KurtBJC

Well, but 100BaseT over copper is 4B5B, MLT-3 encoded. So three voltage states, and a fifth bit added to "words" of four bits. Maximum fundamental is 1/4 the baud rate. 100BaseTX is the most common, and only uses two of the four pairs, one in each direction. So, 100 Mbps becomes 125 Mbps via 4B5B, which gives a 31.25MHz fundamental frequency after MLT-3 encoding. I may have mentioned above -- it's the third harmonic which is important here because it carries most of the energy that distinguishes a sine from a square wave, and sure enough, we test Cat 5e cable to 100 MHz which is just above the 93.75 MHz third harmonic of the 100BaseTX signal. And, Paul: you probably already have noticed, but we do make the 6 and 6a patch in six colors. We'll be adding more, probably to get to ten, over the next year. We just don't have any other than black up on Amazon yet, but that'll change as sales pick up there. The trick with Amazon is that if we send them too many items, we wind up needing to bring some back.

Around here, we spend a lot of money both on probes and on strippers. Between the stripping and the (ultrasonic) welding, sometimes it's pretty much Flashdance over here.

True. And in 5e, at least, it's pretty easy to do high-flex. As you go up, it gets tougher, especially because the good cable stocks use pair-separating splines which are often not especially flexible.

The TIA spec suggests, but does not require, stranded conductors for patch cords. But really, the flex life of solid copper conductors is excellent and I think that's unnecessary. The biggest issue, performance-wise, is that impedance stability is ordinarily better with solid wire. Why? Well, it's a sort of a symmetry/topology issue. A solid wire presents a smooth rounded face to the outside, which means that when a cylinder of insulation is extruded over it, the insulation is very close to the ideal: equal thickness at all points. Think about it in cross-sectional slices of the cable and you see what this means: the relationship of the conductors of the pair to each other, and the amount of dielectric between them, is more consistent in the solid wire case. There are practical issues that can crop up as well; we tend to think of stranded wire as being extremely uniform, but any manufactured product really is not, and so we're not always looking at a really geometrically tight seven-strand bundle with a rounded hexagonal outer profile--we may have spacing within the strands, we may have irregular twist, and things like that which affect the shape of the wire. When I talk about things like this, people tend to think it's a bit silly -- that it cannot possibly matter whether the strands are oriented just so or not. But high-frequency signalling is a strange world, and all kinds of dimensional tolerances that make absolutely no difference in the DC or low-frequency world start to matter. Considerations like these do cause impedance to be a little harder to manage in stranded cable, and when you have products like Cat 6a which are difficult to make compliant, little fractions of dBs matter.

If you have any of those still on hand and would like to know how they test, I'd be happy to test 'em for you. I am surprised to see individual shielded pairs on a Cat 6a, and especially to see what look like outward-facing foils (foil shields are usually bonded to mylar, and either have the conductive face in, or out, though there are also two-faced foils as well), in contact with one another and in contact with a braid. I would expect inward-facing foils as there can be squirrelly results when you've got shields touching each other at these frequencies, especially if they don't touch consistently from place to place. But it's hard to be sure, from the photos, exactly what the deal is. On the "Cat 7" one--I can't test it at Cat 7 but I certainly could see whether it passes 6a. By the way, re: Black being the only color on our Amazon listings: we will be adding more colors. It's just that our experience has been that products start to sell on Amazon very slowly, so we don't want to load them up with dozens and dozens of items that might not sell. As the black ones pick up sales, we'll broaden the offerings. Forums can indeed suck productivity out of one's days. Fortunately for me, though, public outreach is one of my jobs....

Thanks! Glad to hear it. And, as you note, it is indeed a 10GX series cable. For people who are building networks, that's a relevant consideration because while these do meet spec and stand on their own just fine, there are some things about the 10GX system which they support as well. In particular, 10GX patch cord stock and horizontal cable have offsetting crosstalk characteristics -- the "worst" pairs for crosstalk in the patch match the best in the horizontal, and vice versa. One other note re: it being a 10GX cable -- on a question you haven't asked but which people do ask -- while it is a 10GX series cable it does not correspond to any Belden part number you can buy, and it's not even the same stock as the Belden-terminated patch cords which Belden makes in Canada. There are a few patch cord stocks in Belden's custom-order range, and this is one of those. Our Cat6 meets that description in all respects except that it's not a 10GX cable. Here's the deal: When we test Cat 6 assemblies at Cat 6a criteria, they ordinarily pass. In fact, they often pass with larger margins than the 10GX cable does. Now, if that's the case, why are they not deemed "Cat 6a" cables instead of mere Cat 6? The difference is this: all cables must, to be deemed compliant, be made from components which comply to the applicable component specs for the category, and THEN must also pass the return loss and crosstalk specs as assembled. Our plugs are 6a compliant, and our 6a patchcord stock is 6a compliant, but the regular "6" patchcord stock would have to meet 6a alien crosstalk standards to be a 6a compliant patchcord stock. I believe that this is the only reason that our "6" patchcords cannot be deemed "6a." Now, as you will likely already have considered, this is completely irrelevant to you if you are not running the cable in a big bundle of cables. Alien crosstalk is all about what happens when these are run in close proximity to one another, and the standards are written with huge cable trays in mind. I don't know much about Cat 7, and I don't have a copy of the spec, but most of the cable I've seen advertised as Cat 7 seems very suspicious. First off, I'm pretty sure it is impossible to make an RJ-45 plug which is Cat 7 compliant -- there are Cat 7 connectors, but they look quite different and are not plug-in compatible (there are dual-purpose jacks, which will take either an RJ-45 plug or one of these Cat 7 plugs, but the RJ-45 mode is not intended to support Cat 7). To my understanding Cat 7 sees some use in Europe, but it's so rare in the USA that Belden doesn't even make it here. Copper-clad aluminum? That's odd. It is true, however, that skin effect at these frequencies is very strong and so it shouldn't matter much what the core is. I am not sure, however, that there are not issues in drawing wire like that--plating depth is going to be meaningful, and you wouldn't want it to vary much. The Cat 7 standard, as I understand it (again, I don't actually have a copy of the spec), does call for the pairs to be individually shielded. This will cut down on all crosstalk -- alien and otherwise -- but it makes connectorization more unpleasant, and it makes the impedance of the pairs a bit dicier to control because the shield affects the impedance and its wrapping cannot be done quite as consistently as the conductor-to-conductor spacing can be (not to mention drain wire--which messes with symmetry and spacing as well). I don't think I've seen any Cat 6 or 6a with individually shielded pairs, but there's nothing in the spec which would prohibit that sort of construction. Not sure--could be, though. Kurt BJC

Daudio: I think there are indeed glass TOSlink cables available. I've been looking around again and what I'm seeing, in a few cases, are people who report that they are using bundles of many strands -- 280 in one case -- of glass fiber. That would be one way to deal with the huge difference in aperture. I have never seen bulk cable stock of that sort on offer, anywhere, and I know that if I were to have someone here (e.g., Telecast) make it for me it would be very costly. But it may be that there's a factory in China that does the stuff--I have yet to find anyone who does, however. I doubt that glass is superior to POF for TOSlink in any respect other than attenuation. POF is rather lossy. The other advantage of glass would be that because the individual fibers are narrower, there are shorter reflection paths within the fiber and this will mean that transitions are better preserved -- sharper rise and fall -- than in POF. However, the data speed for TOSlink is so slow that I have difficulty imagining how that could matter. In data applications with high bitrates, however, that's a big deal; you want all of the photons traveling the same distance, ergo arriving at the same time. I don't know anything about the quality of the optical/electrical transceiver chips. It is easy to imagine that quality issues there are relevant. I know that we do see that they vary considerably in how much attenuation they'll tolerate, but beyond that I don't know anything on the subject. Kurt Blue Jeans Cable

Well, there certainly should be. But if there is, plainly they (and a few others) know something about this which nobody I've ever met in the glass fiber industry knows. I'd dearly love to know, but the call for long-run TOSlink isn't very great and the easiest thing to do is just convert to S/PDIF in most cases. Kurt Blue Jeans Cable

Pakistan Ordnance Factories tend to be lossy for digital audio transmission, so yes, Plastic Optical Fiber. And thanks for the note re: signature. For some reason I cannot figure out where one inserts a signature--been through the profile and do not see it. So, dumb question, undoubtedly with extremely obvious answer: how do I do that (other than, of course, manually on each message)? Kurt Blue Jeans Cable

Now, there's a mystery. I have looked and looked, at trade shows, through vendor catalogs, and everywhere I can think of, for glass that can be used for TOSlink. I know that some vendors claim to sell them -- and yet everyone I have ever talked to in the glass fiber business tells me that it's nonsense, and that you can't do it because the aperture is all wrong. POF is huge by comparison to any glass optical fiber I've seen, and you've got to have a big aperture at the source or you get hardly any of the light. I have no idea how to do it--it seems as though you'd need a whole lens assembly just to focus the big source light down to a small aperture, and then another to take that and spread it out at the other end. Now, maybe someone has a way to do that, or some sort of wacky glass fiber I've never seen offered anywhere -- but I'm not too sure. I have half a mind to buy some glass fiber TOSlink cables and cut 'em in half to see if they're actually POF. Now, if it can be and is done, the only real advantage would be attenuation. It'd be a lot easier to get signal to travel long distances. But the bitrate in TOSlink is so slow that even the not-so-good attributes of POF are just fine for it, so long as the run isn't too long. Kurt

Fiber's nice, yes. Glass fiber, anyhow; POF isn't so great. I'm planning on getting some fiber patch cords onto our site in the not-too-distant future. Glad to hear you've got some of our cables! We do try to get them into every household...

Oh, heck. You know, I actually have no experience with audio over ethernet--I think it is fair to say that in any application it's good to have well-made product that actually meets spec, and our recent review of ethernet cables found that a horrifying number of them not only don't meet their stated spec but don't meet the 5e spec either. One cable we tested, allegedly a "Cat 6," failed return loss at 5e by 8 dB, and I would not want to rely on that for any data, at any non-negligible speed. Could it make a difference? Yeah. Does it often, in practical fact, make a difference? Dunno. I would not spend a lot of money on Ethernet cables; but being sure you buy something good (not necessarily expensive) can be important. Well, sort of. I think you and I may have the same thing in mind. Yes, the idea is to more or less instantaneously transition between voltage states, holding the voltage state until the next bit begins. That's not what I would term "pulse," which to me implies bursts with spaces between. And it transitions, of course, both up and down. The idea is to generate something akin to a square wave, but an actual square wave is not attainable. If you've ever looked at an "eye pattern" from an HDMI signal, that's the sort of thing (though in an eye pattern the ups, downs and sideways traces are all laid over one another) -- the transitions are sharper than a sine wave but less sharp than a square wave. Yes, true. And one also gets slower effective data rates if packets need re-sending, or if the signal quality is so bad that the whole system downshifts to a lower data rate. But the "native" data rate is the same in all these cases other than the downshift -- when the data are flying, they do it at the specified rate, but there are times when they're not. Kurt

Hi, there, all. I'm Kurt Denke, owner of Blue Jeans Cable in Seattle, WA. I saw this thread and thought I'd throw in a few observations relating to some of the things people have said here. First, on the Megabits = Megahertz thing: not quite. Remember that one "cycle" (if you're an old guy) or one "Hertz" (if not so old) is a swing from the baseline up, then back through the baseline, down, and back up to it again. What that looks like is not one bit, but two: a one and a zero, if "plus" is one and "minus" zero. For this reason, the convention is, when thinking in frequency-domain terms, to regard the frequency of a binary bitstream as one-half the bitrate, so 100 Mbps would be 50 MHz, for example. But it's a bit more complicated than that still, because most Ethernet signals are not sent in a straight ones-and-zeros format. There are different encoding structures which use multiple voltage levels, the most extreme of these in general use being PAM-16 which has sixteen different voltages. This is why we find Cat 6a cable, which is supposed to deal with 10 Gbps, tested only to 500 MHz which would equate to 1 Gbps/pair. The relevance of the frequency range is a bit more complicated, too, by the fact that digital signals are sent in something approximating a square wave. It's impossible to send actual square waves, but the transitions are more sudden than in the pure sine wave which we think of when we think of a diagram of a cycle graphed against time and voltage, and this suddenness is important because it helps us to correctly measure voltages and/or time the transitions. Now, a square wave can be represented as the sum of the primary frequency and lesser-amplitude doses of all of the odd harmonics (e.g., the third, fifth, seventh, ninth, and on to infinity...) thereof; and the behavior of the cable actually shows this -- one reason it's hard to push digital signals through cable at distance is that attenuation increases with frequency, so the higher-order harmonics effectively just disappear. But much of the energy in the augmented "shoulders" of the bit is in that third harmonic. So you want a cable which performs well at the fundamental, and that performs well at the third harmonic at least, if you want to get these transitions through well. That's why, for example, 3G SDI cable is now sweep-tested to 4.5 GHz -- the third harmonic of the 1.5GHz fundamental of a 3 Gbps stream. Some mention was made above of our Cat 6a cable and its floating shield. The floating shield is a bit of an odd critter, and it's not something one sees in a lot of applications. The reason it's there is that in Cat 6a we are concerned not only with crosstalk within the cable but also with "alien" crosstalk, which gets in from neighboring cables in a cable tray. Remember that this stuff is really specified with data centers and the like in mind, where there may be dozens of cables running together in close proximity -- that's why as we get up into higher bitrates alien crosstalk, which is not much of a problem in Cat 5e, becomes a big issue. That floating shield is not really a general-purpose shield, and the fact that hum can be induced through it is no surprise. The function of the shield is to manage alien crosstalk, and it does this by acting as a sort of reflector and dissipator--the idea is that crosstalk hits the shield, flows along it, and dissipates itself, burning up in resistance and/or cancelling out with other crosstalk flowing in the shield. One of the odd consequences of this is that where, with conventional shielding, one wants the shield as heavy as possible for high effectiveness, this shield must be just thick enough--because we want the resistance of the shield to help attenuate the crosstalk, which a thicker foil would actually do less welll. When looking at a floating shield on a Cat 6a cable, one needs to discard a lot of the conventional notions of what it is that makes for good shield effectiveness. If this were not primarily an anti-alien-crosstalk shield, we'd expect it to be thicker; we'd expect it to be tied to ground at both ends of the cable; and if it were expected to deal well with low-frequency, high-energy noise, we might expect the foil to be overlaid by a nice heavy copper braid. There are products like that out there, too -- e.g., Belden's "DataTuff" cables. There are of course people who prefer to use conventionally-shielded Cat 5e, 6, and 6a cables. For whatever reason, in Europe shielded Ethernet cables are very popular, and in the US not so much. Common mode noise rejection on well-made Cat 6 and 6a cable is really superb, so if the sending and receiving circuits are well balanced, a shield tends not to be all that relevant. Alien crosstalk presents special problems because of the proximity of the source, the similar or identical data rate of the source to the signal interfered with, and basic problems of cable layout -- if you've put a dozen runs of the same cable into a tray together, then there are only four lay lengths involved and so every pair in the tray has eleven neighboring pairs that are running at the same lay length. This is problematic for crosstalk because one assumption on which common mode noise rejection rests is that noise will hit the cable symmetrically; but when the source of the noise is twisting at the same rate as the receiver of the noise, it is anything but symmetrical. Kurt Blue Jeans Cable