Discrete opamp bjt front end

[martin griffith]

Ok, this is a classic audio doofer
http://johnhardyco.com/pdf/990.pdf
I was just wondering about the emiiter L's on the LM394s

1) they would seem to reduce the gain at MHzee bandwidthiums

2) ISTR that the LM394 could be a bitch to work with, something to do
with capacitance

3) In real life/breadboard what would be the best way of measuring the
effectiveness of the emitter inductances with normal
transistors(2N4403 etc) as the LTP

4) Dilbert archives here http://pag.csail.mit.edu/~adonovan/dilbert/


martin

 

[Eeyore]

Ok, this is a classic audio doofer
http://johnhardyco.com/pdf/990.pdf
I was just wondering about the emiiter L's on the LM394s

1) they would seem to reduce the gain at MHzee bandwidthiums

Yup. It may also help the phase response at higher frequencies too.

2) ISTR that the LM394 could be a bitch to work with, something to do
with capacitance

Quite likely. It is a large geometry device. Ccb is non-linear too which has
some interesting effects.

3) In real life/breadboard what would be the best way of measuring the
effectiveness of the emitter inductances with normal
transistors(2N4403 etc) as the LTP

Do a bode plot ?

You can get better transistors for this than 4403s btw. 2SA1084 for example.

Graham

 

[martin griffith]

Yup. It may also help the phase response at higher frequencies too.



Quite likely. It is a large geometry device. Ccb is non-linear too which has
some interesting effects.



Do a bode plot ?

You can get better transistors for this than 4403s btw. 2SA1084 for example.

Graham

2sa1316 as well, time to call Profusion. But I was wondering if the
L's will improve RF rejection, or it might be worth putting ferrite
beads on the Bases of the LTP as well


martin

 

[Fred Bartoli]

martin griffith a écrit :

2sa1316 as well, time to call Profusion. But I was wondering if the
L's will improve RF rejection, or it might be worth putting ferrite
beads on the Bases of the LTP as well

The 20uH aren't there for RF blahhh, but just to reduce the 2x30.1R
noise contribution in the audio band.

The LM394 are biased at 2mA, so have a 13R Re, which is probably high
given the LM394 high parasitics.
Emitter degeneration helps stability and also slew rate figures
(remember the BJT vs JFET diff pair in opamps?), but 60R is also just
1nV/rtHz and this will bring the total input noise from 1.13nV/rtHz to
1.5nV/rtHz or about 3dB worse.

Note that the 30R/20uH corner freq is 240kHz, well over the audio BW,
but higher inductance values wouldn't have been practical.
An welcomed side effect of this is a pole-zero pair which boost the GBW
product in the audio band (below 100kHz) by (30+13)/13 and reduce THD by
10 more dB than without the inductors.

 

[martin griffith]

martin griffith a écrit :

The 20uH aren't there for RF blahhh, but just to reduce the 2x30.1R
noise contribution in the audio band.

The LM394 are biased at 2mA, so have a 13R Re, which is probably high
given the LM394 high parasitics.
Emitter degeneration helps stability and also slew rate figures
(remember the BJT vs JFET diff pair in opamps?), but 60R is also just
1nV/rtHz and this will bring the total input noise from 1.13nV/rtHz to
1.5nV/rtHz or about 3dB worse.

Note that the 30R/20uH corner freq is 240kHz, well over the audio BW,
but higher inductance values wouldn't have been practical.
An welcomed side effect of this is a pole-zero pair which boost the GBW
product in the audio band (below 100kHz) by (30+13)/13 and reduce THD by
10 more dB than without the inductors.

Hi Fred,
I'm confused when you say that

60R is also just
1nV/rtHz and this will bring the total input noise from 1.13nV/rtHz to
1.5nV/rtHz or about 3dB worse.

I was thinking that at Low Frequencies the 30.1 R's would be swamped
by the DCR of the parallel inductors, say >1R



martin

 

[Fred Bartoli]

martin griffith a écrit :

Hi Fred,
I'm confused when you say that
I was thinking that at Low Frequencies the 30.1 R's would be swamped
by the DCR of the parallel inductors, say >1R

Yep. What I meant and maybe incorrectly said, is that the 30.1R are
needed for slew rate and stability purpose. But if you let them straight
then they contribute 3dB more noise. Shunting them with the inductors
brings you back to the 1.13nV/rtHz figure and also boost loop gain at
frequencies below 100kHz.

 

[Ban]

Ok, this is a classic audio doofer
http://johnhardyco.com/pdf/990.pdf
I was just wondering about the emiiter L's on the LM394s

1) they would seem to reduce the gain at MHzee bandwidthiums

2) ISTR that the LM394 could be a bitch to work with, something to do
with capacitance

3) In real life/breadboard what would be the best way of measuring the
effectiveness of the emitter inductances with normal
transistors(2N4403 etc) as the LTP

4) Dilbert archives here http://pag.csail.mit.edu/~adonovan/dilbert/


martin

That datasheet seems to be written for those phools, who believe any
technobrabble, Diagrams and important data are missing instead. Some
important data like input bias=2.2uA is listed, but not mentioned in the
text, making it unsuitable for a lot of applications.
Offset voltage and -stability can only be guessed. The noise is mentioned
for the input device only and the capacitor drivel can be read better in the
AVX datasheets.
Beware of shorts, there seems no output protection!
That thing is IMHO useful only for driving transformers, inductive
backcurrent can be absorbed.
The inductors are probably in there for your point 2), they might give some
extra phase margin, but will increase distortion. I do not like those
speed-up diodes across the inputs either.
But I do like the gold plated pins!

 

[Jim Thompson]

[snip]

What a piece-a-crap.

I've got to get into this audiophool business... looks like all kinds
of dummies sitting there ready for the taking ;-)

...Jim Thompson

 

[Jan Panteltje]

[snip]

What a piece-a-crap.

I've got to get into this audiophool business... looks like all kinds
of dummies sitting there ready for the taking ;-)

Well, I dunno, count me in as one if them if you must, nice sales talk
in that pdf, but the thing does what it does.
What is wrong except specifying resistors in Watts, not Ohms?

 

[Jim Thompson]

martin griffith wrote:
Ok, this is a classic audio doofer
http://johnhardyco.com/pdf/990.pdf

[snip]

What a piece-a-crap.

I've got to get into this audiophool business... looks like all kinds
of dummies sitting there ready for the taking ;-)

Well, I dunno, count me in as one if them if you must, nice sales talk
in that pdf, but the thing does what it does.
What is wrong except specifying resistors in Watts, not Ohms?

Well, as one who designs another OpAmp into a chip every few months,
that "design" is junk.

...Jim Thompson

 

[Jan Panteltje]

martin griffith wrote:
Ok, this is a classic audio doofer
http://johnhardyco.com/pdf/990.pdf
[snip]

What a piece-a-crap.

I've got to get into this audiophool business... looks like all kinds
of dummies sitting there ready for the taking ;-)

Well, I dunno, count me in as one if them if you must, nice sales talk
in that pdf, but the thing does what it does.
What is wrong except specifying resistors in Watts, not Ohms?

Well, as one who designs another OpAmp into a chip every few months,
that "design" is junk.

OK, so you are responsible for all those opamps ;-).
There are now so many of those that it is really hard to make a choice.

That design reminds me of some power amps (well headphone power) from
the sixties.
But it should work, I do not see what is 'wrong' with it technically.
Those things (audio people) like to use this sort of thing to drive
transformers.
Now that is a whole different subject, as I got in the seventies a request
to design an audio distribution amp, and my view was that you had to
terminate cables by their characteristic impedance, but the audio guys
did not.
But I was driving those transformers with very much a similar circuit.
Except for the inductors.

So what part of that circuit is junk?

 

[Jim Thompson]

On a sunny day (Wed, 29 Nov 2006 11:15:07 -0700) it happened Jim Thompson
<[email protected]>:

Sunny, but COLD... it got down to 70°F in the house last night, and a
cold wind is blowing today ;-)

[snip]

So what part of that circuit is junk?

Class-B stage for a start. Cruddy bias methods for another.

I'll have to surf, but I think you can buy 0.8nV/rt-Hz off-the-shelf.
If not, adding a bipolar front-end to a single-chip OpAmp is a better,
easier, cheaper way to go.

...Jim Thompson

 

[Jim Thompson]

On a sunny day (Wed, 29 Nov 2006 11:15:07 -0700) it happened Jim Thompson
<[email protected]>:
[snip]

Well, as one who designs another OpAmp into a chip every few months,
that "design" is junk.

OK, so you are responsible for all those opamps ;-).
There are now so many of those that it is really hard to make a choice.

[snip]

Now-a-days, all my OpAmp designs are going into custom ASIC's, not
into jelly-bean stuff.

...Jim Thompson

 

[Jan Panteltje]

Sunny, but COLD... it got down to 70°F in the house last night, and a
cold wind is blowing today ;-)

Here it is winter with summer temperatures, and the first mammoths are beginning
to leave their nests.
That is that noise you hear.

[snip]

So what part of that circuit is junk?

Class-B stage for a start. Cruddy bias methods for another.

So you want class A, or even class D?

I'll have to surf,

Would not do that with those temperatures where you are.

but I think you can buy 0.8nV/rt-Hz off-the-shelf.

Noise comes free here.

If not, adding a bipolar front-end to a single-chip OpAmp is a better,
easier, cheaper way to go.

Yea, I am a bit out of touch with audiophile noise specs.
Are we 48 or 64 bits now?

 

[Eeyore]

Beware of shorts, there seems no output protection!

Actually there is.

Graham

 

[Eeyore]

[snip]

What a piece-a-crap.

I've got to get into this audiophool business... looks like all kinds
of dummies sitting there ready for the taking ;-)

...Jim Thompson

It's an old and very classic design.

It was indeed designed to drive *;arge* transformers.

Graham

 

[Jim Thompson]

On a sunny day (Wed, 29 Nov 2006 12:04:55 -0700) it happened Jim Thompson
<[email protected]>:
[snip]

I'll have to surf,

Would not do that with those temperatures where you are.

Yep. It's 54°F right now, with 17MPH winds.

I'm trying to put up outdoor Christmas decorations in my shirt
sleeves... guess I'd better put on a jacket ;-)

Noise comes free here.


Yea, I am a bit out of touch with audiophile noise specs.
Are we 48 or 64 bits now?

I wouldn't know either... all I own is a JVC "home theater system",
and I'm going deaf anyway :-(

Though I still can hear distortion, particularly in classical
pieces... used to play clarinet, so I know what a woodwind ensemble
_should_ sound like ;-)

...Jim Thompson

 

[Fred Bartoli]

martin griffith a écrit :

4) Dilbert archives here http://pag.csail.mit.edu/~adonovan/dilbert/

One clicked at random from all the list...

http://pag.csail.mit.edu/~adonovan/dilbert/show.php?day=6&month=2&year=1999

 

[Jim Thompson]

martin griffith a écrit :


One clicked at random from all the list...

http://pag.csail.mit.edu/~adonovan/dilbert/show.php?day=6&month=2&year=1999

Bwahahahaha! Good one!

...Jim Thompson

 

[Eeyore]

Class-B stage for a start. Cruddy bias methods for another.

I'll have to surf, but I think you can buy 0.8nV/rt-Hz off-the-shelf.
If not, adding a bipolar front-end to a single-chip OpAmp is a better,
easier, cheaper way to go.

...Jim Thompson

Op-amps typically won't deliver more than +- 30mA.

Graham