(2) dirty/clean ground again [sdiy]

[jhaible]

> >this is true - but only for DC. And at DC, the PSRR of most opamps is
> >excellent.
>
> Sure -- as long as you are talking about decoupling resistors at each op
> amp. I was thinking more of the common practice of decoupling the whole
> board with one pair of series resistors.

I actually had this in mind - 10 Ohm and 100uF for the audio opamps of
a whole board (PS-3200 boards).
Of course this means there is no more RC filtering from opamp to opamp
(only from pcb to pcb, and from the "dirty" opamps to the audio opamps
of one pcb), so it's not the optima solution in terms of filtering at all.
But _even then_ the psu impedance which the opamps see should not
be a problem - that's what I wanted to say. As long as you only run opamps
from these rails. (Discreete circuits with less open loop DC gain would
be much more sensitive here, and definitely need a local RC decoupling.)


> BTW (according to Jung's Op-Amp Cookbook) PSSR is specified for symmetric
> changes in the PS voltages,

I wasn't aware that it's specified this way, yet I'm not surprised, as it's
a convenient
method to make one measurement instead of two. Nevertheless I think on
"classic"
opamps (leaving modern rail-to-rail opamps aside) one supply pin is much
more
sensitive than the other anyway. So with the symmetrical measurement, you
would automatically get the "right" one. (?)

JH.