Grounding and Minimizing the Current Loops

[AndrewT]

At the summing point where the source currents combine, there is a voltage gradient across the copper area causes by copper resistance. You choose one spot as the feedback node and feedback ensures that all parasitics for that spot only are corrected. If you take speaker out from a different point along the gradient, you get distortion. Even inside the gradient tap point there may be a significant voltage gradient, so taking output from one side of the tap point may still result in distortion. Therefore you move the speaker output away from the summing point and closer to the feedback input along the feedback trace; the charge gradient quickly dissipates.

This is much more evident in reservoir grounding where the amps flowing are several times more than the speaker output. There may be a gradient across the star ground bolt if you put star ground right at the reservoir summing point, so even the direction the wire tab faces may cause ground noise. Better to put some thick wire between the gradient and the star ground.

nicely explained.
Maybe now others will remember.

 

[keantoken]

I am now thinking about connecting the rails and ground from the VAS-IPS board to the earlier stage of power supply reservoir caps instead of taking short connection points to the OPS board.

I understand that this will make the current loop much longer. However, since I use a heavy duty passive filter DC-RC-RC-RC-RCCC for the VAS and IPS rails, I am afraid that the switching noise from Diode D gets too close to the output devices. I think I had better route it away from the OPS board so that the switching noise doesn't influence the output device and feedback as much.

What do you think?

Is the D+C grounded on the frontend board? That will act like a rectifier at your frontend. You don't want those currents flowing through frontend ground. I think it should be placed and grounded next to the reservoir/rectifier section, then the intermediate RCs can be grounded at the frontend board or reservoir ground. This is a tradeoff between ground noise and rail noise. I guess that's how RC filtering works.

 

[HiFiNutNut]

What I have done on the IPS+VAS board is one ground zone for the DC-RC-RC-RC, and another ground zone for the last RCCC in the rail filter. Another ground zone is the the VAS TPC ground point, and finally, the ground zone for the input ground and feedback ground. I can use 4 separate wires to connect these ground zones to wherever I want them to go. I think they are best connected to the reservoir capacitor junction before the OPS board (instead of the PSU Common marked in the OPS board).

That is for the ground.

As for the rails, I am thinking about connecting the IPS+VAS rails to the reservoir capacitors before the OPS board as well (instead of the rail tracks in the OPS board) so that the recifier noise is placed further away from the output devices.\

 

[AndrewT]

........... I can use 4 separate wires to connect these ground zones to wherever I want them to go. I think they are best connected to the reservoir capacitor junction before the OPS board (instead of the PSU Common marked in the OPS board).
..............

No.
Create a Main Audio Ground (MAG). This should be near all the audio PCBs and assemblies.
Get all those Flow and Return Pairs correct.
Then connect the PSU Zero Volts to the MAG with ONE wire/cable.

 

[HiFiNutNut]

No.
Create a Main Audio Ground (MAG). This should be near all the audio PCBs and assemblies.
Get all those Flow and Return Pairs correct.
Then connect the PSU Zero Volts to the MAG with ONE wire/cable.

So is the MAG the same as the STAR? Is the PSU Zero the PSU common? If so, this was the way I did it in my power amps.

I agree with you that instead of connecting the 4 wires to the reservoir capacitor junction it is better to first connect the juction with a piece of wire then the 4 wires go to the end of that piece of wire. This should make the "ground" cleaner.

 

[AndrewT]

This latest description sounds correct.
Your post23 description says (to me) something very wrong.