I trust he will not mind that i quote his description of the circuit mods:
three replies to my emails, in order :
-----------------
For the input stages, I'd recommend our JT-11P-1 transformers. The circuit
changes would be minimal. The input coupling capacitors (4 each 22u and
100n) would be eliminated and a single resistor and small capacitor added to
properly load the transformer secondary. I can fax a schematic to show you
how to do this.
For the output stages, I'd recommend our JT-11-EL transformers. One of the
paralleled 100u and 100n capacitor sets would be replaced with a jumper, and
on the other set, the 100u capacitor replaced with a 470u unit. Both 100R
resistors would be replaced with 47R units. The 47K resistors can also be
deleted. Likewise, I can fax a schematic for this.
The input stage modification would have generally the most profound effect.
The reasons for this include much improved noise rejection, especially when
the input is driven by an unbalanced source. Subtle distortions, often
described as a "veiled" or "gritty" sound, as well as RF interference
problems are virtually eliminated by the time-transparent low-pass filtering
provided by the transformer.
The major benefit of the transformer output stage is complete freedom of
interconnection to either balanced or unbalanced inputs. Ground loop
problems, especially prevalent when driving unbalanced inputs, are virtually
eliminated.
I am assuming that the input connector (not shown in the schematic you sent)
is a 3-conductor 1/4" phone jack. If a 2-conductor plug (from an unbalanced
source) is plugged in, the proper connections are made (- input is grounded)
and operation is normal. Of course, the interface is unbalanced and subject
to the ground noise problems shared by all unbalanced interconnections.
If the same 3-conductor jacks are used at the outputs, potentially serious
problems can occur because the 2-conductor plug will short-circuit one of
the op-amp outputs, causing it to current limit. This can not only overheat
the IC, but cause distortion problems in the other (active) output leg.
I hope these are adequate answers to your questions. I'd be happy to
elaborate or answer further questions if you like. And I promise to respond
more quickly, too!
Sincerely,
Bill Whitlock, president
Jensen Transformers, Inc.
------------
Attached please find a schematic of my recommended modification. The DRV134,
like the SSM2142, can drive unbalanced loads without harm. The only
potential problem is instability or distortion if the output cable is not
grounded directly at the output jack (a real transformer allows grounding at
the receive end of the output cable, which largely eliminates ground loop
problems). As shown, you'll need to keep one section of each 5532 operating.
The transformer input stage will have the same gain regardless of whether
the driving source is balanced or unbalanced. Unlike the original output
stage, this one also will not change gain if only one of its outputs is used
to drive and unbalanced line. It imitates a transformer in this regard. Of
course, like a transformer, the other output line must be tied to ground for
this to be true. Do not leave the unused output floating - it will seriously
reduce output on the active output.
----------------
The DRV134 will have the same gain, 2, whether driving balanced or
unbalanced loads (the gain on one pin doubles if the other output pin is
grounded). The old output stage had a gain of 2 for differential (balanced)
loads, but only a gain of 1 for a grounded (unbalanced) load because it
could not sense a grounded pin like the DRV134 does - perhaps the TI guy
didn't know this. Your front-end circuits (both old and new) have a gain of
1. The DBX stage appears to be set for a gain of 1 when control voltage is
zero. So, overall gain should be unity regardless of whether input, output,
or both are balanced or unbalanced.
Input stages are sometimes set for a gain of 0.5 for headroom
considerations. For example, if both input terminals were driven with max
level signals, about 28 volts peak-to-peak each (for bipolar 15-volt power
rails), and of opposite polarity, a unity-gain differential stage would
attempt to output the difference ... some 56 volts peak-to-peak ... and, of
course, go into hard clipping. To prevent this, differential amplifiers are
often operated at a gain of 0.5. Then the differential stage would output 28
volts peak-to-peak. So the input stage would clip at about the same time the
driving source would clip (a good thing for system headroom).
Hope this helps,
Bill
---------------
i am not sure why my images are not coming through, if anyone knows pls contact me and i will correct the original post.
mark