Broadcast & AV

Ground Loops

Steve Lampen

If you work with any kind of cable, certainly audio and video cables, then you are familiar with the shields in these constructions. Some are braid shields, good at low frequencies, starting at 1 kHz and slowly deteriorating around 400 MHz. (The openings in the braid structure begin to look bigger and bigger at shorter wavelengths.) Then we have foil shields. These are superior at high frequencies, starting around 10 MHz and going as far up as we can go. (Up to 20 GHz in some designs.) The foil is too thin and fragile to use it as a connection point, so, if you only have a foil shield, we give you a drain wire, a bare wire touching the foil that is your connection point. The ideal shield is a combination of foil plus braid. And the best of those is the highest braid coverage (around 95%) with a foil underneath.

You can even improve on that. A tri-shield, with an outer and inner foil, and a 95% braid in-between, is even better shield effectiveness. We have one digital video cable like this, Belden 1794A. Quad shields (foil/braid/foil/braid) would probably be even better, but I don't think it has even been tried with two copper 95% braid shields. It would require special connectors since the cable would be much bigger. If you know quad shielding, it's probably in the broadband/CATV world, where the two braids are 40% and 60% coverage, and are made with aluminum wires. This low braid coverage makes a huge difference in shield effectiveness.


In fact, we recently tested one of our off-shore competitors who claimed to make identical cables to our precision digital video cables, like 1694A. Testing dozens of reels bought on the open market, not a single one got close to the 95% shield we make. Their closest was 84% and their worst was below 50%. I was talking to one of our engineers at the IBC show in Holland. "All you have to do is weigh the reel," he told me. We all use the same plastics and same wire (pretty much), so if the reel is 20% or 30% less weight than the Belden stuff, well, that difference has to come from somewhere!


If you have a foil and braid, you don't need a drain wire. The braid is also the connection point for the foil. If you put on a BNC, you would make contact with both foil and braid, although the braid is the main mechanical connection. However, we do make one cable that has a braid shield and a drain wire. That is our top-of-the-line microphone cable, Belden 1800F. It has a flexible tinned copper French Braid shield, with a bare copper drain wire underneath. The drain wire makes it very easy to connectorize this cable. And it's a different color (copper) from the shield above (silver-colored tin). You don't have to separate and un-comb the braid. Just cut it off and use the drain wire. Simple! Here's a picture of that construction.



Now. if you're a brainiac, like many I talk to online or at tradeshows, you might notice one contradiction in what I said at the beginning: foil shields are for high frequencies. But where do you most often see foil-only shields? Inside audio cables like Belden 8451 or 9451, or any of dozens of analog multipair snake cables. If this kind of shield doesn't begin to work until 10 MHz, and that's way past analog audio frequencies, what is it for?


Simple! It's to help prevent RF from getting into your mixer or other device by way of the mic or line connections. So what prevents noise at audio frequencies? The fact that the pairs of wires are twisted together and run as balanced lines! This is why most data cables are UTP, unshielded twisted pairs, because they are run as balanced lines which reject noise. And this works for very low to very high frequencies (Our current high frequency champ is Belden 10GX32, 10 gigabit data cable, at 625 MHz per pair.) In fact, I could make a pretty good argument for the fact that twisted pairs, run as balanced lines, beat any shield in real shield effectiveness. You could make the argument that at the highest frequencies, in the Gigahertz, a loose twisted pair begins to not be so accurate, so a foil shield might beat it at that point. But GHz are a very long way from Hz, and I would be interested to hear any argument about how noise in the GHz region could affect audio at 20 kHz or below. Send me a comment below!


So, I finally get to the core of today's blog: how do you use these shields? How do you hook up those braids or those drain wires? And what if you have a problem where the ground potential at the source end of the cable is different at the load end? That potential difference starts to flow down the shield. And all the noise, RFI and EMI, that the shield would take away is now fed down the entire length of the cable. That shield is now an antenna, feeding hum and buzz into your pair. Of course the pair rejects as much as it can, but no pair is perfect.


The real solution to this is to design your facility so that all grounds everywhere are the same potential. This is sometimes called a "star ground". But if your building is already built, making it a "star ground" would be nearly impossible. So, instead, most installers resort to the next best thing- cut one ground. That instantly solves your ground loop problem, but which end do you cut? The answer is, leave the source (the low impedance connection) and cut the destination (the high impedance end). There are some places, like patch panels, where it is almost impossible\e to tell which is which, so good luck choosing the right one to cut. And, of course, you're now at 50% shield effectiveness (one connection instead of two).


But your problems don't end there! Below is a graph of a shielded data cable where the source ground is connected and the destination ground cut. You will notice that what you have now created is a very nice filter at high frequencies. Lucky you!




So now you see even more advantages to UTP data cable: no ground loops, no filter effects. So why not use UTP for audio. Sure you can, except where you have to have a ground, such as phantom powered microphones. (And there's no cutting one end of any cable running phantom power.) If it's just audio, no problem. Check out Belden 1353A. It's really a Bonded-Pair Category 5e patch cable, but it's in the audio section of our catalog! It's the best audio cable we've ever made. Just don't run phantom power down it - won't work.


If you're running unbalanced signals, such as video coax, then you have to connect both ends. And if you have ground loop problems, there's a cool little tester you can buy. They're at and here's a picture I stole (with permission) from their website if the two units in action. Go check them out.




As you can see, this will tell you if you have a ground loop without disconnecting anything. If you have a story about shields, grounds or ground loops, send it to me or just fill in the comment section below.