[GrimJosef]

I think that the only difference wiring the capacitors as you've shown in the sketch will make will be to the cross-coupling between the channels because of signal currents flowing in what are now the common grounds. To be honest I wouldn't expect that effect to be very large.

The question of whether it's a good idea to put large HT smoothing capacitors at the amp end of the umbilical might depend on the rest of the power supply circuit. Specifically, when you call these 'the main capacitors' are there actually other capacitors back in the power supplies themselves and is there any impedance (a resistor or a choke) in the HT rail between those capacitors and the 100uF ones on the amp chassis ? The reason I ask is that if there is no impedance, and particularly if these really are the main capacitors because there are none back in the power supply, then you run the risk of large pulsed charging currents flowing along the umbilical cores. This can be a recipe for generating a lot of noise at the AC line frequency plus a very broad comb of harmonics. Even if most of the capacitance is back in the psu, significant pulsed currents can still flow into your 100uFs. Of course the choice and layout of the grounding scheme will be critical if you want low-noise operation. Normally in a 2-chassis scheme I think the standard approach is to do as much filtering as possible back in the psu chassis and to transport as near as possible constant DC voltages down the umbilical. The resistance of the umbilical cores should be of the order of an ohm or less. At bass frequencies, where you're likely to need the most power, the effective impedance of realistic capacitors is going to be a lot larger than that, whichever end of the umbilical you put them (100uF at 100 Hz has a reactance of 16ohms). So you might as well put them at the psu end where they won't generate a lot of charging noise.

VB