Differential signals, is there an alternative to a diff pair

[Andrew]

I'm sure there's something to this but I can't see the obvious.

Let's assume we're talking about a fully balanced signal.

Now lets try to amplify it and keep it differential.

We could use a differential pair, of treat each signal as separate single ended signal and amplify it with a normal common cathode or cathode follower, the latter might be useful if we just want drive, not gain.

What's the difference as far as the treatment of the signal is concerned?

Or put another way, what extra does a differential pair do over and above what two common cathodes on an already balanced signal? I know the diff pair can be a phase splitter etc.

Andrew

[Nick]

You avoid a cathode bipass cap. I think there will be some even distortion cancellation, but I can't convince myself its any greater than two separate amps.

[Andrew]

Thanks Nick, that was the same internal debate I was having.

Mostly it related to two design choices and not wanting to limit myself later on.

The first was using bifilar output TX on the phono and if they were useless if I ever decided to do a balanced phono, or even a balanced ouput stage on the phono.

The second and more likely choice, since a balanced LCR ain't gonna be cheap, was how I might use the S&B in a balanced pre with drive configuration. Basically, doubling up on the cathode followers and having the S&B, optionally, do the balanced to single ended on the way out.

All I, or Colin for that matter, since it we were talking about trying this after hearing Little Martha, might need for a balanced pre using TVCs would be a pair of good quality phase splitting input TXs from someone like Sowter.

cheers,

Andrew

[Mike H]

Thinking about it, there's probably 3 or 4 main reasons for using long-tailed pairs

A. it's "fashionable" (i.e. looks impressive)

B. in op-amps for supply ripple rejection, but that only works if both the opposing phase signals are taken and used (common mode signals [ripple] null out)

C. for front-ends of tranny power amps so output DC doesn't drift [much] (both BE junctions have same V-drop due to being at the same temperature. Also, ends up a lot like an op-amp)

D. you need to compare two input Voltages (a comparator)

E. there is no E ( © A. I.)



 

[izzy wizzy]

I think LTP or differential pairs have good CMRR to PSU noise and input noise. With the appropriate configuration, I would see this as an advantage. They are also easy to cascade with direct coupling.

Two separate amplifiers will not have the CMRR of a diff arrangement. To me it would seem to be the worst of both worlds - extra complexity and no great increase in CMRR.

Audio Research used to do the later in their balanced stuff but that's no reason to repeat the idea

cheers,

Stephen

[Andrew]

OK, chaps, but what if you don't want gain, just drive?

Andrew

[Nick]

Two separate amplifiers will not have the CMRR of a diff arrangement.

Why not? If its common, then its still common even if its two amps.

[izzy wizzy]

Why not? If its common, then its still common even if its two amps.

Well to my way of thinking, two separate amps will act differently as there's nothing to force symmetry as in an LTP. The might be balanced but they are not differential because the two halves are not linked.

Now if I woz cleverer, I would hope I could bring some science to this or I should I say, I would hope there is some science to back this up but I guess, I could be wrong. It's what the text books say that I got.

cheers,

Stephen

[Nick]

Why not? If its common, then its still common even if its two amps.

Well to my way of thinking, two separate amps will act differently as there's nothing to force symmetry as in an LTP. The might be balanced but they are not differential because the two halves are not linked.

Now if I woz cleverer, I would hope I could bring some science to this or I should I say, I would hope there is some science to back this up but I guess, I could be wrong. It's what the text books say that I got.

cheers,

Stephen

Yes, I agree with that, but I am not sure (cant find any science either) if the same is true to PSRR. A signal on the anode of both the devices (or cathode) will affect both the same way so the net result will be zero in terms of the difference output signal.

[Mike H]

If you want to drive yourself crazy thinking about long-tailed pairs try one of these ~ like I have been today

I'm a teapot, I'm a teapot .....

No I am the reincarnation of Napoleon I and you are all my slaves