along with distortion measurements and Self's opinions,
which makes for entertaining reading. However, he is far too short
on the theory, not because he doesn't understand it, but because he
choose to keep his book simple: "mathematics has been confined to a
few simple equations." Well, he kept his promise, and despite all
the book's detail, he doesn't even have basic Ebers-Moll. Pity.

Now to the controversy at hand. Self agrees with Kevin.

The alternate to a differential-pair input stage is a single-ended
amplifier. Self does not have much to say about the relative merits
of these two choices, but his book is primarily about amplifier
distortion, and on page 62 he discusses the single-ended input-stage
amplifier. Note the heading of his lone paragraph on the subject:

"Singleton input stage versus differential pair
Using a single input transistor (Figure 4.3a) may seem attractive,
where the amplifier is ac coupled or has a separate DC servo: it at
least has promises strict economy. However, the snag is that this
singleton configuration has no way to cancel the second harmonics
generated in copious quantities by the strongly-curved exponential
Vin/Iout characteristic (1). The result is shown in Figure 4.2
curve-B, where the distortion is much higher, though rising at the
slower rate of 12dB/octave."

(1) this is a reference to Gray and Meyer's 1984 book, page 172,
where distortion predicted by the Ebers-Moll formula is discussed.

Self properly condemns the single-ended input stage alternate to a
differential pair for its excessive distortion, and not primarily
for other reasons, as Paul claimed. So the "Doyen of audio amp
design, Douglas Self," does in fact *agree* with Kevin after all.

Kevin wins the argument.

Kev's argument was based on a misinterpretation of what I said Self
was claiming. He (Self) made *several* arguments in favour of the DA
as a first stage, the least significant of which was the saving of a
large, expensive cap at the input. But if you're adjudicating and you
feel Kev's come out best, then I'm pleased for him. It's about time he
won *some* argument here, given his exceedingly limited command of
English.

 

[Paul Burridge]

It can't be the cost of the capacitor, either, that is the problem. A
10µF electrolytic is cheap indeed.

So what, Steph? Would you seriously capacitively couple audio with a
10uF cap for hi-fi purposes??? I mean, fine if you're talking about
some seriously high-Z coupling, but that's not normally the case in
the subject under discussion.

 

[Stefan Heinzmann]

So what, Steph? Would you seriously capacitively couple audio with a
10uF cap for hi-fi purposes??? I mean, fine if you're talking about
some seriously high-Z coupling, but that's not normally the case in
the subject under discussion.

Yes, I would indeed; in fact I have, and Self himself does it throughout
his book. Together with the 10k input impedance this forms a high-pass
filter with a corner frequency of about 0.6 Hz. For a frequency roll-off
compatible with hi-fi even a 1µF capacitor would suffice. Self obviously
uses the higher value in order to avoid distortion. Read his section
about capacitor distortion in chapter 6 and you'll know why.

 

[Stefan Heinzmann]

Yes, I would indeed; in fact I have, and Self himself does it throughout
his book. Together with the 10k input impedance this forms a high-pass
filter with a corner frequency of about 0.6 Hz. For a frequency roll-off
compatible with hi-fi even a 1µF capacitor would suffice. Self obviously
uses the higher value in order to avoid distortion. Read his section
about capacitor distortion in chapter 6 and you'll know why.

1.6Hz, sorry!

 

[normanstrong]

Oh?

Why?

I'm not talking about the input coupling capacitor here, but rather
the emitter bypass capacitor. Since a single-ended input stage
requires that both the signal and the negative feedback must appear
between the base and emitter of the input transistor, a large bypass
capacitor is necessary. A differential input takes a much smaller
capacitor since the gain of the second transistor increases the
impedance in the feedback circuit.

So I've given you one advantage of the differential input stage. Now
it's your turn to give an advantage to the single-ended input.

Norm

 

[Kevin Aylward]

On 11 Jul 2004 06:10:20 -0700, Winfield Hill

much higher, though rising at the

Kev's argument was based on a misinterpretation of what I said Self
was claiming.

Ho humm.. You claimed that Self claimed that one, essentially, uses a
diff pair to save a cap. I claimed that was claim was crap.

He (Self) made *several* arguments in favour of the DA
as a first stage, the least significant of which was the saving of a
large, expensive cap at the input. But if you're adjudicating and you
feel Kev's come out best, then I'm pleased for him. It's about time he
won *some* argument here,

Err.. and just what would be some of those arguments that your claimed I
lost.?

Kevin Aylward
[email protected]
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

 

[Paul Burridge]

Ho humm.. You claimed that Self claimed that one, essentially, uses a
diff pair to save a cap. I claimed that was claim was crap.

No I didn't! Self explains *many* advantages conferred by the use of a
DA stage, only *one* of which is the saving of a large and expensive
capacitor!

Err.. and just what would be some of those arguments that your claimed I
lost.?

Pretty much most of the stuff you post on neo-Darwinism and sexual
selection. I'm unable to judge your credibility on matters
electronique, so I'll stick with what I know.

 

[Kevin Aylward]

No I didn't! Self explains *many* advantages conferred by the use of a
DA stage, only *one* of which is the saving of a large and expensive
capacitor!


Pretty much most of the stuff you post on neo-Darwinism and sexual
selection.

Oh?

This would be very unlikely as most of what I say is all *standard* and
accepted Darwinian evolution, which is the cornerstone of all of biology
and zoology. Its not debateable. What is observed mostly, is what
replicates the most. Its a tautology, i.e. correct by construction.
Period. All we have do is look to see if we can identify what will
generate the most offspring. e.g. For males, its pretty much sex at very
opportunity with as many as possible, for females its chose the best
they can to avoid being in the club for 9 months with bad gene stock. If
you have an argument to refute what is actually experimentally proven,
lets hear it.

I'm unable to judge your credibility on matters

electronique,
so I'll stick with what I know.

Which don't appear to be much.

Kevin Aylward
[email protected]
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

 

[John Larkin]

Oh?

This would be very unlikely as most of what I say is all *standard* and
accepted Darwinian evolution, which is the cornerstone of all of biology
and zoology. Its not debateable. What is observed mostly, is what
replicates the most. Its a tautology, i.e. correct by construction.
Period. All we have do is look to see if we can identify what will
generate the most offspring. e.g. For males, its pretty much sex at very
opportunity with as many as possible, for females its chose the best
they can to avoid being in the club for 9 months with bad gene stock. If
you have an argument to refute what is actually experimentally proven,
lets hear it.

A few interesting twists:

The grandmother theory: unlike most animals and pre-homo-sapiens,
humans live far beyond their active-reproductive ages, so that they
can protect and teach their kids. Social structure then becomes a
survival mechanism, and we have social evolution as well as
biological. Males have a primary family and tribe which command their
first loyalty. Females want not just good genes, but a good provider
and protector, too, so the beautiful brats can survive. So mating
isn't just lets-go-for-it, but lets-make-a-deal.

Inter-species gene transfer is an interesting idea, possibly mediated
by viruses. Why restrict the available gene pool to your own species?

Macros: DNA is complex enough to have macro-evolution mechanisms,
genes whose function is to manage the evolution process itself.

John

 

[John Woodgate]

I read in sci.electronics.design that John Larkin <jjlarkin@highSNIPland
THIStechPLEASEnology.com> wrote (in <csk5f05oen2bk9525a2odrd0rfcnvqstlj@
4ax.com>) about 'Question on audio amplifiers for a change...', on Mon,
12 Jul 2004:

Macros: DNA is complex enough to have macro-evolution mechanisms,
genes whose function is to manage the evolution process itself.

Such as a 'homosexuality gene' or a 'monk/nun gene', to terminate a gene
line for some hidden reason?

 

[Genome]

| Paul Burridge wrote:
| > On Mon, 12 Jul 2004 06:44:18 GMT, "Kevin Aylward"
| >
| > I'm unable to judge your credibility on matters
| > electronique, so I'll stick with what I know.
|
| Which don't appear to be much.
|

Burridge had his five minutes of fame when he pissed off Win Hill. He
was a troll and now he's a trolling has been.

You can't call a wanker a wanker, it simply doesn't penetrate.

Paul, you're a Wanker..... see, didn't work.

DNA

 

[Keith Williams]

A few interesting twists:

The grandmother theory: unlike most animals and pre-homo-sapiens,
humans live far beyond their active-reproductive ages, so that they
can protect and teach their kids. Social structure then becomes a
survival mechanism, and we have social evolution as well as
biological. Males have a primary family and tribe which command their
first loyalty. Females want not just good genes, but a good provider
and protector, too, so the beautiful brats can survive. So mating
isn't just lets-go-for-it, but lets-make-a-deal.

I'll buy that. Indeed... ;-)

Inter-species gene transfer is an interesting idea, possibly mediated
by viruses. Why restrict the available gene pool to your own species?

Umm, a little here? The the ability to procreate and
have fertile offspring pretty much defines a "species".

Macros: DNA is complex enough to have macro-evolution mechanisms,
genes whose function is to manage the evolution process itself.

Macros, or check-digits? It seems to me that the DNA
hamming-distance is pretty large. Though I guess it's
how you look at it.

 

[Paul Burridge]

| Paul Burridge wrote:
| > On Mon, 12 Jul 2004 06:44:18 GMT, "Kevin Aylward"
| >
| > I'm unable to judge your credibility on matters
| > electronique, so I'll stick with what I know.
|
| Which don't appear to be much.
|

Burridge had his five minutes of fame when he pissed off Win Hill. He
was a troll and now he's a trolling has been.

You can't call a wanker a wanker, it simply doesn't penetrate.

Paul, you're a Wanker..... see, didn't work.

Be off with you, you ruffian! Be off with you, I say!!