help needed: hi Zin, lo Zout, hi swing driver for experiments

[aletheian]

Hello.

I am experimenting with a few different discrete power amp topologies/zero-gain power buffers and followers, and I need to piece together a driver circuit so that I can take output measurements on scope, computer and meter. I don't need a spoon-fed schematic, but if there are any ideas for a general topology, I (with the help of SPICE and my breadboard) can take it from there... unless someone has a killer circuit squirreled away!

Here are the general design parameters:
-single polarity output is fine... for now. I can build two and invert the input if need be for differential signals

-High input impedance (at least hundreds of k-ohms) so that I can drive it from a variety of sources (signal generators, soundcard, CD-player, various preamps, etc.) without loading or distortion

-low output impedance in case I need to drive a circuit with a high input capacitance

-LOTS of voltage swing (probably > a dozen volts) so that I can drive, for example, a complimentary MOSFET output stage, etc to full output... and even beyond the rails for torture testing

-low enough distortion not to interfere with FFT measurements

-low enough noise not to interfere with noise measurements


I would very much like to keep it in the discrete realm and avoid ICs if possible. I have no desire to run around sourcing parts and paying top-dollar for exotic opamps... besides, what is the fun in that... AND I tried a few already based on 'buffer->ne5534->buffer' and also 'opa2134->buffer' and they don't have enough swing (they got ugly over 10v or so) + drive for low impedance circuits in and of themselves. I have tried a few typical data-sheet-application-note instrumentation amplifier circuits, but none of them seem to have a wide enough clean swing. I won't bore you with the circuits that I tried, but they are pretty typical. I got closest with a high GM FET input feeding an emitter follower with a decent amount of feedback, but I could not get it to swing enough without taking the rails out of safe territory for the parts I had.


Anyway, thank you for taking the time to feed my time-consuming hobby!!

[lineup]

So that is what you want

Hehe.
That is what we all want, isn't it ....
No kidding, but there are some really advanced preamplifiers.

PMA - Pavel Macura is in the same division as you.
See his DISPRE = discrete pre amplifier.
Unfortunately all details are not public. Probably cost a small sum of money.

A discrete version of AD797 would come a little bit towards your goal.
I have done some AD797 clones schematics .. even really high voltage versions.

jcx is our forum expert on AD797 topology.
He has studied and corrected some minor 'defects' of the Analog Devices chip version.
Along with Scott Wurcer, one of the original designer of AD797.

There are other extreme creators.
G Kleinschmidt is one and several others, too.

John Curl is perhaps The King .. but he can not tell you his best secrets.
But he gives us hint after hint ..without any serious details revealed, of course

I hope they all post here, to get you started
( as if you hadn't already a plan )

Regards, Lineup

[aletheian]

HA! thanks for the pointers Lineup. I actually hadn't looked at it THAT way, but I suppose you are right. I don't relish the thought of building a discrete opamp to run on a pair of 50v rails, but I guess when all is said and done... who knows.

I didn't consider AD797 because I honestly don't need performance THAT good (I honestly considered using an LM3886 with an input driver for a brief moment before discarding the idea), and I would need to jack the Vout swing up regardless... but I'll rifle through posts and also application notes for it to see what I can turn up.

Thanks again.

[AndrewT]

do a search on "discrete opamp".
I got 682 threads as results.

B1 and Pedja and Dispre and various PASS references all appeared on the first page.

[DigitalJunkie]

Is this for single-ended,or P-P topology?

I know it's a chip,but perhaps an LM4702/LME49810 or similar,with a driver/buffer after it would work well?
There's plenty of swing,and the driver/buffer after it can be built to supply enough current for anything you might need.
Just a quick thought.

[aletheian]

Originally posted by DigitalJunkie
Is this for single-ended,or P-P topology?

Both/either/neither. I just need to build a test jig that can cover all bases.

I know it's a chip,but perhaps an LM4702/LME49810 or similar,with a driver/buffer after it would work well?
There's plenty of swing,and the driver/buffer after it can be built to supply enough current for anything you might need.

I had forgotten about those! I'll keep that in mind in case I can't get anything else up and running. An IC takes the DIY fun out of building it, but at least the performance would be predictable. They seem to have a very narrow usable range though within their intended use.

Thank you for the thoughts.