Originally Posted by DonH50
Hmmm... The resistor to ground should match (maybe "equal" is a better word?) the impedance of the driven output (which will normally have a series resistor as was stated), but since it is not driven then there is essentially no rejection to ground on the receiver, obviating the high common-mode rejection a truly differential signal would provide.
Not exactly. The interference rejection is entirely provided by the differential receiver. The balanced output improves the SNR by delivering a 2x signal to the differential receiver. The differential receiver provides cancellation even if the resistor is shorted out.
I call that circuit "pseudo-differential". It all depends upon how the ground is routed, natch, since to provide any sort of rejection the buffer opamp's input signal must reference the same ground (return path) as the other side (grounded through the resistor).
I don't know about the pseudo part. The differential input has to be real to provide any benefit.
The usual, or at least one, LF (audio) approach is to simply use another opamp to invert one side, creating a quasi-differential design. Fully-differential circuits are harder to find (not a normal opamp).
A true differential input can be made from general purpose op amp(s) - from 1 to 3 of them depending on what sort of performance you desire. A purpose built differential input op amp chip includes all or most of the active and passive components required, usually including some resistors that need to be carefully matched.
Aside: If nothing else changes, a balanced output using a second amp to replicate and invert the signal should provide 6 dB voltage gain (twice the signal), the higher signal level Arny is referring to (I assume?) I could not say how much that matters for a given system (depends on the noise floor and headroom).
Agreed. The differential input costs you 3 dB of noise performance (2 uncorrelated noise sources instead of just one with an ordinary unbalanced input) but the balanced source gives you 6 dB more signal or if you will puts the noise floor 6 dB further below the signal, for a net gain of 3 dB less noise.
FWIWFM, I have used a number of Mackies and other similar systems and IME they do not do well at isolating ground or CM noise the way their XLR or truly balanced connections do.
Remember the different ways of building a balanced input with 1, 2, or 3 op amps or using a dedicated chip? The more op amps the less dependence on matching resistor values. The dedicated chip usually has factory-trimmed resistors and keeps them tightly thermally coupled. Improving the matching of these resistors improves common mode rejection.
But, experience (including mine) varies widely, or maybe wildly...
What you saw is what one gets with the lower cost implementations of active balanced inputs. The really nifty chips usually end up in mic preamps, but one sometimes sees them in premium equipment. For example among audio interfaces the M-Audio Delta 44 and 66 use impedance balanced outputs, the Delta 24/192 uses active balanced outputs, and the LynxTWO uses dedicated chips for both inputs and outputs. The output chip senses when one output line is grounded and doubles the voltage on the other output, which simulates what a transformer output does. The audio interfaces cost $100, $200, and $800 respectively and have dynamic range to match.