Where is best to locate the star-earth?

[bm_mode]

I'm assembling my Leach amp in a new chassis. The tranformers and rectifiers are shielded from the PCBs and output transistors by a partition down the middle of the case. Should I locate the star-earth point on the transformer side or the shielded side of the case?

Cheers,
Joseph.

[hifiZen]

how about right on the partition?

Trannies, rectifiers, and maybe first stage filter caps can go on one side, sensetive circuitry on the other. If you build the star ground using a bolt through the partition, you can attach noisy trannie and rectifier leads to one side, and quieter signal leads to the other side...

[jan.didden]

Yes, sounds like a good idea. In fact, the beauty of a star ground is that the location becomes less important. Look for a spot that minimizes wire lengths as well, especially the high current ones.

Be carefull with the mains ground, if you connect that to the chassis at the mains socket, that can give rise to unwanted ground currents. Best in this case is to run the ground lead to the same starground as well.

I would isolate the star ground from the chassis altogether, ground the mains at the mains socket, then connect the star point to the mains ground point through 10 or 12 Ohms. That still maintains a ground on the system, while absolutely avoiding mains ground loops getting into the system.

Jan Didden

[Ryder]

HAve you read the short piece by Randall Aiken?

http://www.aikenamps.com/StarGround.html


Cheers
Craig

[jan.didden]

Craig,

Thanks for that, excellent piece. [Except that I don't agree with the use of a double or triple ground point].

Jan Didden

[Ryder]

Hi Jan:

You're welcome.
I have a fairly large collection of online tube-diy articles/resources. If you're interested in a copy e-mail me. (I need to be able to send an attachment)

Cheers
Craig

[traderbam]

The number of star points isn't important as such. In any circuit you'll find multiple points where grounds come together. The key is to avoid current loops sharing a single, common conductor when you don't want them to. The time when you don't want them to is when the voltage drop along the conductor matters - beacuse the voltage drop will be related to the sum of the shared current loops. It just happens that one or more star points emerge when current loops need to be separated.

[jan.didden]

Traderbam,

I fully agree, it is not avoidable. Although in one of my supplies I managed to get every component a separate ground lead. But that was a special case, probably overdone.
But I feel uncomfortable by the explicit recommendation to use double or triple ground points.
As an example, I would return the ground of the 2nd cap in a Pi-supply filter to the central ground, not to a separate ground with other (low level signal grounds) and *then* to the central ground. That will surely give rise to ground modulation in the way you mentioned.

Jan Didden

[traderbam]

Jan,
Yes you are probably right about the pi. In general, I suppose one limitation of running a separate ground line for everything is the inductance of the wiring and noise pick-up.

[promitheus]

take a look at this photo: