If you're an old geezer like me, the idea of using unshielded cable to run audio is still a pretty new idea. Sure, those IT people who installed your computer network have been using unshielded twisted pairs (UTP) for over a decade, but that's computer data, not audio. And now there are other crazy people telling you to run video on UTP. What's with that?
If you attend NAB, AES, SMPTE or other professional audio or video tradeshows, you'll notice a lot of Category data cable everywhere. But that's probably machine control stuff, right? Just design engineers taking advantage of Ethernet that's a lot cheaper and easier to install than the RS-232 or 422 that you're used to.
But, wait a minute. There's a guy running four channels of analog audio down a CAT-5E or CAT-6. Is that possible? And there's a pile of surveillance stuff, cameras and controllers wired with RJ-45 connectors — the same connector as UTP.
And then there are the balun manufacturers. They make little black boxes that convert balanced twisted pairs to unbalanced devices. And they have dozens of applications: analog audio, digital audio, analog video, digital video, RGB, VGA and so on. Have they all gone off the deep end?
Before you throw yourself across the high voltage in your transmitter, things are not that bad. And this stuff actually works. All you need to know is how it works and why. Is UTP better than coax? Not necessarily. The secret is in those unshielded twisted pairs.
Back at the dawn of time (20 years ago), twisted pairs were pretty crude things. Sure, there were impedance-specific twisted pairs, called Twinax, but they were big and expensive, and you probably didn't even know they existed. About 15 years ago, people began to play with UTP. First they put it in levels (invented by the distributor Anixter), then the standards group EIA/TIA 568 put them in categories, such as CAT-3 and CAT-4. (The only one of those still in the standard is CAT-3. The FCC made it the standard for telephone wiring.) Then along came CAT-5. And this was the cable that made everyone sit up and take notice. For one thing, it had a bandwidth of 100MHz to carry Ethernet. They called the signal 100baseT. (The T stands for twisted pair.)
Figure 1. This test of far-end crosstalk on stranded Belden CAT-5 patch cable 1752A at audio frequencies shows the average of all pair combinations with crosstalk from 1kHz to 50kHz. Click here to see an enlarged diagram.
Right away there were improved versions — some with the pairs spread out for ultra-low pair-to-pair crosstalk and some with bonded pairs for excellent impedance stability. But even the standard generic CAT-5 was pretty amazing, and some people started to look at it for non-data applications. So, 10 years ago at Belden, we ran some tests. Yes, 10 years ago!
Figure 1 is a test of crosstalk done on stranded Belden CAT-5 patch cable 1752A (even worse than standard solid CAT-5 cable) at audio frequencies. Shown is the average of all pair combinations showing crosstalk from 1kHz to 50kHz.
Braid and foil
You can see the worst case is around 45kHz, where the crosstalk is just a bit better than -95dB. In the human ear range, the worst case is around -97dB, and for much of the range it's -100dB or better. A 100dB crosstalk? Hey, what's the trick?
Well, those of you who knew something about data cable probably noticed it said FEXT above Figure 1. FEXT stands for far-end crosstalk. That's where the signal is the weakest, at the far end of the cable. Therefore, the other end, where the signals are strongest, has probably really bad crosstalk. Figure 2 shows NEXT, or near-end crosstalk, on the same cable.
Figure 2. This test of near-end crosstalk on stranded Belden CAT-5 patch cable 1752A at audio frequencies shows the average of all pair combinations with crosstalk from 1kHz to 50kHz. Click here to see an enlarged diagram.
So why didn't we use unshielded pairs before? And why all that shielding, braid and foil? Well, the phone companies have been using unshielded pairs since they began. And they could easily deliver -60dB SNR at 15kHz, as anyone who rented a line could tell you.
And shields? Well, there are two kinds of shields: braid and foil. Braid shields are good at low frequencies (up to around 10MHz); foil sheilds are good at high frequencies (10MHz on up). So all those foil pairs were really just giving you RF protection. And no shield has much effect below 1000Hz.
Cable for digital
So what got rid of the noise? It was the twisting of the pair. It was the balanced line! And the better the balance of the source and destination devices, the more noise was rejected. That's called common-mode rejection ratio (CMRR), something the phone company already knew about. Poorly balanced (or worse, unbalanced) pairs pick up noise or radiate their signal. Good balance would reject almost anything — 60Hz, RF — you name it. Instead of putting a Band-Aid (shield) around a poor pair, we fixed the pair.
So then we tested some super-CAT-5 with bonded spread-out pairs (Belden 1872A MediaTwist). The lab technician was crying on the phone because he couldn't provide the test data. Why? Because he couldn't read it! The crosstalk was below the noise floor of the $60,000 network analyzer. And what was the noise floor? -110dB! The average pair-to-pair crosstalk at analog audio frequencies on a really good cable (such as today's CAT-6) is unreadable.
Brave souls, or those who wanted a cheap way to send such audio as background music, immediately jumped on UTP. And slowly, some professionals have moved that way as well. Then along came digital audio.
Now AES digital is different from analog. It has cable specs, for one thing: shielded pairs with an impedance of 110Ω. Obviously, if you have to meet the spec, then you can't use UTP. But if you're the keeper of the spec or if your only question is “Will it work?” then we can take a look.
Double-, triple-duty cable
Category cables are all 100Ω impedance. Is that close enough? If you calculate the mismatch, it comes out to a return loss of -26dB and a match of 99.75 percent (i.e., 0.25 percent reflected). Sounds pretty good to me. In fact, the most popular audio use for Category cables is AES audio. The highest sampling rate is 192kHz (24.576MHz bandwidth), but it's unlikely you'll be running past 48kHz sampling (6.144MHz bandwidth).
And, since this CAT-5 or 5E cable is made to run 100MHz, then 6MHz or 25MHz is easy. CAT-6 is tested to 250MHz, so it's easier still. And, by the way, even though there are a billion feet of installed CAT-5, it's no longer part of the standard, so CAT-5 is hard to buy these days. You'll have to settle for 5E (enhanced 5) or CAT-6.
And, when you think about it, it just makes sense. You can use one cable to run all the computer stuff (and there are computers galore in any broadcast facility), and then you use the same cable to run all the digital audio. Hey, if you're brave enough, it can run the analog audio. And while you're at it, you can run the phone, fax and modem down the same stuff, too.
So now we have vendors selling us on audio running down CAT-5E or 6 as an Ethernet bit stream. Peak Audio has CobraNet, which can run 128 channels down a piece of CAT-5. Digigram has EtherSound. And now everyone is getting into the act. At Gibson, the guitar folks have maGIC, which will do 32 channels of audio. And if you run 1000baseT (Gigabit Ethernet), then it'll carry 320 channels. How big was your facility, again?
Then we have the Telos folks and their Axia networked audio. They just convert everything — mic level, line level, control — to 100baseT, and away you go.
The Pair Splitter from ETS breaks out each pair in a four-pair UTP run to a single pair in separate RJ45 jacks.
And the really cool part of all this 100baseT is that the routers and other gear are cheap, off the shelf and at your local electronics store. Easy? You bet! So what's the downside? There must be a downside.
The downside: If you're running analog audio, the four-pair format of data cable does not lend itself gracefully to single-channel stuff. Often, you will waste three of the pairs to get a signal going. There is two-pair cable, but it's hard to find.
However, there is a cool little box made by ETS, one of those balun manufacturers. I call it the pet rock of the electronics industry because there is nothing inside except wire.
Each of the four jacks has one live pair. And there's a single jack on the other side where all four pairs are combined. Then you can run four signals down four pairs, even when they didn't start that way.
So, I hear you asking, “What about video?” Well, I've run out of time and space, so we'll have to continue next month with a second installment about UTP. And we'll look at more than just video. How about RGBHV, VGA, broadband/CATV or SDI/601? We'll look at all these and much more.
Steve Lampen is the multimedia technology manager for Belden. He holds an FCC Lifetime General License, is an SBE Certified Radio Broadcast Engineer and is a BICSI Registered Communication Distribution Designer. His latest book, “The Audio-Video Cable Installer's Pocket Guide,” is published by McGraw-Hill.