There are many qualitative statements on the web, that say that EMI sensitivty decreases when going down the list mentioned above (in other words: UTP is most prone to EM interference, fiber is least or not at all prone to EM interference).
I'm looking for references, where I can find a table that gives concrete values or value ranges, such as a book or a web site. Google came up with a lot of results on related issues, none of them actually listing concrete values.
Finally, some discussion which kinds of environments (engines etc.) may cause interferences that affect data transfer via these cable, would be most appreciated.
I doubt you will find many reliable references let alone numbers. A test like that is extremely hard to produce and maybe harder yet to verify. Also, cabling manufactures are rarely if ever have access to equipment and facilities required to conduct such tests, especially RF. I have seen tests done (even took part in one many years ago) but the results were very dependent on the environment the tests were conducted in.
The one I've assisted to conduct was strictly RF, so I would not necessarily project the results onto low frequency noise from likes of power cables and fluorescent lamp ballasts. However, for RF it boils down to this: before your UTP (CAT5E and CAT6 - have not seen this done to CAT3) cable will pick up enough noise your equipment -PC and pretty much everything else - will fail (and people flee ;-)) because you are talking way beyond FCC and EC allowed levels. You will never encounter such high levels of EM fields in commercial and residential environments so there is really no need to worry about the cable. Let me re-phrase that: worry about the electronics before worrying about the cable.
If you are in a special environment (industrial, science lab, ship, plane, spaceship etc.) - you are on your own. Feel free to conduct your own tests and come up with your own results - no cable manufacturers will ever guarantee anything or even conduct such (very expensive to setup) test.
As far as fiber is concerned, for all practical reasons there is no EMI effect.
STP or ScTP will provide some degree of RF protection if properly grounded on both ends which it never will due to ground loop concerns. It will not provide any degree of protection for low-frequency EMI because most of its energy is in magnetic component of the EM field and aluminum foils used in STP are transparent to magnetic fields.
Coax cables are really for so different applications that I don't know why would you even throw them into the mix, but all the STP considerations apply here as well.
Best Regards, Dmitri Abaimov, RCDD
Cabling Guide, Cabling Forum, color codes, pinouts and other useful resources for premises cabling users and pros
-- ##-----------------------------------------------## Article posted with Cabling-Design.com Newsgroup Archive
Web and RSS interface to your favorite newsgroup - comp.dcom.cabling -
2650 messages and counting! ##-----------------------------------------------##
The EMI issue began when Ethernet over UTP was really new and I've never known it to be anything but theoritical. I don't even recall "CAT3" as a specification when I first put Cabletron hubs in service. At the time somebody did some practial test, wrapping loops of what we would call CAT3 cable around zerox machines,arc welders, small power tools, etc, while running BERT tests. As I recall, the BERT error level was ZERO. I don't think I've ever heard of EMI-induced errors in commercial use of UTP. If I were installing a network near an MRI machine or near an electric arc furnace I'd use fibre, but I'd have lots of other issues, too.
Even if there were an occasional noise shot, A TCP/IP network would resend the packet very nicely, thank you.
UTP cable is twisted pairs of wire. One byproduct of that is that any external EM field that would induce voltage in one wire induces an identical, opposite voltage in the other wire of the pair. They cancel out. For perfectly symetrical pairs EMI induces zero net voltage.
As I understand things, UTP wire, in each step from CAT3 to CAT 6 has improved the exactness of the twists. This makes CAT6 much more EMI-resistant than CAT3, especially what was in use when the tests I recall were done.
AFAICS, Cat3 to Cat5 was an improvment in EMI (all twists much tighter), but since then, most of the improvements have been to reduce crosstalk (twists slewed). I think Cat5/5e/6 are all pretty much the same for external EMI, although some of the siamesed-pairs might be slightly better.
Well, one is that it depends much on the transformers at each end of the cable. It is very easy to pick up a common mode (the same voltage or current on both wires of a pair) signal with UTP cable. The transformer at each end keeps that signal away from the rest of the electronics. There has been discussion that UTP ethernet should work with a 240VAC 60Hz common mode component on the cable. I have thought about testing this, but still haven't done it.
The twisting of the pairs keeps out most external differential mode signals. Signals with a wavelength shorter than the twist pitch, and localized so they don't average out over the cable length could get through. With the usual twist, that might be about 10GHz, and at fairly high power to couple in enough to interfere with the signal.
Coax grounded at no more than one point is, for similar reasons, pretty good against common mode signals. If you ground at more than one point it is fairly easy to get large currents through the shield (especially at 60Hz) with no compensating current in the center conductor. Ethernet transceivers for coax are electrically isolated (with a transformer coupled DC to DC converter, and transformers on the AUI signals).
The physics of the coupling is different for different cable types, and it depends very much on the isolation (especially transformers) used. There is also some frequency dependence. For all those reasons, it can't be given as a property of the cable alone.
It can also be looked at this way: the cable is only providing path for electromagnetic signal that would otherwise have to travel through the air in a form of radio waves. You wouldn't say: "this air is too susceptible to EMI", wouldn't you? You'd say: "this receiver is not sensitive enough" or "its signal to noise ratio is not high enough".
An interesting way to look at it, but continuing the analogy, the cable is a _huge_ bundle of antennae. It might be nice if it were well-balanced so the differential signals would show through equal interference.
Well, agreed - not the best analogy. The point being that cable is passive. It does not introduce any more EMI than is already present in the environment. Moreover, it actually helps somewhat by attenuating the noise picked up in the beginning of the line before it hits the receiver. Of course it has its unique properties not present in the open space, such as delay skew, but please don't take the analogy I made literally.
As far as "balancing" of the cable is concerned, it means vastly different things to different people. For EMI susceptibility purposes it usually means "symmetrical" and is only as good as the "balance" or symmetry of both transmitting and receiving opamps and transformers. In any case, it is hard to over-emphasize importance of the active electronics in this entire EMI issue.
Good point. Us cable guys cannot get too easy off the hook. Well, gotta carry at least some responsibility in order to get paid, you know ;-)
I'm not sure what you are referring to here. I was talking about external EMI, and a differential signal. If the cable (pair really) is perfectly balanced (ie receives the EMI equally), then how will it have difficulty with ANY interfering frequency?
To put a little real-world into all this theoretical talk. I've encounter only 2 situations where EMI caused a problem, both were with Cat3 UTP when that was new and replacing coax. And both were related to the installer not following basic guidelines. The first was when the tech didn't have a shorter cable and tightly coiled up about 20ft of Cat3UTP along with some power cables that were lying around. This was in the Hub days, so the whole hub went down but only when that user turned on his computer. The second was a bitch to find since it only occurred when a storage closet light was left on for an hour or more. The light's transformer was failing and causing interference on the Cat3UTP cable that had been conveniently laid right on top of it. Again it was a hub and the excessive retransmits killed the whole segment. I had one issue with coax (excluding all the bad crimps) but it turned out to be RI, Rat Interference, the cable ran along an I beam in an industrial machine shop and they would get the strangest results when the equipment encountered excessive errors when accessing instructions on the server. This happened whenever the rats chewed on the cables.
"Al Dykes" wrote in message news:e7pt76$9te$ firstname.lastname@example.org...