Counterpoint - A Fully Symmetrical I/V stage

[Russ White]

Hello Folks,

Some of you may already know I have had a new I/V stage in the works for a while now. Some of you have probably been a bit frustrated it took me so long, but this is has been a labor of love and I have poured a lot of thought and care into the design.

The initial design was a voltage feedback design similar to a THS4131. This design is more like (If I am not mistaken) a current feedback design. It is not Super Symmetrical in that the feedback in not taken at the bases of the input transistors, but rather at the emitters, but it still is fully symmetrical in operation. It is not really an opamp per se either, but a true transimpedance stage. While it does employ feedback, it is not subject (at least to a great degree) to the same negative issues that plague most opamp based I/V designs.

Here were some of my basic design goals:

1) The design should not just be simply balanced, but fully symmetrical for use with balanced current output DACs. Good candidate DACs include PCM1794A, ES9008 (ESS Sabre), and many others.

2) The design should not be subject (at least less subject) to the usual slewing limitations of a inverting or conventional VFB fully symmetrical opamps.

3) The design should be approachable for DIY use. What I mean by this is it should use commonly available parts and not be expensive to make while still offering excellent results. No exotic or expensive parts are required.

4) It should be very low distortion with a high dynamic range. This it is, but I do not have any solid numbers yet. I have a volunteer who will measure it for me once I provide him the PCB. What I do know is the test circuit had THD too low for EasyLab to measure, what I mean is it was at the floor which is about 110db. So I have to have someone with better test gear test it out for me.

5) It should be unique! I have not seen anything exactly like it, but it does share similarities with other designs.

6) It should allow input voltage bias adjustment so the the input can be at AVCC/2 for chips like the ES9008 or at 0V for chips like PCM1794A.

7) It should have no DC at the outputs either between the outputs or from the outputs to GND. The integrator handles this and it also allows for single ended output without any significant THD penalty. There is a potentiometer to trim the output DC to 0V. This is important as without it you would need very very good parts matching to obtain good results. As it is, while good part matching helps it is actually not actually required.

8) It should have effective HF filtering. It does

OK enough about the design goals. Here is my cct.

I have tested the cct (on protoboard) , but I have not received the final PCB yet, when I do I will have a better idea how it sounds.

The power supply should be +/- 9-12V rails with the parts shown. I would not go above 12V without adjustments.

Cheers!
Russ

[Russ White]

Here is the PCB:

[Russ White]

Here is a simulation file for people interested in how it works.

I know some of you will want to simulate it, so I figured I would make it easy for you.

Cheers!
Russ

[PigletsDad]

Very neat.

As I understand it, it has (near) unity current gain, and a high open loop output impedance - it basically turns the DAC into a near perfect current source. The open loop bandwidth will be super wide; so it should tolerate nasty switching edges pretty well.

You then can do the filtering at the output with C4.

If you wanted, you could also do some filtering at the input, shunting the virtual earth points (X1.1 and X1.7) to ground via a cap, so that when the feedback loop eventually runs out of puff in the HF, the impedance falls, rather than rising.

This would doubtless mean ages fiddling to get it stable again!

[Russ White]

Originally posted by PigletsDad
Very neat.

If you wanted, you could also do some filtering at the input, shunting the virtual earth points (X1.1 and X1.7) to ground via a cap, so that when the feedback loop eventually runs out of puff in the HF, the impedance falls, rather than rising.

This would doubtless mean ages fiddling to get it stable again!

Thanks PD, I actually thought of this too, but I found that adding the caps at the inputs alone seems to produce the opposite of the intended effect. It actually caused more gain at high frequency, and not less... at least in simulation.

But perhaps the use of both C4 and those caps would have some positive effect, I am not sure. It does appear to remain stable in this setup. At least with 1nf at the inputs to GND.

Cheers!
Russ

[coloradosound]

nice work Russ!

[rjbaldwin]

As ever Russ, it is the care and thought you put in that most impresses me. Mind you I'm looking forward to hearing it too : )

Russ

[Knorfski]

Good job Russ!

Can't wait when the Counterpoint will be available. I just obtained a Buffalo today but I don't have a I/V Stage and they are not available at the moment.

This one looks very promising! When do you think when it will be available ?

[mikelm]

mmmmm,

I see 2 points labelled "out"

I see no points labelled "in"

perhaps X1.6 & X1.2 are input's but above those u have a cap labelled "out"

X1.1 & X1.7 could be inputs but they are just connected with a resistor to the points labelled "out" . . .

your right Russ I didn't see a circuit like this before !

or is it just too early in the morning here . . .

mmmmm . . . ??????

[BrianDonegan]

X1.1 and X1.7 are the inputs. X1.2 and X1.6 are the outputs. The "+out" and "-out" are connections to the output nodes (X1.2 and X1.6).