Input Transformer Operation

[ziller]

I'm having some difficulty understanding how an input transformer operates, in terms of CMRR. I've read over the excellent info in the Meta threads, along with lots of other info (maybe have info overload). In particular, I've been reading the Jensen artilcle

http://www.jensen-transformers.com/an/an002.pdf

and am hoping someone can explain some of the concepts in the paper. For instance, it states that symmetry has nothing to do with noise cancellation... But don't the signals of each line have to be 180 deg out of phase, lest the signal itself be canceled in the transformer? Or is it saying that it doesn't matter in terms of CMRR, but it does matter in terms of signal transfer to the secondary? Did I just answer my own question?

Also, on some schematics, the center tap in the primary on the input transformer is tied to one end of the secondary winding. In the Jenson schematics, there is no primary center tap, and the signals just float. Seems like that would make a difference, but I'm not sure what it is...

If it's the case that the signals have to be 180 out of phase, and the primary is not center tapped, then I think I get how any CM signal would be canceled in the transformer: both ends of the primary would be at the same voltage and so no current flows in the primary, and so no mutual induc to the secondary.

I think I'm missing something basic here, but can't pin it down...


Thanks.

 

[Larrchild]

The noise cancellation and CMRR occurs in the twisted and balanced mic cable. The floating input winding merely allows it no connection to ground. (Which would un-balance it).

 

[ziller]

Thanks Larrchild.. I am still a bit confused, however. From what I get from the readings, it is importent that each signal line have the same impedence (balanced) so that any voltages induced on the lines from interference are equal on both signal lines. This allows the receiving device (active or transformer) to reject the induced signal, as the receiver functions as a differential amp with an equal voltage on each input pin (outputing 0 volts).

Close ?

 

[PRR]

> it states that symmetry has nothing to do with noise cancellation...

No, it says "Signal symmetry has NOTHING to do with noise rejection."

Transformer symmetry IS critical.

> But don't the signals of each line have to be 180 deg out of phase, lest the signal itself be canceled in the transformer?

They can't be in-phase or they will, of course, cancel. However they do NOT have to be equal to get high CMRR. In fact we often work an unbalanced signal (one side zero signal volts) into a differential input (such as a transformer) for high CMRR.

> the center tap in the primary on the input transformer is tied to one end of the secondary winding.

Shouldn't be. The input center-tap may be bonded to Shield to improve rejection of very large crap, such as adjacent power lines and especially RF. One end of the secondary often returns to Internal Signal Ground. Shield and Ground are often connected together somewhere in the box. But dumping the input CT directly into the signal ground defeats the point.

 

[ziller]

Thank PRR... I think what I was missing is that the input tranformer is functioning as a differential input device.

On the schematic I was refering to, the CT of the primary is connected to one end of the secondary, and they are both grounded to the receiving chassis.

 

[CJ]

As soon as you connect one end of a line to ground, your signal is no longer balanced.
Thats all you need.

 

[Larrchild]

Yeah man, ground one end of your phone line for vivid proof.

 

[NewYorkDave]

On the schematic I was refering to, the CT of the primary is connected to one end of the secondary, and they are both grounded to the receiving chassis.

I scanned an article from 1950 Audio Engineering magazines that will shed some light on the matter of why you sometimes see grounded centertaps on input transformers. It's called Longitudinal Noise in Audio Circuits, by two Bell Labs engineers, and it's a large (3Meg) PDF file. Anyone want to host it?

 

[Kev]

3 meg is big

if there are no takers I'll consider it - definately want to see it
if it is texted based it could be re-written and made very small ?
(txt file and add a couple of jpgs)

Longitudinal Noise in Audio Circuits is a new term for me
does that inmply a lateral noise ?

 

[ziller]

I may be able to host it, for a while anyway, if I can get it... I don't know if I can receive emails attachments that large though. Maybe zipped first?

 

[ziller]

NewYorkDave - check your PM...

 

[PRR]

> It's called Longitudinal Noise in Audio Circuits, by two Bell Labs engineers, and it's a large (3Meg) PDF file.

Longitudinal Noise in Audio Circuits

This is pretty dense stuff, not for casual reading.