boji said:
47k and 28k values?
Frankly they are arbitrary and from other people's design suggested values. :-[
The only reason for using such high value for the injection resistors is that it reduces the load imposed to the pan-pot and aux pot(s). I don't see any justification why they couldn't be scaled down to 10k. That would drop the bus impedance down to about 200r, which would benefit from the use of a VLN opamp/DOA (about 1nV/sqrtHz).
Now, is there a justification for making the summing amp less-than-unity-gain?
In terms of gain margin, since the opamp is in non-inverting mode, effective BW does not change below unity.
A commonly given reason is that it gives an additional xdB of headroom for the summing amp. This is often based on the ASSumption that a number of channels mixed equally would sum algebraically, so 2channels->6dB, 4channels->12dB, and 10 channels->20dB, which would clip the summing amp. A more conservative approach is that channels would combine quadratically, so 2 channels-3dB, 4channels->6dB, and 16 channels->12dB; for a large mixer, say 48 channels, the cumulative level would be about 16-17dB above nominal, which does not leave much headroom.
In fact, the actual math is in-between, since transients actually may combine algebraically.
But mixing is not math; a typical mix would involve about 4-6 dominant channels, that would constitute most of the energy, the rest being almost negligible in terms of signal strength.
I've never seen a case where,
with all individual channels properly set, one had to move the faders down because of summing amp overload.
In the current state of SS electronics, hitting the summing amp hard (within reason) is beneficial to S/N ratio, without any adverse effect.
That may be significantly different for vacuum electronics.