Current Feedback IC input in a power amp

[steph_tsf]

Unfortunately if one tries to embed a conventional op-amp, it all gets unstable.

At first glance, the attached Tina 7 Texas Instruments simulation file works fine :
- Analysis / DC Analysis / Table of DC results : everything okay
- Analysis / AC Analysis / AC Transfer Characteristic : very nice bandwith

With a 0,5V sinus input at 1 KHz, the Fourier Analysis of VOUT gives about 0,003% distorsion. Quite decent, isn't ?

However, with a 0.5V squarewave input at 10 KHz, the simulations delivers an oscillation.

I have tried different things, for getting more useful results, but no luck.

Any help appreciated.

[steph_tsf]

Finally I managed to embed a conventional op-amp, and get it stable.
See attached Tina 7 Texas Instruments simulation file : it is working fine.
But the frequency compensation is ugly, and not possible using one IC. We don't have access to the terminals where the frequency compensation gets applied.
With a 0,5V sinus input at 1 KHz, the Fourier Analysis of VOUT gives about 0,028% distorsion. This is 10x worse than without the frequency compensation scheme.

We better don't use a OPA-134 in such topology.

The true current feedback approach arrangement seems to be more efficient.
It is amazing to see the level of performance delivered by a few discrete (well matched and thermally coupled) transistors.

Next try will be using a AD8033 IC as front-end. It is not a current-feedback device, but it is much faster than the OPA-134.
How to simulate an Analog Devices AD8033 FastFET Op Amp with 80 Mhz bandwith and 80V/µs slew rate ? What topology ?

I should run a test of my OPA-134 implementation alone. Maybe my simulated OPA-134 already has the speed of the AD8033. This would mean that using an IC as front-end is a dead end.

Any suggestion welcome.

[Cauldron Amplifiers]

Hi, Just stumbled across your thread while looking for people who would like to build an input buffer stage using discrete. I have a nice little buffer amp I designed, but maybe a different thread would be better.
In any case: I took a peek at where your going on your current design process, and offer this example of a clean way to do an IC front end, using current feedback in the main amplifier.
The first image is the schematic proper. The second image details small signal bandwidth and phase. And the third image is an FFT of distortion at 100hertz.
This is a nice clean topology, and is amendable to the OPA134, or OPA627, with small changes in compensation.
Enjoy!

Paul Vonharnish
Phoenix/Cauldron amplifiers

[steph_tsf]

Originally Posted by Phoenix Audio

This is a nice clean topology, and is amendable to the OPA134, or OPA627, with small changes in compensation.

Hello, thanks for your reply. At first glance, it looks more elaborate than the Kuroda 1982 amplifier. See here Kuroda TL071 60W FET Amp (1982)
I'm not talking about the output stage, but about the way the opamp gets interfaced with the VAS. Did you know about the Kuroda 1982 amp upon designing yours ? Did you know about the Alexander amp ? If yes/yes, can you describe the features of your implementation ? Thanks. Steph.

[Bonsai]

If you read th e Alexander amp app note mentioned earlier, you will see that if you return the compensation cap, which in a conventional CFA goes from the VAS output to GND, back to the feedback summing junction, you get quite an improvement in phase margin.

However, looking quickly at your first cct diagram in your initial post, this looks like a voltage feeback topology. CFA will look a lot different. Take another look athe Alexander amp and how the feedback is configured - this should help point you in the right direction.

[Cauldron Amplifiers]

Hi Steph,
In answer to my knowledge regarding the amplifier topologies you mention: No, and no. ;-) At first blush, nether amplifier would appear to be very desirable in an audio application. I see lots of probability for ringing and overshoot, and generally don't care to use FET's, unless they are included in an actual IC. Fet matching is always a serious problem in discrete design, and frankly, I don't care for the sonics either.

The amplifier I published is my own take on the Harman Kardon Citation 16 of yore. (Circa 1976) The input to the VAS section is primarily through Q5 and Q6, which are biased by cascode current sources. A person could go with some simpler compound current sources, but I prefer the cascode treatment.
The VAS gain in my example is controlled primarily by R26, R27, R28, R29, and compensation is augmented by C13, C14.

I also see potential power supply rejection issues in the examples you mentioned, whilst the topology I posted yields supply rejection figures in the
-180db range.

Also, I believe a person could drive an IC section with a simple input buffer, (somewhat like the example I will post here) but feedback in closed loop mode, would need further attention.

Enjoy!

Paul Vonharnish
Phoenix/Cauldron Amplifiers

[Cauldron Amplifiers]

Hello again,

Didn't realize you also used LT spice, so here are zip files of the two schematics I posted. Hope you find them helpful.

Paul Vonharnish
Phoenix/Cauldron Amplifiers