Oddly, that's not true. The upgrade from 100baseT to gigabit was accomplished by (1) changing the signalling from two-level (zero or one) to five-level (+2,+1, 0, -1, -2) and (2) by running four signal pairs in each direction (while 100baseT uses one SEND pair and one RECEIVE pair).
So, gigabit doesn't have any faster timing than 100baseT, and at 100m (maximum length) a 5% propogation delay between channels is only 20 ns. That's a two-bit effect.
I think each channel is independently received in self-clocking fashion, so the skew isn't important. Four pairs, four embedded clocks...
doesn't seem very spectacular, no different than an 8bit bus versus a
16 bit one once you run out of speed you add more wires and move bits in parallel
the iso model has always been a bit artificial. My point was that before you can even start to think about packets, retransmissions etc. you need a physical layer that can some what reliable move bits from A to B
The whole local area network thing was based on the idea that a purely serial - one bit - transmission scheme was a lot simpler, and could be a lot faster (at the time), than any kind of parallel scheme when communicating over distances longer than a metre or so.
Once you run out of serial speed you don't have any option but to go to parallel connections, but you lose a lot of simplicity and ease of implementation in the process.
Back in the day, various token ring schemes were a competing with Ethernet, and none of them were very successful. Building Ethernets with cheap BNC tee connectors was popular, but the official systems used cable-piercing connectors, which were apparently rather more reliable for those places that could be bothered to find and buy the specified connectors.
Physicists do have this enthusiasm for improvisation, and an equal enthusiasm for damning the system design as unreliable when they've bodged their example together out of components that they could find in their desk drawers.
Whereas AlwaysWrong is the kind of idiot who shoots his mouth off without doing the calculation. The wire gauge were were talking about has a worst case maximum diameter of 0.15mm. We can take a twisted pair of this a having a diameter of 0.3mm - it will pack down a bit more, but not much.
The EP13 former in mentioned in the post you are reacting to (but snipped from your response)
formatting link
has a winding width of 7.5 mm, which means that you can get in 25 turns per layer. You can wind the layers 1.8mm deep, which - without any interleaving - would let you accommodate 6 layers, for a total of
150 turns. The mean length of turn is 23.8mm, so 150 turns is 3.57 metres - about 11 and a half feet.
A sensible person would sacrifice one of the layers of winding by interleaving each layer with 60micron thick Mylar tape, which would reduce the length to 2.975 metres - nine and three quarter feet. The EP7 former is almost certainly too small to take three feet of 43 swg twisted pair. An EP10 former would do it, or you might go for even finer wire.
I couldn't care less how you solve your design problems. I do get peeved when you exaggerate the difficulties of working with fine wires and twisted pair transmission lines - they aren't a good as coax, but it's easier to get a really thin transmission line in twisted pair.
In principle you can buy really fine coax - but when last I looked you had to buy a couple of thousand dollars worth from a Japanese supplier who only responded to requests for quotation in Japanese. The details may be wrong but degree of difficulty was definitely high.
Quoting: "The wire CANNOT perform at full cat6 speeds unless it was properly wired."
Kinda look like you're saying that the wire is responsible for performing something. My CAT6 wire is passive, just sits there, and has never performed for my amusement.
The wire does not govern, perform, or negotiate. Hint: CAT6 is a wire specification.
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
Once you wind up a twisted pair on a bobbin, the prop delay and impedance will change, and you'll get crosstalk. So spend another day iterating on all that, fiddling terminations and tuning the length of wires you can't handle except under a microscope.
All to replace four 0603 parts worth maybe 40 cents. Insanely expensive and impractical.
--
John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators
The RG8 cable piercing thing was messy and insanely expensive. A DEC cable tap block was about $2000, and the Unibus interface board was about $4K or so more. Networking got popular with cheap ISA boards and RG58/BNC "cheapernet" hardware.
Why are you so consistently nasty? What a way to live.
--
John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators
A question you should ask yourself. ...Jim Thompson
-- | James E.Thompson, CTO | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at
formatting link
| 1962 | I love to cook with wine. Sometimes I even put it in the food.
We use micro-coax to make transmission line transformers, where the transformer is key to a product's performance. Joy Signal #014 is available from stock. We use even smaller stuff, mostly assemblies connectorized with tiny H.FL connectors, available from Digikey.
No problem buying this stuff.
formatting link
You can push pretty big, fast pulses through a coax-wound transmission line transformer:
formatting link
The really fast stuff these days is RSPECL and then CML. We use a lot of LVDS, too. All positive power supply stuff.
You'll blow up all sorts of stuff if you drop ground leads all over the place. And you'll hurt your thumb if you hit it with a hammer. People learn to be careful when they do this stuff which, of course, you don't.
--
John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators
which at 1.3mm OD. is a bit thicker than the 1.22mm OD Filotex cable I suggested that you might buy from Farnell
formatting link
It's roughly four times thicker than 43 swg twisted pair, which is to say it has roughly 18 times the cross-sectional area.
connectors, available from Digikey.
The stuff I had in mind was significantly finer - some kind of weird mineral insulated cable that - IIRR - was made by feeding a rod of one metal into a tube or another, getting the interface hot enough that they fused, then drawing the combination out to form a very fine wire, and finally exposing this wire to a hot oxidising atmosphere so that part of the inner core oxidised to form your dielectric.
I never laid hands on any of it, and it may have been one of those products - like oerientation-insensitve mercury-wetted relays - that nobody could actually make reliably enough to make money out of
So what else is new?
Idiot. I was careful when I was doing this kind of stuff - and mostly used a high frequency probe with a sprung ground pin right next to the probe tip which doesn't get around as much as a conventional ground clip - but I did get to see that not everybody was quite as careful.
DEC was gouging. The cable piercing connectors came from the cable- television industry, and should have been cheap.
I'm not consistently nasty, but I do call a spade a spade. BNC T- connectors were always a bit dodgy - IIRR BNC was a MIL spec but it didn't cover T-connectors. Calling the whole Ethernet system unreliable because people tended to cheap-skate on the connectors was downright silly.
Why would we buy anything from Farnell, when we're half an hour from a zillion distributors in Silicon valley? Heck, I can order something from Digikey at 5PM and have it on my desk the following morning.
connectors, available from Digikey.
Laser drivers, 40 ps edges, time stampers, VME/VXI crate controllers, LVDS simulators, EUV sources, gradient drivers, EOM pulsers, all sorts of weird stuff.
I'm sure not going to put in a bunch of negative switching supplies, and level translators, for NECL logic, because I'm worried about dropping scope probe grounds onto PECL signals. Makes as much sense as going back to RTL.
--
John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators
Where does that statement declare that the wire performs any negotiation, you retarded piece of shit?
The wire WILL throttle the speed and ultimately any negotiation that does take place. Simply by its construction and lack of compliance with the new spec.
Watch what happens when you put a cat3 patch link in place on a GbE link.
So, even with your pathetic quotes all you do is prove that your previous remarks were bullshit, just as I said they were. You cannot even comprehend what you read. I made no such claim whatsoever, FUCKHEAD!
Again, you retarded piece of shit... I never said a goddamned thing about the wire negotiating anything. If you want to discuss something, stop accusing people falsely, asswipe. Otherwise you will be touted as the characterless bastard you are.
The way the wire performs, however WILL clamp a negotiated link to a lower rate if it does not comply with the new requisites.
Placing the wrong pairs in place in the connector DOES cause failure modes. If you swap the brown pair and the blue pair, it will not operate the same way. And no, I do not refer to crossed wiring. I am talking about pin for pin wiring using the wrong pairs in the sequence of pin 1 though 8. It DOES matter.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.