[Henry Pasternack[_2_]]

"Eeyore" wrote in message
...
Ian Thompson-Bell wrote:
The problem I see with that is that all output currents must flow from
the power supply, through the load and back to the PSU 0V but they share
a common ground with the inputs. How do you separate these?

Clever local decoupling can achieve this for all but very low frequencies.
A
technique used at Neve in fact.

Once again, agreed. The way I do it is as follows:

The main capacitor board has a short ground bus running through it. The
power transformer center tap connects at one end of this bus. All the
charging currents associated with the rectifier are contained at this end.
The other end of the ground bus faces the signal circuitry.

The B+ lead to the output transformer comes off the first filter cap.
Current
flows to the transformer, through the output tube(s), plate to cathode, and
back to the filter cap ground bus. The wires in this circuit are routed
close
together (or twisted) to minimize loop area.

The second filter cap (or set of filter caps, or whatever) sits on an
extension
of the ground bus/ No output stage current flows in this extension. The
far
end of the extension is the input/driver circuit ground reference. On the
B+
side, RC or preferably LC decoupling is used to isolate the two loops at
signal frequencies.

The input jacks are electrically isolated from the chassis and connected to
the input circuit using a twisted pair. A separate ground bus runs the
length
of the input/driver stage. The input twisted pair ground is connected to
this
bus at the input tube end, and a wire connects the driver stage end (or the
signal circuit bus itself is extended) to the cap board signal ground.

There is a solid connection from the rectifier side of the power supply
ground
bus to the chassis. Then, the input jack grounds are optionally connected
to
the chassis by small (0.01uF) capacitors with short leads. These (in
principle)
bypass RF currents on the input cable shields to the chassis.

This is a pretty good approach that has worked well for me. A major source
of hum is unbalanced capacitance between the power transformer windings
and ground. With this arrangement, these currents flow through the chassis
and back into the supply through the rectifier side of the power supply
ground
bus, avoiding the signal circuitry.

Sometimes, though, in spite of careful design, the damned thing still hums.
In
that case, you have to experiment until you find the cause of the problem.

-Henry