output buffer

[peter purpose]

Gentlemen,
Would this work OK as an output buffer, using the transformer to flip the polarity?

Cheers
peter

 

[NewYorkDave]

When driving a reactive load, it's generally a good idea to include a small (100R or less) resistor in series with the output, but outside the feedback loop.

Otherwise, the circuit looks OK as long as the DC offset at the opamp output is always going to be very low. Otherwise, it's coupling cap time.

(PS: Does the diagram really need to be that huge? :wink: ).

 

[gyraf]

What Dave says, but I'd put in a 100K from non-inverting-to-ground to get a tad better DC balance between inputs..

Jakob E.

 

[NewYorkDave]

Good point, JE. I think 50K would be even better, since (assuming the driving source impedance is low) that's about the same as the resistance seen by the inverting input (R1 || R2).

 

[Larrchild]

the DC offset present at the opamp output is always going to be very low

With the output dc grounded, it's gonna be Very low!, lol. I'm just worried the 2520 is gonna try to leave the circuit board on it's own steam under less than ideal conditions (like asymetrical power rails during powerup).

A cap there might not be so bad.=)

 

[Greg]

Yup, I've always used the || combination of R1 and R2 from non-inverting to ground... which is exactly what Dave said. Helps with the DC offset. :thumb:

 

[peter purpose]

Thank you gents.. :thumb:

 

[bcarso]

And you can bypass R3 with a C for a bit of noise reduction. Also, do you really need the 100k input impedance? That choice of R's is giving you 8 uV of thermal noise in a 20kHz BW before any contribution of the op amp.

The 100k does make it pretty bulletproof though.

I'm assuming that you are connecting the input to a d.c.-coupled output of something, and that it has low output offset voltage. Then the 50k is appropriate for bias current balance. If a.c. coupled then 100k is correct.

But if the source has a d.c. offset, beware of putting d.c. through your iron.

 

[Samuel Groner]

Of course for lowest distortion R4 could/should be bypassed by an air-coil inductor. See JT-OLI-3.

As Brad said, 10k or similar for the other resistors is probably a wise thing to consider.

If you run the output into an unbalanced input, expect higher distortion as the opamp now sees a capacitively coupled short.

Samuel

 

[peter purpose]

To what it would be nailed.....

 

[gyraf]

from Channel out? Then use 33K resistors like they do already.

 

[peter purpose]

Thanks Jakob.. :thumb:

 

[NewYorkDave]

In that case, you don't even need R1 on the buffer. Take the output of the 30K resistor from the channel strip directly into the inverting input of the op-amp. Make R2=30k, R3=15K and you're golden.

 

[peter purpose]

Thanks Dave... :thumb:

 

[peter purpose]

Thanks also to Brad, Greg, Larr and Samuel... :thumb:

 

[clintrubber]

[quote author="peter purpose"]Gentlemen,
Would this work OK as an output buffer, using the transformer to flip the polarity?[/quote]

Come to think of it, what's actually the idea behind the double polarity flipping i.s.o. using it non-inverting with the TX also non-inverting ?

Just curious.

That opamp isn't a monolytic with JFET-input, so that won't be the reason.

As far as noise matters here, the drawn topology (shunt feedback) has as a disadvantage that its noise gain is higher (just +1 though) than its signal gain w.r.t. the non-inverting (series feedback) topology. So that won't be the reason either, on the contrary.

But hey, maybe it just sounds better this way :thumb:

 

[bcarso]

Although the garden-variety monolithic JFET amps are well known for common-mode distortion, other amps suffer as well in varying degrees, and hence can benefit from inverting mode operation.

 

[clintrubber]

[quote author="bcarso"]Although the garden-variety monolithic JFET amps are well known for common-mode distortion, other amps suffer as well in varying degrees, and hence can benefit from inverting mode operation.[/quote]
I see, hadn't heard of that before. But the 2520 as shown has a discrete BJT input ... - still ? :roll:

Regards,

Peter

 

[bcarso]

Oh sure---you can still get common-mode distortion from a discrete bipolar input.

 

[clintrubber]

[quote author="bcarso"]Oh sure---you can still get common-mode distortion from a discrete bipolar input.[/quote]
Yes, you're right - why shouldn't it. Mmmm, right, while noise isn't any longer an issue here at the signal levels this stage will live at then indeed best doing it as drawn.
Hi Peter P, this also originally your intention ? What were your thoughts ?
Just curious.

Thanks,

Peter