Mixed Jacks

[Martin Czech]

>    But I don't like the idea of multiple inputs to one output at all,
>    because the outputs need some resistance for protection and stability,
>    and the inputs should have no very high resistance, because the input
>    summing amp needs extremely high impedance in this case. This would
>    mean quite a load for the output and some voltage drop, which is
>    undesirable e.g. in the case of frequency control voltage. I'd prefer
>    low offset voltage buffers, with one input und , say, four outputs,
>    each with it's own protection circuit.
> 
> It's the responsibility of the modular designer to handle this
> situation in a reasonable way.  Ie., place the current limiting
> resistor within the feedback loop.

Yes.

But: This would mean to forward external voltages directly to the
inverting input of an opamp. I don't like this. This won't help much
with closed loop stability problems either.  Of course, some protection
network could be applied only for the - input, the series resistance is
not so important in this case.  So two protection circuits, one for the
output and one for the input path are necessary. That's the way we do
our CMOS-ICs all day, this is called "force-sense" I think.  If it is
carefully designed, the additional capacitance of such protection
should not affect the buffer stability. But what happens, if a long
patchcord is connected? The large capacitance will slew the output and
thus phase shift the feedback. What phase margin will be left?
That's the reason for using some output resistance to protect the feedback
loop against phase shift, of course sacrificing some accuracy.

The LF356 seems to be very stable, even with large capacitive output loads.
Are there any more modern op-amps with this feature?

m.c.