Re: Ground Busses

Eeyore wrote:
> 
> robert casey wrote:
> 
> > Avoiding ground loop issues you need to understand the paths the
> > currents thru such a ground are taking.  And remember that all real
> > conductors have resistance, and that "noise" voltages will develop
> > across those current paths.  Using dedicated wires for each current
path
> > avoids this.  That's what you do in a star ground system.  One
im****tant
> > current path is the one in the power supply.  From the center-tap of
the
> > high voltage secondary to the filter cap is a high current spike every
> > 120th of a second.
> 
> 100th of a second in most of the world actually.
> 
> > SO to keep that out of the rest of the amp, you
> > connect the secondary center-tap directly to the negative side of the
> > filter caps, and then what becomes the B- (at the filter cap negative)
> > then becomes a wire leading to the star ground.  I've connected my
> > output stage cathode resistors to this B- point, and then use the star
> > ground for low current low signal level work.
> 
> Spot on and well explained.

He's right, and the reason is that the current spikes from diode
switching and charge pulses
are not flowing in the OV wire between the -ve bottom of the reservoir
cap and the star point or buss 
near the input.

If one uses a CRC or CLC type of supply, the same applies, with two caps
-ve brought together to the 
0V rail to which the cathode circuit MUST be connected to avoid signal
currents causing any 
voltage developed in the 0V wiring.

Keep 0V wiring to the star point or other "referencing 0V wires" or NFB
wiring 
away from the underside of any magnetic component such as PT, OPT, or
chokes because induced 
voltages are quite likely to spoil the noise performance.

The CRO should have difficulty displaying the amp noise at the output.
If there is lots of medium F junk noise or RF, 
tem****arily bypass the output with 0.47F, and this should reduce the HF
blurry 
junk on the CRO, and leave the 50Hz and related harmonics and spikes
better able to be seen.

Ground the input grid to 0V during noise testing.

Turn up gain controls if any to -6dB, their noisiest position.

After any wiring changes and adjustments of wire positions, watch the
CRO and the noise changes.

Make sure the CRO is itself grounded to the amps 0V rail and that 
when you probe the 0V rail no noise appears.
Its no use blaming an amp when your test methods are crap.

Consider yourself WRONG until you proove otherwise to yourself.

Patrick Turner.

> 
> And Andre Jute wonders why I criticised his 'design by rote' post !
> 
> Graham