improved opamp I/V

[ABO]

In an attempt to improve my opamp I/V, I tried an modification to the standard circuit. The idea was to add error-feedforward to the circuit.

Any voltage on point A in the attached schematic is the difference between the ideal and the actual output voltage(assuming perfect opamps otherwise). If this voltage can be added to the output signal, this should improve performance.
I therefore added another opamp to the circuit in the way shown in the picture.

Although I've never seen this solution before I'm sure it is not a new idea, as it is so simple. The results however are hardly subtle. The sound was improved in every way thinkable. It was easily heard on first switch on.

I think for those who are reluctant to build discrete circuitry this can be a very worthwhile and easy improvement of their cd player.

Best,

Abo

[darkfenriz]

At first glane- looks like perfect simplicity. Wonderful ! A more interesting comparison would be using something in TL072-range of performance.
Do you use balanced output, un-grounded DAC of other type of subtraction out+ and out-?

best regards
Adam

[ABO]

My DAC is a 1704. It is used with oversampling (DF1704 SRO).

I use no other output circuitry. The outputs shown in the drawing are going directly to my pre.

I used to have a 42 kHz bessel filter after the I/V. However the sound was better without it. I don't think output filtering is really necessary in my case.

I now use the opamps from the former output filter for the feedforward circuit.

Best,

Abo

[fastcat95]

ABO:

If I read your idea correctly, the output of the second OPA627
is simply combined with the output of the I/V op-amp as it goes
to the output jack + pin? And you have some sort of first order
filter on the I/V op-amp with the capacitor paralleled with the
I/V resistor? Do you think that the feed forward op-amp must
be of the same type as the primary I/V op-amp? And have you
tried this configuration with any other op-amp than the OPA 627?

Thanks!

fastcat

[BrianL]

Abo,

Before you decide that this is going to do a lot of good for you, I'd suggest doing the math to figure out what sort of signal you actually see on point A. The transimpedance amp will try to force that point to be zero (save for DC input offset). Most likely the signal that the lower op-amp will see is lost below its noise floor.

[Dave]

the signal at the -OUT will not be an inverted signal of the +OUT.
If the opamps were ideal the voltage at point A will be zero and hence the output at -OUT would also be zero.

[Bare]

Silly Me.. Had I seen this earlier I would not have removed the opamps from My CD player.
But getting output directly out of the DAC chip really works well.

Best opamp ever made is NO opamp.. at least to my ears :-)

[ABO]

Originally posted by BrianL

Before you decide that this is going to do a lot of good for you, I'd suggest doing the math to figure out what sort of signal you actually see on point A. The transimpedance amp will try to force that point to be zero (save for DC input offset). Most likely the signal that the lower op-amp will see is lost below its noise floor.

Well,

If there is no voltage on point A, then the output voltage is equal to Iout x R and there can be no improvement of the circuit, as it is already perfect...

The signal of the feedforward opamp is indeed almost zero. With a multimeter I cannot measure any signal on this point. However, if you look into the theory of error-feedforward, you will learn that this is almost textbook application.

Of course there is noise contribution of the second opamp and I'm sure that such a circuit will not improve subjective performance in every case. However in my case, using OPA627s (which were already on the PCB) it is definately an improvement.

I don't think there will be any debate among those who hear my DAC with and without it as the sound is better in every aspect, and most notable bass extension.

Bare, I don't know if you are taking a stab at me or not.

I know my I/V is better than the standard circuit. If it can be bested by other topologies, I don't know. The difference will certainly much smaller than ever before.