Talk:10BASE2

WikiProject Computing / Networking (Rated Start-class, Low-importance)
This article is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
Start  This article has been rated as Start-Class on the project's quality scale.
Low  This article has been rated as Low-importance on the project's importance scale.

debugging 10base2 networks

The "expensive" equipment mentioned is probably a fluke network analyzer or similar. This device can give you the distance to a fault in the wire, which can greatly help. However, you don't need that, and shorts are not difficult to find really. There are two easy methods:

1. play Pass the Terminator: Grab a terminator, and replace one end of the cable at the T with the terminator, and see if that half of the network now works. Do this until you find the fault.
2. Multimeter test At the located fault: Set your multimeter to volts and check the voltage at the open end of the cable. If it is zero, there are no live computers on that wire, or you have a short or open. (If there is voltage, STOP!) Then switch to ohms and test the cable end again. It should show about 50 ohms. If it reads much higher, you have an unterminated line (open). If it reads zero ohms, you have a short. Note that if you measure at the T where the computer connects rather than where a cable connects, it should read near 25 ohms (two terminators in parallel, 1/(1/50+1/50) = 25) NOTE!! This method is a bit DANGEROUS -- if you have voltage on the wire and use your multimeter in ohm mode, you could damage the meter (or more likely, blow its fuse). This is why you MAKE SURE the line is 0 volts before measuring ohms.

Despite the ease of this, I'm glad my 10base2 days are long over. It's so much easier to plug in the CAT5 cable and look for the blinky lights. The biggest problem with 10base2 cables is that computers are in a chain, and one kicked out cable takes out your whole network. --ssd 05:04, 30 Jul 2004 (UTC)

Another thing i heared from a collage lecturer was that students would half-unplug thinnet cables so they looked connected from a distance but didn't actaully make a connection. Plugwash 16:50, 30 September 2005 (UTC)

Anyone who ever maintained such a network for years will probably loathe them as much as I do. Our home (and company!) network (at it's 10base2 peak in late 2003) consisted of 2 iMacs via switches, one Router, one Notebook and 5 PCs all on one chain of BNC cables. Though the cables were *mostly* shielded from direct harm (cats chewing through network cables isn't a problem you usually consider when putting up a network) eventually (after about 5 years) they started to act quirky. Interestingly, the problems never occured in the patched up bits cats had gnawed on, but always at apparently random spots all over the network. The "pass the terminator" game had a tendency to be completely unreliable and the network would phase in and out with no interaction whatsoever, sometimes curiously spliting the network in half (which was technically impossible except for the PCs sharing the same switch.

As some PCs were already connected to the same switch via Cat5 (which was then connected to the rest of the network as part of the 10base2 chain), we eventually decided to get rid of the whole mess and replaced it with what must have been more than 100m of Cat5 cables. We also waved the company which set our network up good bye for good. It hasn't been down once ever since and if any cable would fail now, it would be both easy to replace (the BNC cables were sunk into the walls (some places apparently even OUTSIDE the house) and all but easy to replace -- we had to leave some good length of cable in place, just cutting off the ends because we couldn't just tear down the entire wall to get rid of the cables) AND easy to track down. Also, impact would be minimal, as the switches effectively create sub-networks.

It's not the best solution, but even though it's a bit of a hackjob, it's so much better than the Voodoo we had to deal with before. — Ashmodai (talk · contribs) 00:12, 1 November 2006 (UTC)

10base2 and 10base5 compatibility

Were 10base2 and 10base5 compatibile with each other? I've taken a quick look at 802.3 and other than some specs being tighter for 10base5 and the use of better quality coax (which between them allow bigger networks) I can't see any real difference. Plugwash (talk) 02:24, 8 March 2009 (UTC)

You are correct, they use the same electrical signaling (or if not the same, close enough as not to matter). Although the specs didn't suggest it, I did hear of cases where people took the terminator off of the end of a 10base5 run, replaced it with an N-to-BNC adapter, and just continued the run with RG58... eventually ending in a terminator, of course. And they worked fine. Jeh (talk) 23:59, 15 April 2009 (UTC)
Yes you can do that. I had a net running that way for a little while, while we were waiting for router parts. It is a little harder to figure out how close you are to the limits, though, so should only be done when you are not so close to the length limits. The most important term is the cable resistance, which limits collision detection. Also, not to have taps too close together, as they generate a larger reflection. Also, the early 10base2 transceivers were actually 10base5 transceivers with N to BNC adapters at each end. Gah4 (talk) 03:54, 3 October 2016 (UTC)

rounding

I made a change which I thought made it more readable, but was reverted. The original, and now reverted text, says: and the 2 supposedly refers to the maximum segment length of 200 meters, though in practical use it can only run up to 185 meters. (The IEEE rounded 185 up to 200 to come up with the name 10BASE2, for consistency with the general standard). The maximum segment length in the standard is 185m. It will work on longer runs with better cable, and shorter with lower quality cable, but the number is 185m. They could have named it 10BASE1.85, but decided to round it to 10BASE2. (I prefer lower case, as in 10base2, the IEEE seems to disagree, but that is a different question.) The 2 comes from rounding 1.85 up, the standard never says that it is 200m, or that in practical use it can only run 185m. The current text seems a complicated way to say that it is rounded, while implying something that the standard does not say. Can we do better? Gah4 (talk) 03:49, 3 October 2016 (UTC)

Perhaps a simple "and the 2 refers to the rounded up maximum segment length of 185 m"? You're right that the text is inacurate as it says 10BASE2 is supposed to support 200 m but doesn't in practical use. --Zac67 (talk) 09:33, 3 October 2016 (UTC)

Network design section, termination, reflections ,and all that

It is possible to have a 10BASE2 segment that is so short (say, half a meter between just two stations) that reflections wouldn't be a problem even without terminators. Nevertheless terminators are required. Collision detection on coaxial Ethernet works by putting your own signal on the wire (through a current source) and then checking the resulting signal voltage. The signal voltage you should be seeing depends on the fact that the cable presents a 25 ohm load. If two or more stations are transmitting, then two current sources are sourcing, the voltage will be too high, and a collision is detected. So if you leave the terminators off "because I just have two stations that are half a meter apart, and that's not long enough to worry about reflections" you'll still get "collisions" all over the place even though the reflections are close enough in time to the original signal so as to not matter from a simple distortion standpoint. So explaining the problem purely in terms of signal reflections is inadequate. Jeh (talk) 20:44, 24 October 2016 (UTC)

I see 10BASE5 doesn't explain this either. Jeh (talk) 21:00, 24 October 2016 (UTC)
Well, what actually happens is that the reflection adds, so the voltage is twice as high as it would be without reflections. For short SCSI cables, you can get away with terminating on one end, but not for coaxial ethernet. I suspect that for a short enough 10base2 cable, you could get away with two terminators at one end, and none at the other, so there would still be reflections, but also the appropriate 25 ohm load. The wavelength is 30m, so you might get away with 0.5m. I once knew someone with a three way (tee connector with branches two three different rooms) 10base2 network, installed by a supposed professional installer. It worked often enough, but sometimes was just too slow. Over terminated, you get no collision detect, but signals still might get through. Gah4 (talk) 16:09, 14 November 2016 (UTC)
I had a three terminator network once, on a customer site several hundred miles away. The building was T-shaped, which explained some of it. It worked quite, but not always, well. It was very difficult to diagnose this problem remotely! It was before cheap cable testers for one thing. Only after insisting that they checked the terminators yet again and having the exasperated answer, "yes, all three of them!" did I realise what was going on.
A further complication was their use of a proprietary "no break" connector which applied an equivalent loading to the cable of about the same as three conventional taps. Andy Dingley (talk) 17:33, 14 November 2016 (UTC)

Relative costs

I seem to recall that an adapter card with 10BASE2 capability could cost significantly more than a 10BASE-T only unit, meaning the saving from equipping 10baseT adapters was probably enough to offset the cost of the hub. 212.159.121.228 (talk) 13:28, 14 November 2016 (UTC)

There is a lot of supply/demand and economy of scale going on, and prices change over time. For some years, I was buying a lot of used networking equipment for school network projects. Early 10baseT hubs were expensive, but the prices dropped when 100baseTX started to catch on. The circuitry for a 10base2 card is similar to a 10baseT card, and so is the cost to build one. More usual were AUI/10base2 and AUI/10baseT, and the combo (three way) cards were somewhat more. I do remember with my first (486 days) laptop, buying a combo (10base2 and 10baseT) adapter, as I had 10base2 at home, and that it cost more than the 10baseT only. As 10baseT got more popular, the price for the rare new 10base2 cards went up, but then the price for used ones went down. Then the 10baseT hubs with one 10base2 port allowed one to mix them for a convenient transition. Gah4 (talk) 15:35, 14 November 2016 (UTC)

although the underlying logic may still be present.

although the underlying logic may still be present If you mean for x86 machines, I don't ever remember built-in 10base2 ports. By the time they got around to putting ethernet on board, it was already 10baseT and even 100baseTX. Some workstations did, though. NeXTstation, which I still have, and some Sun machines. I now have a Sun3/e with its associated SCSI/ethernet board, and the latter has AUI and 10base2 ports. But many Sun systems had only AUI, or AUI/10baseT ports, in later years. Now some NIC may have had the PC wiring, but the circuits not installed. It is fairly common to build a PC board with the features of the high-end model, and not install all the components when building the low model. For 10base2, you need a DC to DC converter to isolate the transceiver. That may be the bigger cost of a 10base2 port. Gah4 (talk) 12:09, 15 January 2017 (UTC)

It's unclear what "underlying logic" is referring to. On the logic side, 10BASE2 and 10BASE-T are identical (as seen from the MAC). --Zac67 (talk) 13:38, 15 January 2017 (UTC)
I agree. I never saw a PC mobo with a built-in 10base2 port. I do have a workstation (a VAXstation) which like Gah4's Sun3/e has both AUI and 10base2. Back in the early days of PC networking I did buy quite a few ISA NICs (mostly 3com's) that had both coax and 10baseT. it would make sense that they'd sell a variant without the coax connector and leave off the other associated expensive parts, like the transformer, but use the same board - no reason not to. otoh a board with every component needed for 10base2 except the connector would make far less sense IMO (if you're going to leave off the connector, why not other stuff?). In any case the text in question was speculative and plausibly questionable, so without a reference I can't see keeping it. Jeh (talk) 15:53, 15 January 2017 (UTC)
I used loads of them with 10BASE2 connectors - but they were all small format mobos for embedded systems (mid-late 90s), not intended for desktops. Some didn't even have graphics, at a time when integrated graphics were usual.
I never saw a PC mobo with AUI on it, this was more common for workstations and by the time that PC mobos had integrated networking, AUI had gone. Andy Dingley (talk) 17:31, 15 January 2017 (UTC)