[time-nuts] Ground loops in measurements?
joegwinn at comcast.net
Tue May 21 10:20:20 EDT 2013
time-nuts Digest, Vol 106, Issue 101
> Message: 6
> Date: Mon, 20 May 2013 17:16:02 -0400
> From: "Bob Camp" <lists at rtty.us>
> To: "'Discussion of precise time and frequency measurement'"
> <time-nuts at febo.com>
> Subject: Re: [time-nuts] Ground loops in measurements?
> Message-ID: <5654054C635D47F2A251C902A875F035 at vectron.com>
> Content-Type: text/plain; charset="us-ascii"
> The braid becomes less effective at lower frequencies due to skin effect.
> Once that happens you have a magnetic loop that is not as effective at
> rejecting signals as a twisted pair.
> Looking at it another way, the threat signal is induced on the shield
> (braid) and it penetrates to couple to the center conductor.
> There are a couple of other ways to look at it, they all ultimately come
> back to the same point. There are a few other odd things that happen as well
> as you drop frequency, none of them really very helpful for moving signals.
> Now - is it actually a problem or not? As always, that depends. Most of the
> nasty stuff in a 1 pps is in the edges. That energy (with a fast edge) is
> nicely captured by the coax. The low frequency stuff is not captured as
> well, but it also doesn't couple all that well either. Making a transformer
> work to 1 Hz isn't very easy. So, coax is fine for getting the signal to
> it's destination. It's only when you look at the ground loops and threat
> signals that there could be an issue.
Another way to think about it is to notice that skin effect applies
independently to both inside and outside surfaces of the shield. If
the frequency is high enough, the other and inner shield currents are
independent of one another, being coupled only by holes in the shield.
This leads to a critical issue: At power-line frequencies, ground-loop
currents in the shield cause a voltage drop that appears between center
and shield. So, one can have really good coax, with solid copper tube
shields, and still the ground-loop current gets in. The only solutions
to prevent the ground loop currents from flowing in the shield and/or
to make the system immune to power-frequency (and harmonics)
interference coming in on the center conductor.
An inline RF transformer can be used to do both at the same time. But
you often need a small RF capacitor in series to prevent transformer
core saturation due to the ground loop current - the varying core
saturation can cause the loop current to be modulated onto the signal
passing through, causing phase-noise spurs at the harmonics of the
This can be a particular issue for DMTD systems - the 1 Hz difference
signal is hard to contain, and tends to leak into undesired places,
causing synchronous jitter. Ground-plane islands and RF transformers
help a lot. One can carry the 1 Hz signal by LVDS over shielded
twisted pair, with shield interrupted at one end (so the island remains
an island), the island being DC connected to the mainland only by a
As for books on grounding, shielding, and ground currents, my favorite
is Ott's book on EMC Engineering:
<http://www.hottconsultants.com/index.html>. Ott goes over the issues
in handling mixed analog and digital signals in great detail.
More information about the time-nuts