Running dedicated AC mains .... need advice ....

My instinct would be to use ONE copper ground rod, close to the system, with the grounds of the three Romex runs coming together at one and only one physical point, at that ground rod.

As you obviously realize, the upside of dedicated runs that are separate for digital and analog components is reduced cross-coupling of digital noise into the analog components. The downside is possible voltage offsets between the chassis (and consequently the signal grounds) of the different components in the system, particularly at high frequencies (such as typical noise frequencies), due to the inductance in the wiring. If your components are unbalanced, meaning that they use single-ended interconnects, these voltage offsets result in extraneous current flow in the shields of those interconnects, that is in common with the current flow of signal return paths.

So to minimize that downside, you want to have the chassis and consequently the signal grounds of all of the components as electrically common as possible, with as little inductance between them as possible. DC resistance will probably be sufficiently low regardless of what you do; the concern is with inductance, that can dramatically increase ground run impedance at noise frequencies.

I think that using one ground rod as I suggested will accomplish that. As long as the grounds are tied together at only one point, so that no ground loops exist, there is no reason to "keep the grounds separated." And having three separate rods in relatively close proximity doesn't really keep them separate anyway.

Regards,
-- Al

I agree with all of the excellent cautionary statements that have been made. I want to clarify that the suggestions in my previous post (the third post in this thread) did not mean to suggest anything contrary to these cautions.

I assume that there is and will continue to be a bond between ac neutral, safety ground, and earth ground at the main entrance of power into the house, near the breaker panel. For all existing runs, and for the new ones that may be added. The earth ground is essential for lightning protection, and the bonding of ac neutral and safety ground at that physical location is essential for effective circuit breaker operation, as the paper Jea48 linked to makes clear. And, yes, it seems likely that not having outlet receptacles may be a code violation.

But to restate the point I was trying to make: The downside of having dedicated runs to different parts of the system is increased voltage differential between the chassis of the different components, particularly at high frequencies, which can result in noise currents flowing through the same cable shields as signal return currents, thereby effectively summing the noise into the signal (the very thing that having multiple ac runs is intended to improve).

Connecting the grounds of the different runs together, near the system, will minimize those voltage differences. Jea may very well be correct that tying that local system ground to an earth rod will not accomplish anything, but I don't envision that it would have any negative effects either (in terms of either safety or performance).

A better approach all around, though, may be to utilize isolation transformers, as described by Zargon in this thread:

http://forum.audiogon.com/cgi-bin/fr.pl?htech&1228780109

Regards,
-- Al

The earth shall not be considered as an effective ground-fault current path.
From 2005 NEC 250.4 (A) 5

True, but the presence of such a path will do no harm as long as the proper path is also present, at the service entrance (that proper path presenting a much lower impedance to fault current than the extraneous path, as diagrammed in Whitlock's paper).

The earth does not have some magical mystical power that sucks AC noise from our audio systems.

Agreed. As I admitted, you are likely correct that a ground rod near the system would accomplish nothing.

More and more audio equipment manufacturers are building their equipment with double insulated AC power wiring thus eliminating the need for an equipment ground. They are finding the use of an equipment ground causes more problems than the added cost of the double insulated power wiring.

That's interesting; maybe one of the problems they are finding is exactly what I was describing -- voltage offsets in safety grounds causing noise currents to flow through signal return paths, a problem which would be eliminated if signal ground and ac safety ground were not both tied to chassis.

Stray voltage

Interesting paper, but it doesn't seem to be particularly relevant, and I don't think it makes its case completely. His theme seems to be that cost reduction measures taken by utility companies cause ac return paths to power company equipment to be partially through the earth, instead of entirely through their wires, and that that is somehow harmful to people and animals. But I didn't see any explanation of how that current flow might result in significant voltage difference across any individual animal or person, other than brief mention of in-ground swimming pools (and I'd want to see more evidence or quantitative explanation before concluding that he is right in that case).

In any event, I don't think his paper has relevance to connecting an audio system ac safety ground to earth, when a very nearby connection to earth of that same ac ground would exist at the service panel.

Thanks for the interesting read, though!

Best for the holidays and new year,
-- Al

I provided the link to show what can happen if the equipment ground for a branch circuit was connected to an earth ground electrode ONLY.[emphasis added]

Yes, I agree completely, that would be a definite no-no.

Thanks for the good references!

Regards,
-- Al

Timtim,

As has been said, it is essentially an absolute must to have a ground rod at the service entrance, that serves ALL runs. For reasons of code compliance, effective circuit breaker operation, lightning protection, etc.

No one in this thread has advocated independent ground rods for each line that is run to the system. And I think that Jea has effectively made the case that having even a single ground rod dedicated to the system is pointless, as well.

The question of tying the dedicated ground runs (not rods) together at the system is a little less clear cut. As I indicated earlier, there is both potential upside to keeping them separate (improved isolation) and potential downside (offset voltages between chassis causing noise currents to flow in common with signal returns).

That downside would pretty much be negated, I think, in the case of a system that had balanced interconnects. But most systems have some or all of their interfaces single-ended. For those cases, I am not in a position to offer generally applicable guidance as to whether the upside or downside is more likely to prevail. It could be that trial and error is the only way to tell in a particular case. And as I said earlier, a more ideal solution would be to make use of isolation transformers as discussed in the other thread I referenced. It seems pretty widely agreed that both the isolation they provide and the filtering that results from their limited bandwidth make them a very effective approach.

Regards,
-- Al