quiescent current in audio MOSFET class AB output stage

[Mac]

On Wed, 10 Dec 2003 09:12:48 +0000, Ban wrote:

[snip]

The LM833 is utterly unsuitable to drive a MOS output stage, with a 2k load
the O/P voltage swing is only +/-10V.

I just looked over the datasheet again, and I noticed that with a 10k
load, which should be doable, it swings +/- 12, which is just enough, I
think.

Best regards,

Mac

 

[Kevin McMurtrie]

What is the best way to set the quiescent current in a class AB MOSFET
output stage? I plan to use an LM833 for the gain stage with feedback from
the output back to the LM833, so I am not worried about offset drift or
anything like that. I am a little worried about the quiescent current
getting out of hand from a heat dissipation perspective.

I plan to use +/-15 volt rails and am shooting for 10 watts into 8 ohms.

Is it safe to just use a trim pot in the biasing circuit, to set the
quiescent current or is it liable to drift a lot if I do that?

How about putting small source resistors in combination with the hand
trimming approach?

What I would really like is a way to get a reasonably bounded quiescent
current by design without trimming. Again, I don't care if the current
drifts a bit, as long as the power consumption doesn't get out of hand.
I'm thinking that 200 mA is a reasonable upper limit.

Best regards,

Mac

There some things to consider:
Cold bias current
Hot bias current
Crossover distortion compensation with negative feedback

With any luck, you'll find the magic current that works cold without
distortion and hot without runaway heating. Increasing the Source
follower resistors makes it more stable at the cost of efficiency. Gate
to Source bypass resistors can reduce crossover distortion at the cost
of a greatly increased drive load (can't do that with LM833).

Between the >4V losses in the LM833 and the >5V losses in MOSFET gates,
you're looking at pretty much no output from +/- 15V. You'll need
boostraping or another pair of rails with a higher voltage op-amp.

 

[Tony Williams]

Study the Transfer Characteristics plots in the data sheet of
your favorite amplifier FET, and unless it's a lateral type,
you'll see what I'm talking about. You'll see crossing curves
at two or three temperatures. For Vgs below the crossover gate
voltage the tempco is positive, above it's negative. OK, now
notice the high drain current at crossover. Consider what the
power dissipation would be for a linear power-amp FET with say
50V operating at that current... Ouch! Impossible!

It is "Fig. 4" on almost every IR Hexfet data sheet.

We had a short thread here last year where some chap
stabilised a Class AB push-pull rf power amp by
extracting the typical dVgs/dT (for a constant Id)
from his Fig. 4 and used a thermistor the generate a
compensating dV/dT of the gate bias.

He had two identical n-types, so could use one thermistor,
but a p-type and n-type pair would probably each need
their own temperature-sensitive biassing arrangement.

 

[default]

default wrote...

You probably made a good choice. What are the part numbers?

Thanks,
- Win

whill_at_picovolt-dot-com

2SJ49 and 2SK134

 

[John Woodgate]

I read in sci.electronics.design that Walter Harley
com>) about 'quiescent current in audio MOSFET class AB output stage',

Although, as I said elsewhere, it doesn't talk much about MOSFETs except
to assert their inferiority to BJT's.

It's not just an assertion; he shows credible evidence. But IMHO, it's
not conclusive.

 

[Walter Harley]

It's not just an assertion; he shows credible evidence. But IMHO, it's
not conclusive.

The commercial success of both kinds of output stage demonstrates that each
is viable, though one might be preferable in some regards.

I didn't mean the term 'assertion' with any pejorative connotations! I just
didn't want to make a stronger statement than was necessary, in lieu of Mac
actually reading the book. Self can speak for himself just fine; he makes a
well-reasoned, well-tested, and articulate case.

I would say that the main criticism one might level against Self's book
would be one that he himself addresses, namely that the pursuit of
vanishingly small distortion is not necessarily a useful design approach in
terms of pleasing consumers, marketers, or bean-counters. The book might
better be titled "Handbook of Very Low Distortion Audio Power Amp Design."

(Personally, I also have some criticisms of Self's approach on the broader
topic of objectivism in audio. I think that he occasionally overstates his
case. But that's minor in comparison to the value that he provides.)

 

[John Woodgate]

I read in sci.electronics.design that Walter Harley
com>) about 'quiescent current in audio MOSFET class AB output stage',

The commercial success of both kinds of output stage demonstrates that each
is viable, though one might be preferable in some regards.

H'mm. Not necessarily. We had a lot of 'successful' amplifiers with
slewing troubles, until Matti Otala blew the whistle. Unfortunately,
people who already knew about the problem suggested that he used an
illegal pea. (;-)

The same applied to certain pickup cartridges, with very much better
reputations than linearity.

But you are right in this case, I think.

I didn't mean the term 'assertion' with any pejorative connotations!

Understood, but other might have thought so. It sounded as though he
didn't present any evidence.

I just
didn't want to make a stronger statement than was necessary, in lieu of Mac
actually reading the book. Self can speak for himself just fine; he makes a
well-reasoned, well-tested, and articulate case.

I would say that the main criticism one might level against Self's book
would be one that he himself addresses, namely that the pursuit of
vanishingly small distortion is not necessarily a useful design approach in
terms of pleasing consumers, marketers, or bean-counters.

Consumers and marketroids are very pleased with ridiculous distortion
figures.

The book might
better be titled "Handbook of Very Low Distortion Audio Power Amp Design."
Agreed.

(Personally, I also have some criticisms of Self's approach on the broader
topic of objectivism in audio. I think that he occasionally overstates his
case. But that's minor in comparison to the value that he provides.)

Also agreed. I have often been 'pleased' by reproduction that I knew
was very significantly inaccurate. 'Pleasing' and 'accurate' are not 1:1
related.

 

[Mac]

There some things to consider:
Cold bias current
Hot bias current
Crossover distortion compensation with negative feedback

With any luck, you'll find the magic current that works cold without
distortion and hot without runaway heating. Increasing the Source
follower resistors makes it more stable at the cost of efficiency. Gate
to Source bypass resistors can reduce crossover distortion at the cost
of a greatly increased drive load (can't do that with LM833).

Between the >4V losses in the LM833 and the >5V losses in MOSFET gates,
you're looking at pretty much no output from +/- 15V. You'll need
boostraping or another pair of rails with a higher voltage op-amp.

Well, it's a bit premature of me to say so, since I haven't worked out all
the biasing details yet, but my plan was to allow the LM833's output to be
pretty close to 0 V when the amp's output is at 0V. Then I'd have
capacitors connected to the bias chain which would allow me to drive the
gates above and below the supply voltages during the peak swings.

So the LM833 headroom reduction will be more like +/- 3 V, and the Vth
none at all. That leaves +/- 12 V of swing, assuming 15 V rails. That's
enough for 10 Watts.

By the way, how did you come up with 5 V of loss for the MOSFETS? Vth is
min 2V and max 4V. Of course, that's for some ridiculously low current,
but still...

Anyway, the way things are going, I may just add small source resistors and
adjust the bias with a potentiometer when the amp is good and warm.

I plan to take overall feedback from the amp output back to the LM833 (or
whatever). I wasn't planning on using any other negative feedback.

Thanks for the inputs.

regards,
Mac

 

[Mac]

Douglas Self's "Handbook of Audio Power Amplifier Design". It will really
open your eyes to a lot of issues you might not have thought about (at
least, it did mine), and at the same time it is very concrete, readable, and
contains schematics that you can build yourself.

Although, as I said elsewhere, it doesn't talk much about MOSFETs except to
assert their inferiority to BJT's. But even if you're committed to building
with MOSFETs, I would recommend it; many of the issues are the same.

-w

Thanks. Maybe I can by myself a copy for Christmas. ;-)

Mac