[Eeyore]

Arny Krueger wrote:

"Phil Allison" wrote

** Such a power stage has a rated output of over 500
watts and so HAS to use at least 8 large output devices.

Irrelevant.

Very relevant indeed since each device is working at a fraction of the total
load current.

Graham

[Eeyore]

Arny Krueger wrote:

"Eeyore" wrote
Arny Krueger wrote:

We all know that speakers can be capacitive-reactive and
can jack the load current to unexpectedly high levels.

Entirely calaculable though.


A common power amp will have +/- 80
volt DC rails and may need to deliver up to 15 amps peak
to the load.

15 Amps ? I'd design for 40 !

OK.

In your opinion Graham, what would the quiescent dissipation of a
traditionally-designed output stage like this be?

Phil's already covered that and I'm in accord with what he says.

Graham

[Eeyore]

Arny Krueger wrote:

"Eeyore" wrote
Arny Krueger wrote:
"Eeyore" wrote

I'm concerned about the transconductance 'modulation'
with signal level since this introduces non-linearity.

Again, covered in
http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm

Which is what I'm designing out.

Although I'd hoped this thread might be more about any
audible benefits of Class A rather than a discussion of
design principles ( I'm quite au-fait with those ! ).

Like I said before, I've never done any DBTs involving Class-A amps. This is
partially because I've never seen a true Class A amp that was capable of
interesting power levels, in real life.

Fair enough.

I've heard enough positive comment on them to pursue my line of thinking on the
matter already btw.

Graham

[Eeyore]

Arny Krueger wrote:

"Eeyore" wrote
Phil Allison wrote:
"Eeyore"

In fact I've a sort of grudge to bear in fact.

** So THAT is what " Eeyore " really is -

a bear with some sort of grudge ??

Long story. Only a small grudge really. I must learn to
lie in job interviews too.

** Oh - that job Doug got with Soundcraft ?

Uhuh.


He did the power amp stage of their "PowerStation".

Joke isn't it ?

What's wrong with it?

Do tell!

It's fairly 'routine'. That's the joke.

Graham

[Phil Allison]

"Eeyore"
Arny Krueger wrote:

"Phil Allison"

Yes, there are two problems with class A - both related
to the fact that with class A, the output stage always
has far more current flowing in it. (1) Power transistors tend to be
less
linear at high
currents - the beta falls off. Running an output stage
class A approximately doubles the current that the
output stage has to handle. So, you move the operating
point way out on the output devices. (2) Less SOA from the perspective
of
the load, because
the output devices are pulling so hard against each
other.


** What a load of complete DRIVEL !!!!

Arny - leave commenting on power amp design to people
who know something about it cos they spend their lives
dealing with it at component level.
Cos YOU do not have a bloody clue.

Phil, its real handy for that you seem to be so mentally incapable of
framing a proper technical reply. Everybody is going to dismiss your
ranting.

I'd personally like to see you claim that power transistors get more
linear
when run at very high currents and that there is no such thing as beta
fall-off at high currents.

Oh there is but since a classic Class A output has such high standing
dissipation you use more devices in parallel and you're actually working
them at
lower currents than in A/B.



** Correct.

Arny has made the same ASININE error in FOUR posts.

A "class A amplifer" is one DESIGNED to work correctly in that mode.

His wacky claim about peak device currents being higher in class A is FALSE.




......... Phil

[Phil Allison]

"Arny Krueger"


Agreed again. Building a truely Class A power amp that
can deliver significant amounts of power output is
really a pretty awesome thing. The OPT stage quiescent
current has to be equal to the *peak* current that is
delivered to the load.

** Absolutely FALSE.

The peak load current is *exactly* double the idle
current for an amp operating in class A.

Agreed.


** Self contradiction - right here.



The idle current flow in one device of a pair increases
to double while the other drops to just under zero at
peak level.

Agreed.

That double value current peak flows entirely via the
load to the common point.

Agreed.

Very basic stuff.

So what's your point?


** You are WRONG !!



You just broke down what I said down into steps.


** I JUST COMPLETELY CONTRADICTED YOU

- YOU ****WIT !!.




We all know that speakers can be capacitive-reactive and
can jack the load current to unexpectedly high levels. A
common power amp will have +/- 80 volt DC rails and may
need to deliver up to 15 amps peak to the load.

** Such a power stage has a rated output of over 500
watts and so HAS to use at least 8 large output devices.

Irrelevant.


** Shame how it is MASSIVELY relevant.



So, now the OPT is dissipating 2400 watts per channel
just sitting there, not counting internal losses.

** What a load of complete DRIVEL !!!!

What's your number, Phil? Remember, we are talking about a class-A output
stage that delivers 80 volts peak and 15 amps peak.


** Therefore it runs at 7.5 amps idle.

It also dissipates 1200 watts spread over at least 20 large devices.


Arny - leave commenting on power amp design to people who know something
about it cos they spend their lives dealing with it at component level.

Cos YOU do not have a bloody clue.





........... Phil

[Ruud Broens]

"Eeyore" wrote in message
...
:
:
: Don Pearce wrote:
:
: Eeyore wrote:
: Don Pearce wrote:
: I'm concerned about the transconductance 'modulation' with signal level
: since this introduces non-linearity.
:
: Graham
:
: That would be greatest in the transition region between class A and
: class B.
:
: Exactly spot-on Don and readily visible when looking at a distortion analyser's
output.
: This is why Trevor's ideas baffle me.
:
: My intention is to entirely eliminate this with a kind of 'hybrid' output
stage.
:
:
: Graham
:
what? you're going to use tubes ?

;-)
R.

[Arny Krueger]

"Eeyore" wrote in
message
Arny Krueger wrote:


I'd personally like to see someone claim that power
transistors get more linear when run at very high
currents and that there is no such thing as beta
fall-off at high currents.

Oh there is, but since a classic Class A output has such
high standing dissipation you use more devices in
parallel and you're actually working them at lower
currents than in A/B.

I think I saw the effects of beta fall-off in my testing. I was working
with 8 devices up and 8 devices down. VCC was about 80 volts. The devices
were complements from the MJE 1502x series.

Admittedly, one go add enough devices so that beta fall-off wouldn't be an
issue.

[Bertie the Bunyip]

Eeyore wrote in
:



Trevor Wilson wrote:

"Eeyore" wrote in message

Have you ever spent any time listening to a true Class A solid
state amplifiers ?

**We all have. Pretty much every amplifier operates Class A to a few
tens of milliwatts.

That's not what I meant by true Class A though.


Plenty loud enough for quiet listening. Some high bias Class A/B
designs operate to a couple of Watts in Class A. Even with 90dB/W/M
speakers, this is pretty loud.

Sure enough but what you need to be aware of is that working in the
crossover region the output device transconductance is varying widely
and putting non-linearity into the transfer characteristic


For my own part, I am very familiar with an
amplifier which has user switchable Class A operating points. Since
the design eschews the use of Global NFB, it is far more sensitive to
bais current changes than most high Global NFB designs. Once the
Class A operating point goes beyond 10 Watts Class A, there is bugger
all difference.

In this case the crossover point has 'moved up the scale' and has less
overall effect as a percentage of signal level. It doesn't totally go
away though.


Except, when using the thing on a US power receptacle. At 100
Watts Class A, the power consumed from the mains supply is too high
for regular US outlets, without losing mains Voltage.

Really ?


At 100 Watts Class A, in
the US, the sound becomes slightly 'compressed'.

I find some difficulty believing that.


Pretty much anywhere else
on the planet is fine.


Any comments on how it sounded compared to any other types ?

**IMO, Class A is used by so-so designers who can't figure out how to
design an amplifier properly. Once bias current is set above the
'knee' of the device (around 100mA for BJTs and 1 Amp for MOSFETs)
any more Class A bias is superfluous.

There are still some advantages in not turning the devices hard off
though.


Think of it this way: A badly designed Class A/B amplifier will
probably sound better when operating in Class A. A well designed
Class A/B amplifier will probably sound slightly worse when operating
in Class A.

I can't understand the rationale for your 2nd idea there.

Graham



fjukkwit netkkkop


bertie

[Bertie the Bunyip]

Eeyore wrote in
:



Trevor Wilson wrote:

"Eeyore" wrote in message
Trevor Wilson wrote:
"Eeyore" wrote in message

Have you ever spent any time listening to a true Class A solid
state amplifiers ?

**We all have. Pretty much every amplifier operates Class A to a
few tens of milliwatts.

That's not what I meant by true Class A though.

**'True Class A' is a very rubbery definition. A 'True Class A'
amplifier rated at (say) 50 Watts @ 8 Ohms, will only be 25 Watts
Class A, when driving 4 Ohms, 12.5 Watts @ 2 Ohms, and so on. Given
the fact that a typical 8 Ohms speaker can easily exhibit impedance
minima down to 3 Ohms, you'll see the problem.

My understanding of the textbook definition is that the device(s) are
always conducting for the entire cycle i.e. never switch off at all
output power levels.


Plenty loud enough for quiet listening. Some high bias Class A/B
designs operate to a couple of Watts in Class A. Even with
90dB/W/M speakers, this is pretty loud.

Sure enough but what you need to be aware of is that working in the
crossover
region the output device transconductance is varying widely and
putting non-linearity into the transfer characteristic

**Well, you've managed to narrow your choice of devices down,
significantly, since the amplification factor of BJTs is expressed
as hFE. Current gain. The variation in current gain of modern BJTs is
spectacularly low, over very wide Collector currents. You need to
examine the curves on this page:

http://www.futurlec.com/Transistors/2SC5200.shtml

Pay close attention to the hFE/Ic curves. You'll note that the device
is linear from 10ma all the way through to 3 Amps.

Current gain doesn't much bother me. It'll be coming from a very low
impedance

fjukkwit netkkko[


Bertie

[Arny Krueger]

"Eeyore" wrote in
message
Arny Krueger wrote:

"Phil Allison" wrote

** Such a power stage has a rated output of over 500
watts and so HAS to use at least 8 large output devices.

Irrelevant.

Reason being that my amp running off VCC = +/- 80 volts (no-load) is more
like a conservative 250 wpc @8 ohms amp, not 500 watts. I get about 300 wpc
at clipping presuming perfect power supply regulation which of course is
impractical.

The power supply was based on a toroidal power transformer from a commerical
power amp. My test amp used 16 large (MJE 1502x) output devices, which is
obviously *twice* Phil's 8.

Very relevant indeed since each device is working at a
fraction of the total load current.

Agreed, in the general case.

[Bertie the Bunyip]

Eeyore wrote in
:



Trevor Wilson wrote:

"Eeyore" wrote in message
Trevor Wilson wrote:
"Eeyore" wrote in message
Trevor Wilson wrote:
"Eeyore" wrote in
message

Have you ever spent any time listening to a true Class A
solid state amplifiers ?

**We all have. Pretty much every amplifier operates Class A to
a few tens of milliwatts.

That's not what I meant by true Class A though.

**'True Class A' is a very rubbery definition. A 'True Class A'
amplifier rated at (say) 50 Watts @ 8 Ohms, will only be 25 Watts
Class A, when driving 4 Ohms, 12.5 Watts @ 2 Ohms, and so on.
Given the fact that a typical 8 Ohms speaker can easily exhibit
impedance minima down to 3 Ohms, you'll see the problem.

My understanding of the textbook definition is that the device(s)
are always conducting for the entire cycle i.e. never switch off at
all output
power levels.

**And, unless the precise load is specified, then the definition of
Class A is meaningless.

It can be specified as a minimum load impedance.


Plenty loud enough for quiet listening. Some high bias Class
A/B designs operate to a couple of Watts in Class A. Even with
90dB/W/M speakers, this is pretty loud.

Sure enough but what you need to be aware of is that working in
the crossover
region the output device transconductance is varying widely and
putting non-linearity into the transfer characteristic

**Well, you've managed to narrow your choice of devices down,
significantly,
since the amplification factor of BJTs is expressed as hFE.
Current gain.
The variation in current gain of modern BJTs is spectacularly low,
over very
wide Collector currents. You need to examine the curves on this
page:

http://www.futurlec.com/Transistors/2SC5200.shtml

Pay close attention to the hFE/Ic curves. You'll note that the
device is linear from 10ma all the way through to 3 Amps.

Current gain doesn't much bother me. It'll be coming from a very
low impedance
stage.

**Current gain SHOULD bother you, since that is what we're
discussing.

It's not what I'm discussing. I've had excellent results with 'super
beta' output stages. Commercial ( pro ) amplifiers simply can't used
matched devices for practical production and servicing reasons so you
design the issue out.


The
linearity of the devices is exemplary, over a very wide current range
and down to quite low currents.

For my own part, I am very familiar with an
amplifier which has user switchable Class A operating points.
Since the
design eschews the use of Global NFB, it is far more sensitive
to bais current changes than most high Global NFB designs. Once
the Class A operating point goes beyond 10 Watts Class A, there
is bugger all difference.

In this case the crossover point has 'moved up the scale' and
has less overall effect as a percentage of signal level. It
doesn't totally go away though.

**More bull****. If one goes to the trouble to match output
devices, then,
combined with the excellent linearity I previously mentioned,
crossover distortion does not exist. It will be swamped by other
forms of distortion and noise.

You can do all this without matched devices.

**Using matched devices eliminates all possiblity of problems.

I don't want to used matched devices. In any case the gm still varies
hugely at small currents.


I simply can't conceive that the
change in gm won't ever have an effect.

**It probably would. However, I am suggesting that modern BJTs have
such excellent current linearity, that no problems will occur when
using in Class A/B.

That's not my experience. I should point out that I am targeting
*ultra-low* distortion figures.


Except, when using the thing on a US power receptacle. At 100
Watts Class A, the power consumed from the mains supply is too
high for regular US outlets, without losing mains Voltage.

Really ?

**Yes, really. US 117VAC receptacles are rated for 1,500 Watts. In
reality,
I have measured significant Voltage drops with as little as 1,000
Watt loads. IMO, it is for this reasona that manufacturers such as
Krell stopped building Class A amplifiers for domestic
consumption.

At 100 Watts Class A, in
the US, the sound becomes slightly 'compressed'.

I find some difficulty believing that.

**After you try it, get back to me. US 117VAC power receptacles
and associated wiring is pretty ordinary, compared to almost
anywhere else on the planet.

How would ac power voltage sag cause a compressed sound ?

**Think about it. Think about it in the context that few power
amplifiers use a regulated power supply.

It's a crappy design that's supply rail sensitive !


Pretty much anywhere else
on the planet is fine.


Any comments on how it sounded compared to any other types ?

**IMO, Class A is used by so-so designers who can't figure out
how to design an amplifier properly. Once bias current is set
above the 'knee' of the
device (around 100mA for BJTs and 1 Amp for MOSFETs) any more
Class A bias is superfluous.

There are still some advantages in not turning the devices hard
off though.

**No. However, you can name those alleged advantages, if you wish.

Elimination of switching effects.

**Those effects can be elimintated through the use of matched linear
devices, constant operating temperatures and modest amounts of Class
A bias current. No full Class A operation is required.

I'm talking about *switching effects* - carrier storage and stuff.


Think of it this way: A badly designed Class A/B amplifier will
probably
sound better when operating in Class A. A well designed Class
A/B amplifier will probably sound slightly worse when operating
in Class A.

I can't understand the rationale for your 2nd idea there.

**Douglas Self explains it very nicely.

Call me a sceptic then !

**Read Self's work.

I've browsed it occasionally I don't consider him to be especially any
guru though. In fact I've a sort of grudge to bear in fact.

Graham



fjukkwit wannabe ****


Bertire

[Bertie the Bunyip]

Eeyore wrote in
:



Phil Allison wrote:

"Eeyore"

In fact I've a sort of grudge to bear in fact.

** So THAT is what " Eeyore " really is -

a bear with some sort of grudge ??

Long story. Only a small grudge really. I must learn to lie in job
interviews too.

Graham


Lying ****



Bertie

[Arny Krueger]

"Eeyore" wrote in
message
Arny Krueger wrote:

"Eeyore" wrote
Arny Krueger wrote:

We all know that speakers can be capacitive-reactive
and can jack the load current to unexpectedly high
levels.

Entirely calaculable though.


A common power amp will have +/- 80
volt DC rails and may need to deliver up to 15 amps
peak to the load.

15 Amps ? I'd design for 40 !

OK.

In your opinion Graham, what would the quiescent
dissipation of a traditionally-designed output stage
like this be?

Phil's already covered that and I'm in accord with what
he says.

I see no numbers, just a lot of fussing and fuming.

[Bertie the Bunyip]

Eeyore wrote in
:



Phil Allison wrote:

"Eeyore"

In fact I've a sort of grudge to bear in fact.

** So THAT is what " Eeyore " really is -

a bear with some sort of grudge ??

Long story. Only a small grudge really. I must learn to lie in job
interviews too.

** Oh - that job Doug got with Soundcraft ?

Uhuh.


He did the power amp stage of their "PowerStation".

Joke isn't it ?

Graham



fjukkkkwti netkkkop



Bertie

[Bertie the Bunyip]

Eeyore wrote in
:



Arny Krueger wrote:

"Eeyore" wrote
Phil Allison wrote:
"Eeyore"

In fact I've a sort of grudge to bear in fact.

** So THAT is what " Eeyore " really is -

a bear with some sort of grudge ??

Long story. Only a small grudge really. I must learn to
lie in job interviews too.

** Oh - that job Doug got with Soundcraft ?

Uhuh.


He did the power amp stage of their "PowerStation".

Joke isn't it ?

What's wrong with it?

Do tell!

It's fairly 'routine'. That's the joke.

Graham




fjukkwt enenenentkkkkkunt



Bertie

[Arny Krueger]

"Eeyore" wrote in
message
Arny Krueger wrote:

"Eeyore" wrote
Don Pearce wrote:
Eeyore wrote:

I'm concerned about the transconductance 'modulation'
with signal level since this introduces non-linearity.

Graham

That would be greatest in the transition region between
class A and class B.

Exactly spot-on Don and readily visible when looking at
a distortion analyser's output.

Agreed.

This is why Trevor's ideas baffle me.

The point that Trevor alludes to and Doug Self describes
in detail, is the fact that there are numerous other
sources of nonlinear distortion that can be bigger
problems in poorly designed amps.

If they're poorly designed !


Don't be proud Graham, go spend some time with Self's
article:

http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm

I've been there before Arny. There's nought Self can
teach me there.


My intention is to entirely eliminate this with a kind
of 'hybrid' output stage.

How? Other than class-A you'd need to come up with some
kind of Gm-halving circuit.

It is a 'form' of Class A that indeed meets the classic
definition but without the very high idle current. It
involves quite a radical rethink of the output stage.

In the past the usual approach to this has been to use a bias circuit that
keeps the output devices from ever turning fully off.

[Bertie the Bunyip]

Eeyore wrote in
:



Arny Krueger wrote:

"Eeyore" wrote

I've browsed it occasionally I don't consider him to be
especially any guru though.

I've read Self's little article about power amp distortion mechanisms
several times, and have been struck by the benefits that I might have
received from it, had it been available about 20 years ago, when I
was
trying to design a SOTA power amp.

In fact I've a sort of grudge to bear in fact.

Do tell.

See Phil's post.

Graham




fjukkwit wannabe planespotter

bertie

[Bertie the Bunyip]

Eeyore wrote in
:



Trevor Wilson wrote:

"Scott Dorsey" wrote
Trevor Wilson wrote:

**There are plenty of reasons NOT to go pure Class A and very few
to do so.

Other than size and heat?

**Cost, reliability and the fact that high bias Class A/B is lower in
distortion.

Lower ?

Graham



fjukkwit netkkkoping ****


Bertie

[Bertie the Bunyip]

Eeyore wrote in
:



Arny Krueger wrote:

"Trevor Wilson" wrote
"Eeyore" wrote
Trevor Wilson wrote:
"Scott Dorsey" wrote
Trevor Wilson wrote:

**There are plenty of reasons NOT to go pure Class A
and very few to do so.

Other than size and heat?

**Cost, reliability and the fact that high bias Class
A/B is lower in distortion.

Lower ?

**Yup. Self has provided convincing proof of this.

Agreed.

I would hope that this paper would be required reading for any
self-appointed power amp guru:

http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm

Meow !



Fjukk you, you hypocritical netkkkopin gturd



Bertie

[Arny Krueger]

"Eeyore" wrote in
message
Arny Krueger wrote:

"Eeyore" wrote
Phil Allison wrote:
"Eeyore"

In fact I've a sort of grudge to bear in fact.

** So THAT is what " Eeyore " really is -

a bear with some sort of grudge ??

Long story. Only a small grudge really. I must learn
to lie in job interviews too.

** Oh - that job Doug got with Soundcraft ?

Uhuh.


He did the power amp stage of their "PowerStation".

Joke isn't it ?

What's wrong with it?

Do tell!

It's fairly 'routine'. That's the joke.

IOW, despite the use of a "big name" designer, the wine came out tasting
pretty ordinaire.

Is it class G?

I notice that Self makes some heavy claims about a proprietary implemention
of class G.

[Arny Krueger]

"Eeyore" wrote in
message
Arny Krueger wrote:

As my distortion measuring gear progressed to residuals
in below 0.01% there was always a mixed bag of
measurable differences. But they were arguably quite
small, given that its often darn hard to hear distortion
below 0.1% or so.

I had a 'revalation' that blew away that idea about 30
years ago.

Do tell.

I made my last try at making small amounts of nonlinear distortion audible
about 5 years ago, and the 0.1% number was the fruit of that effort. My
results aren't that dissimilar from those found in J. Robert Stuart
(Meridian Audio), "Digital Audio for the Future", Audio, 3/98 pp 30-37.
Stuart isn't exactly conservative on this topic. I know of nobody credible
who has claimed lower thresholds for audibilty.

[Bertie the Bunyip]

Eeyore wrote in
:



Trevor Wilson wrote:

IOW: The design is more important than the Class of operation. Class
A will help a bad design and, at best, do nothing to a good design.

You're assuming they all start as Class AB output stages there.

Graham



netkkkoping piece of ****/



Bertie

[Phil Allison]

"Arny Krueger"
"Eeyore"

In your opinion Graham, what would the quiescent
dissipation of a traditionally-designed output stage
like this be?

Phil's already covered that and I'm in accord with what
he says.

I see no numbers, just a lot of fussing and fuming.



** Open your bloody eyes - asshole.

Fools like YOU only see what they want to.




........ Phil

[Arny Krueger]

"Phil Allison" wrote in message

"Eeyore"
Arny Krueger wrote:

"Phil Allison"

Yes, there are two problems with class A - both
related to the fact that with class A, the output stage always
has far more current flowing in it. (1) Power
transistors tend to be less
linear at high
currents - the beta falls off. Running an output
stage class A approximately doubles the current that the
output stage has to handle. So, you move the operating
point way out on the output devices. (2) Less SOA
from the perspective of
the load, because
the output devices are pulling so hard against each
other.


** What a load of complete DRIVEL !!!!

Arny - leave commenting on power amp design to people
who know something about it cos they spend their lives
dealing with it at component level.
Cos YOU do not have a bloody clue.

Phil, its real handy for that you seem to be so
mentally incapable of framing a proper technical
reply. Everybody is going to dismiss your ranting.

I'd personally like to see you claim that power
transistors get more linear
when run at very high currents and that there is no
such thing as beta fall-off at high currents.

Oh there is but since a classic Class A output has such
high standing dissipation you use more devices in
parallel and you're actually working them at
lower currents than in A/B.



** Correct.

Arny has made the same ASININE error in FOUR posts.

And as is your style, you've done nothing but posture about it, Phil. Why
not show us a worked-out example? You're capable of it, I've seen you do it.

A "class A amplifer" is one DESIGNED to work correctly
in that mode.

Whatever that means.

His wacky claim about peak device currents being higher
in class A is FALSE.

It's not wacky at all - I've seen it on the bench. It makes perfect sense.
The maximum current through an output device at peak current is increased
when you increase the quiescent current. Something about Kirchoff's law as
applied to the junction of the top and bottom of the output stage, and the
load.

[Eeyore]

Arny Krueger wrote:

"Eeyore" wrote in
Arny Krueger wrote:
"Eeyore" wrote
Phil Allison wrote:
"Eeyore"

In fact I've a sort of grudge to bear in fact.

** So THAT is what " Eeyore " really is -

a bear with some sort of grudge ??

Long story. Only a small grudge really. I must learn
to lie in job interviews too.

** Oh - that job Doug got with Soundcraft ?

Uhuh.


He did the power amp stage of their "PowerStation".

Joke isn't it ?

What's wrong with it?

Do tell!

It's fairly 'routine'. That's the joke.

IOW, despite the use of a "big name" designer, the wine came out tasting
pretty ordinaire.

Is it class G?

The one I saw was just A/B.


I notice that Self makes some heavy claims about a proprietary implemention
of class G.

I haven't heard that. I'd be curious to know how it can be made proprietary !

Graham

[Eeyore]

Arny Krueger wrote:

"Eeyore" wrote in
Arny Krueger wrote:
"Eeyore" wrote

My intention is to entirely eliminate this with a kind
of 'hybrid' output stage.

How? Other than class-A you'd need to come up with some
kind of Gm-halving circuit.

It is a 'form' of Class A that indeed meets the classic
definition but without the very high idle current. It
involves quite a radical rethink of the output stage.

In the past the usual approach to this has been to use a bias circuit that
keeps the output devices from ever turning fully off.

It's that kind of thing.

What do you know about previous examples of this. I've just generally heard
that they hadn't been brilliantly succesful.

Graham

[Ruud Broens]

"Eeyore" wrote in message
...
:
:
: Arny Krueger wrote:
:
: "Eeyore" wrote
: Arny Krueger wrote:
: "Eeyore" wrote
:
: I'm concerned about the transconductance 'modulation'
: with signal level since this introduces non-linearity.
:
: Again, covered in
: http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm
:
: Which is what I'm designing out.
:
: Although I'd hoped this thread might be more about any
: audible benefits of Class A rather than a discussion of
: design principles ( I'm quite au-fait with those ! ).
:
: Like I said before, I've never done any DBTs involving Class-A amps. This is
: partially because I've never seen a true Class A amp that was capable of
: interesting power levels, in real life.
:
the Manley 500 W monoblocks, using 10 KT90's, do 275 W in triode mode,
500W in penthode mode, presumably very substantially class A
it *does* use a staged power switch on
at steal at about 150 USD/kilo
hehe
(over on RAT, Patrick Turner reported working on a 6*GM70 design,
that could definitely do 100 % class A all the way to 275 W

Rudy

: Fair enough.
:
: I've heard enough positive comment on them to pursue my line of thinking on the
: matter already btw.
:
: Graham
:
:

[Arny Krueger]

"Phil Allison" wrote in message

"Arny Krueger"


Agreed again. Building a truely Class A power amp that
can deliver significant amounts of power output is
really a pretty awesome thing. The OPT stage quiescent
current has to be equal to the *peak* current that is
delivered to the load.

** Absolutely FALSE.

The peak load current is *exactly* double the idle
current for an amp operating in class A.

Agreed.

** Self contradiction - right here.

Oh, I get it, this time.

Yeah, Phil I had things wrong.

Had you not been so extreme and violent in your initial response, I would
have probably gotten it, the first time.

[Eeyore]

Arny Krueger wrote:

"Eeyore" wrote
Arny Krueger wrote:

I'd personally like to see someone claim that power
transistors get more linear when run at very high
currents and that there is no such thing as beta
fall-off at high currents.

Oh there is, but since a classic Class A output has such
high standing dissipation you use more devices in
parallel and you're actually working them at lower
currents than in A/B.

I think I saw the effects of beta fall-off in my testing. I was working
with 8 devices up and 8 devices down. VCC was about 80 volts. The devices
were complements from the MJE 1502x series.

Admittedly, one go add enough devices so that beta fall-off wouldn't be an
issue.

Those devices start showing significantly reduced beta @ around 3A.

In any case the output topology I have in mind is quite insensitive to beta.

Graham

[Bertie the Bunyip]

Eeyore wrote in
:



Trevor Wilson wrote:

"Eeyore" wrote in message
Trevor Wilson wrote:

IOW: The design is more important than the Class of operation.
Class A will
help a bad design and, at best, do nothing to a good design.

You're assuming they all start as Class AB output stages there.

**That's because there are essentially no Class A push pull designs
(anymore). They're all Class A/B with different bias currents.

That's where I plan to differ.

Now. To get back to the original question.... Have you heard a
noticeable difference between Class A and A/B ?

Graham


Fjukkwit netkkkkop



bertie

[Bertie the Bunyip]

Eeyore wrote in
:



Trevor Wilson wrote:

"Eeyore" wrote in message
Trevor Wilson wrote:
"Eeyore" wrote in message
Trevor Wilson wrote:

IOW: The design is more important than the Class of operation.
Class A will help a bad design and, at best, do nothing to a
good design.

You're assuming they all start as Class AB output stages there.

**That's because there are essentially no Class A push pull
designs (anymore). They're all Class A/B with different bias
currents.

That's where I plan to differ.

Now. To get back to the original question.... Have you heard a
noticeable difference between Class A and A/B ?

**Not in a properly designed Class A/B amp. In a poorly designed
Class A/B amp, increasing bias will make it better.

So do tell me, what happens to the delta Vbe in an A/B output stage ?
( note its relationship to current )

Graham



fjukkwit netkkkop



bertie

[Eeyore]

Arny Krueger wrote:

"Phil Allison" wrote in message

snip


Arny has made the same ASININE error in FOUR posts.

And as is your style, you've done nothing but posture about it, Phil. Why
not show us a worked-out example? You're capable of it, I've seen you do it.

A "class A amplifer" is one DESIGNED to work correctly
in that mode.

Whatever that means.

I think it means "A "class A amplifer" is one DESIGNED to work correctly in
that mode. "


His wacky claim about peak device currents being higher
in class A is FALSE.

It's not wacky at all - I've seen it on the bench. It makes perfect sense.
The maximum current through an output device at peak current is increased
when you increase the quiescent current. Something about Kirchoff's law as
applied to the junction of the top and bottom of the output stage, and the
load.

Obfuscation.

Graham

[Bertie the Bunyip]

Eeyore wrote in
:



Trevor Wilson wrote:

"Eeyore" wrote in message
Trevor Wilson wrote:
"Eeyore" wrote in message
Trevor Wilson wrote:

IOW: The design is more important than the Class of operation.
Class A will help a bad design and, at best, do nothing to a
good design.

You're assuming they all start as Class AB output stages there.

**That's because there are essentially no Class A push pull
designs (anymore). They're all Class A/B with different bias
currents.

That's where I plan to differ.

Now. To get back to the original question.... Have you heard a
noticeable difference between Class A and A/B ?

**Not in a properly designed Class A/B amp. In a poorly designed
Class A/B amp, increasing bias will make it better.

So do tell me, what happens to the delta Vbe in an A/B output stage ?
( note its relationship to current )

Graham



fjukkwit netkkkkop



Bertie

[Bertie the Bunyip]

Eeyore wrote in
:



Arny Krueger wrote:

"Phil Allison" wrote

** Such a power stage has a rated output of over 500
watts and so HAS to use at least 8 large output devices.

Irrelevant.

Very relevant indeed since each device is working at a fraction of the
total load current.

Graham


Lying kkjunt




Bertie

[Bertie the Bunyip]

Eeyore wrote in
:



Arny Krueger wrote:

We all know that speakers can be capacitive-reactive and can jack the
load current to unexpectedly high levels.

Entirely calaculable though.


A common power amp will have +/- 80
volt DC rails and may need to deliver up to 15 amps peak to the load.

15 Amps ? I'd design for 40 !

Graham



fjukkktard wannabe



Bertie

[Arny Krueger]

"Eeyore" wrote in
message
Arny Krueger wrote:

"Eeyore" wrote
in
Arny Krueger wrote:
"Eeyore" wrote

My intention is to entirely eliminate this with a kind
of 'hybrid' output stage.

How? Other than class-A you'd need to come up with
some kind of Gm-halving circuit.

It is a 'form' of Class A that indeed meets the classic
definition but without the very high idle current. It
involves quite a radical rethink of the output stage.

In the past the usual approach to this has been to use a
bias circuit that keeps the output devices from ever
turning fully off.

It's that kind of thing.

What do you know about previous examples of this.

Used in some Japanese amps from maybe the late 70s and early 80s.

I've just generally heard that they hadn't been brilliantly
succesful.

I think the usual phrase used to describe this is "sliding bias".

It seems to me that sliding bias can make power amps less reliable, by
turning minor faults into situations where lots of power is dissipated in
the output stage. The trick might be to back out the sliding bias feature
when things start going awry, like a shorted load or some such.

Graham

[Bertie the Bunyip]

Eeyore wrote in
:



Arny Krueger wrote:

"Eeyore" wrote
Arny Krueger wrote:

We all know that speakers can be capacitive-reactive and
can jack the load current to unexpectedly high levels.

Entirely calaculable though.


A common power amp will have +/- 80
volt DC rails and may need to deliver up to 15 amps peak
to the load.

15 Amps ? I'd design for 40 !

OK.

In your opinion Graham, what would the quiescent dissipation of a
traditionally-designed output stage like this be?

Phil's already covered that and I'm in accord with what he says.

Graham




fjukkwit netkkkop



Bertie

[Bertie the Bunyip]

Eeyore wrote in
:



Arny Krueger wrote:

"Eeyore" wrote
Arny Krueger wrote:

We all know that speakers can be capacitive-reactive and
can jack the load current to unexpectedly high levels.

Entirely calaculable though.


A common power amp will have +/- 80
volt DC rails and may need to deliver up to 15 amps peak
to the load.

15 Amps ? I'd design for 40 !

OK.

In your opinion Graham, what would the quiescent dissipation of a
traditionally-designed output stage like this be?

Phil's already covered that and I'm in accord with what he says.

Graham




fjukkwit netttkkkkop

Bertie

[Bertie the Bunyip]

Eeyore wrote in
:



Scott Dorsey wrote:

Trevor Wilson wrote:

**There are plenty of reasons NOT to go pure Class A and very few to
do so.

Other than size and heat?

That's what I'm working on you see !

Graham



lying kjuuunnntt



Bertie