[LAU] balanced <--> unbalanced

[Chris Caudle]

On Tue, January 10, 2017 12:15 pm, Ralf Mardorf wrote:
> We are never talking about
>   unbalanced out --> balanced in
> we are only talking about
>   balanced out --> unbalanced in

Correct, NOT unbalanced out to balanced in, only balanced out to
unbalanced in.

(whether connection between - and GND is made close to ouput, or close to
next device input)

> IMO it doesn't make a difference.

For noise reduction it can make a difference when cables are long, the
connection difference changes whether the voltage difference between the
two devices due to power line leakage currents is higher at the unbalanced
input, or higher at the balanced output.  For some types of balanced
outputs which can float effectively on the common mode voltage the power
line noise is lower with that connection.

The other way that connection point makes a difference is that connecting
at  the input of the next device changes the capacitance which loads the
grounded output.  That could affect stability in some cases.

By the way, what was the output level from the device with bad sound when
connected to the unbalanced input?  Section 1.2 of the paper I referenced
earlier points out that when grounding one side of that style of output
circuitry, the ungrounded leg will be driven +6dB to keep the nominal
output level the same, which is twice the output voltage.  If that drives
the output into clipping, large currents can flow through the grounded
side.  If the output was near the listed device maximum it could have been
clipping just slightly on peaks.  I would not expect that, usually an
unbalanced input is designed for lower limits, so I would expect the
unbalanced input to clip before the balanced output even in single ended
connection, but hard to know without checking.

-- 
Chris Caudle