Any pointers to a calculation of what the CMMR would be? It really doesn't work very well if the source and drain are unbalanced and have a big impedance difference.
Hi Julf,
Remember my world is RF, not audio, so I was thinking more of the LF RFI rejection though it should work at audio as well (mainly for power supply noise). As for calculations, not off-hand; I use a little red book by Walker (can't recall the title, have to look it up later) or the old ITT
Reference Data for Radio Engineers book. Don't know if either has calcs or curves. One of those things I've "known" for decades from a college EM class plus a lot of work in the field then and since; at one point a couple of us made a bunch of twisted pairs using different tpi and measured their CMRR but I have no idea where that data is now. As others will be quick to point out, I could be completely wrong and making the whole thing up. But, it was also taught in a Besser Associates EMI class a few years ago.
It'd be interesting to repeat the measurements using our VNA but not sure when I'll have the time. My workweek is pretty crazy right now. I am wondering about your assertion on the impedances. Not disputing it, do not know, and am curious. RF tends to use matched circuits, and most of my current world uses differential pairs, but audio tends to have low driving impedance and very high load impedance for interconnects. I am suspecting my experience is (again) not valid for audio though I know folk do it.
IIRC the idea is that the twisting lets the ground act like somewhat a shield even for single-ended singles and that the twists allow signals to impinge "equally" on signal and ground (return) conductors to provide the rejection. Not nearly as effective compared to differential, I agree with you on that! And pretty much a fail at HF where the wavelength (or fractional 1/2 or 1/4 wavelength) approaches the twist distance.
I have no idea the actual CMRR in practice, however. I do it all the time when adding test wires and such, more or less by habit, and very recently a younger engineer hooked up a probe connector without twisting, got lots of noise, then twisted and saw a significant reduction. Didn't measure it, alas; I just saw the noise, looked at the wires, asked him to twist them, and the noise went away (or was at least reduced below the noise floor of the DSO). So I have what we on ASR would call subjective evidence. But not like it's all that hard to get below the noise floor of a DSO...
There are some online references so a search might help. The young engineer I was working with found several but I didn't follow up. I have a vague memory from one of the classes that the twisted pair idea was actually patented by Alexander Graham Bell back in the late 1800's! Would never have guessed that.
Anyway, you can treat all my experience as anecdotal, but I'll keep doing it out of habit if nothing else.
I would like to hear your thoughts on the impedance impact, however. It makes sense but I have not really thought about it.
HTH - Don