Who Put the Gory in Category?
I’ve spend more than a few columns in the past extolling
the virtues of Category 5 premise/data cables. And more than a
few readers have taken the plunge and are using it not just as
data cable but also for things like RS-422 and RS-485 machine
control, digital audio or even analog audio.
This last application, analog audio, is probably the most controversial.
After all, I am advocating the use of unshielded cables ("UTP")
to run signals that have always been shielded, right?
Well, the phone company has run audio signals on unshielded
cable for years. Of course, their cable is not data cable; it’s
POTS, plain old telephone service, cable. And it works up to a
high frequency of 3500 Hz. So let’s ignore it for now.
We’ve done some testing at the Belden Engineering Center
running analog audio down UTP, just to see how good it is, especially
in crosstalk. Part of the problem is that, for data, the specs
don’t start until 772 kHz, long after analog audio. So I
told the engineer in charge of the testing, Dave DeSmidt at dave.desmidt@belden.com,
to test the most generic Category 5 UTP he could find.
Analog and CAT-5
He ended up using Belden 1752A, which is a stranded Category
5 patch cable. Stranding makes any cable seriously worse at high
frequencies. However, at analog audio frequencies, solid vs. stranded
doesn’t make much of a difference. The only bell or whistle
this cable has is that the conductors are bonded together. This
improves impedance at the data frequencies, but, as you should
know by now, impedance at analog audio frequencies is not important.
(A quarter wavelength at 20 kHz is more than 2 miles.)
The bonded pairs do improve twist consistency and therefore
probably improve crosstalk down at those low analog audio frequencies
(20 Hz to 20 kHz) since the pair remains more "balanced."
So 1752A was probably was a good choice for all the testing.
So what did we find?
Maybe I forgot to tell Dave just where to stop testing, because
he came back to say that he had tested all the different pair
iterations. There are six combinations in a four-pair cable. You
know, Pair 1 to Pair 2, Pair 1 to Pair 3, Pair 1 to Pair 4, Pair
2 to Pair 3, Pair 2 to Pair 4, Pair 3 to Pair 4.) He then averaged
the crosstalk between all of them.
The worst case was a crosstalk of -95 dB at 50 kHz! Below 20
kHz, average crosstalk was around -100 dB or even lower. This
was done with cables of 100 meters, or 328 feet, in length, which
is the maximum cable length in the standard data specs. So what’s
the problem?
Even generic Category 5 cable will do just fine running analog
audio. The only caution flag I would put up is to say that the
balance of the source and destination devices also influences
these crosstalk numbers. The better the balance, the lower the
crosstalk will be.
There are many people out there who still don’t understand
what this means. If you need CD-quality noise-floor, generally
around -90 dB, you can get this kind of performance, and better,
from an unshielded generic Category 5. And, if you need better
than -95 dB ...
We also tested our top-of-the-line cable, Belden 1872A ("MediaTwist").
The problem was, at analog audio frequencies, we couldn’t
read the crosstalk! Now our Network Analyzer has a crosstalk "floor"
of -110 dB, so the crosstalk at all audio frequencies was below
-110 dB.
I hear some of you old-timers say, "Well then, why did
we fool with those shields for all those years if we didn’t
need them?"
Why, indeed? Most shields in install cable are foil shields.
They don’t even begin to have any shield effectiveness until
well into the Megahertz. In other words, they are RF shields.
If you put in a heavy-duty high-coverage braid shield, you can
get that effective shielding down to 100 kHz. Below 100 kHz, there
is no standard shield that has an effect at all!
In other words, down at the audio frequencies themselves, especially
if we’re comparing shielded and unshielded cables, those
foil shields are doing absolutely nothing.
Therefore, if you have poorly designed and poorly twisted pairs,
you’ll get crosstalk. And this is where the advantage of
Category 5 data cable comes in, because those pairs are precisely
twisted, sometimes even bonded, so their crosstalk performance
for analog audio is amazing.
A more weighty issue
I used to weigh a lot more than I do now. In fact, I have lost
over 100 pounds in the last three years. So, you’ll understand
why I used to be called "Lumpy" by a lot of my friends
at Belden. I often play "Stump the Lump" when I give
a question-and-answer session at one of the talks I give to broadcast
engineers.
Well, I did get stumped with a comment made by Dave Obergoenner,
who works for the Zimmer Radio Group. I was giving a presentation
to the St. Louis section of the SBE, and we were in a free-ranging
discussion of Category 5 and its enhanced versions. Dave said
there was one problem with Category 5 UTP, which could only be
solved by shielded Category 5. You can tell from my diatribe above
that I was ready with a dozen answers why the unshielded version
was better.
"Oh, yeah," I said, "What is it?"
"Lightning!" said Dave.
That’s right, lightning! Seems Dave was working at a station
that was struck by lightning. The equipment attached was ruined;
at least 80 percent of the equipment was damaged or destroyed.
The stuff with shielded cable gave the lightning something to
conduct down, and a greater percentage survived. Dave said only
20 percent of that equipment got fried.
So I can’t honestly say UTP is the answer to every problem.
And there is a lesson here, if you do use shielded cable. I’m
not so sure the "telescopic" grounds you often use would
help with lightning. You know, I’m talking about connecting
the ground at one end to prevent group loops.
That’s one of the great advantages of UTP: no shield, no
drain wire, no ground, no ground loops. In fact, go to your computer/data
manager and ask, "What is a ground loop?" I’ll
just bet they can’t tell you. Wouldn’t that be a wonderful
world to be in?
Unless, of course, you want to be lightning-safe. In which case,
you might want to use shielded cable. Then you have to make sure
you have a very good ground. To avoid ground loops, you should
really put in a star ground system, so that you can ground both
ends of the cable.
Dave says that lightning exhibits a frequency range between
2 kHz and 2 MHz. You have to present a low-impedance (R-C-L) path
for these frequencies. That would also mean the Neil Muncy "Pin
1" problems comes in and you’d better hope that Pin
1 is grounded to the chassis, and not to something on the circuit
board inside.
I’d bet that those devices that have a ground trade or
bus bar inside the chassis are among the 20 percent that were
fried by lightning, even with a good ground connection.
I’m no meteorologist, and would be delighted to hear from
anyone with more information on lightning.
Steve Lampen is technology specialist, multimedia products for
Belden Electronics Division in San Francisco. His book "Wire,
Cable, and Fiber Optics for Video and Audio Engineers" is
published by McGraw-Hill. Reach him at shlampen@aol.com.
