Unity gain input buffer

Hi y'all,

I came across this balanced input buffer schematic. I've been staring at it for a while, but I really am a bit stuck as to how it arrives at a unity gain. Can someone walk me through it?

I don't like it from a quick glance... so don't want to stare at it either.

JR

Maybe a better question.  Is there a “typical” balanced input buffer that can be trimmed a few dB?

dogears said:

Maybe a better question.  Is there a “typical” balanced input buffer that can be trimmed a few dB?

Click to expand...

There was a decent one going around designed by Ted Fletcher (of Alice Mixers) years, decades(?)  ago. I prefer two op amp balanced inputs though.

Ted's design was discussed here and around the Internets.

JR

Search only brings up this thread, and the tech docs for Alice don't show it unfortunately.

Thread about the Ted Fletcher circuit: https://groupdiy.com/index.php?topic=12586

Schematic attached since PRR's link is expired.

> don't want to stare at it

It's not bad, just drawn bad.

Draw it balanced like it should be. If the back-end loss is 1/7.5, then the 10K at the opamp inputs "match" the 75K coming in. On thumbs I get "few-K" cross resistor. Everything loads everything but there is clearly a unity-gain solution.

That was how it was in the schematic I saw, but yes - seeing it that way helps quite a bit!

If unity-gain is the goal, a 4-pack of resistors does the job; with slightly less noise-gain.

Since the four resistors "should be matched" for good CMRR, they will match fine for unity gain.

You can even buy resistor-arrays, made on one substrate and nominal 2% (34dB CMRR), usually run much better inside one part. However the selection of through-hole is getting slim. Fortunately 1% individual resistors are now cheap in quantity.

I think the goal in the original is to be able to trim the circuit a few dB up or down to match two channels.

dogears said:

I think the goal in the original is to be able to trim the circuit a few dB up or down to match two channels.

Click to expand...

Note that there are two distinct levels of performance with a balanced input like this. One level is simple debalancing which cancels ground noise. The second level is a good Common Mode Rejection Ratio (CMRR) which cancels common mode noise like interference picked up over a long cable. Getting good CMRR is much harder. AFAIK it's basically not possible with a simple single op amp circuit like you're using. And you definitely would not be able to trim it because doing so would throw off the impedance balance. You would need a two op amp circuit of which there are many. But if you're using short cable runs and don't need good CMRR, then the circuits posted would would fine and you can trim by just trimming part of the feedback resistor.

The original here comes from a Sontec design.

But looking at how PRR drew it, if the 75k are matched and all 10k are matched, wouldn’t the crossing resistor not affect the CMRR?

The crossing resistor has no effect on CMRR.

Source impedance has significant effect on CMRR due to different impedances for differential and common-mode signals.

Gain trim, yar.... put a standard "volume control" after a fixed-gain diff-amp, perhaps with stopper to limit the available range. For few-dB trim it is not worth putting gain-trim inside the diff-amp.