Curious about Jensen advice on output tx

From the helpful Jensen 'Insandouts' pdf:
Large capacitance between [bi-fillar] windings needs some
precautions: Do not use an output transformer to reverse
polarity. It puts a huge capacitive load on the line driver
amplifier.

• If you must capacitively couple to primary, use very large
(>470uF) cap, to reduce sub-sonic resonance peak.

I've not looked at a ton of different skiz, but in the ones I have, they almost always reverse polarity with a tx. Some also have coupling caps lower than 470uf. Would someone care to enlighten me on that apparent discrepancy?

Does quadfillar negate some of the issue or something?

I think they  are referring in the first bullet to capacitive coupling between windings.

They recommend OLI. Output load isolator.
Not sure if  the
Recommendation is to not flip polarity by flipping primary connections (most likely)

Or not flipping secodary ...

Earlier in the doc:

using input transformers
Balanced primary can be reversed to obtain
polarity reversal.

Input transformer secondary can be loaded as required
for best performance, since user/designer has control of
the load.

Maybe that is the key distinction. The load on the output side is unknown, and could be less than ideal. So swapping the polarity after the output transformer could change the capacitive loading of the output drive stage.  In some cases swapping the polarity of an output transformer does change the sound of a preamp because the capacitance to ground is different.

Most preamp designs have the phase on the input transformer.

As to the capacitive coupling, read about the 'servo' design of the 990 preamp (John Hardy M1).  This eliminates the blocking capacitor using a servo that nulls the DC offset of the drive amp.
http://www.johnhardyco.com/pdf/M1M2M1p.pdf

The other way to eliminate this blocking capacitor is to inductively couple, like the Neve design (1290, 1073 etc) or many tube output designs (RCA BA2, etc).

When capacitive coupling is used, the interaction between the blocking cap and the transformer is complicated and forms a frequency dependent HPF  (the impedance of the transformer to ground is freq dependent). Lowering the knee (or resonance peak) reduces bias, phase shift, as well as the rollof point. A larger capacitor will push this knee down. 

we had a guy who designed for Jensen at the last brain trust meeting.

He mentioned that if you want flip polarity of a signal it is always best to do it on the input side before any input transformer. So in the case of a mic pre have before the mic input.

pucho812 said:

He mentioned that if you want flip polarity of a signal it is always best to do it on the input side before any input transformer.

Click to expand...

The 'why' is more interesting.

Most of it was covered by dmp above. But I think the designers of the past may have been playing with some known constant problem that was forgotten in the documentation. On paper we have symmetrical balanced I/O where no one defined ground. What did they have. Perhaps dmp had this covered too.

Kingston said:

The 'why' is more interesting.

Most of it was covered by dmp above. But I think the designers of the past may have been playing with some known constant problem that was forgotten in the documentation. On paper we have symmetrical balanced I/O where no one defined ground. What did they have. Perhaps dmp had this covered too.

Click to expand...

It is odd. No real explanation is given. Neve always connected phase inversion switches to its output transformers.

Cheers

Ian

I can see it making a difference if it's unbalanced to ground, if it's floating or balanced I don't see the problem.