Originally published at: San Francisco metro says goodbye to 5.25" floppies - Boing Boing
…
3 Likes
Replacing a fragile “loop cable system used for sending data” with less bandwidth than a dial-up modem is “more pressing than replacing the floppies,”
Waiting on a “Ksssseeeeyyy errrrr beeeep ong dee ong waaahhh urrrrrr” sequence to understand what the hell could’ve been slower than a dial-up modem. (“loop cable system” isn’t getting a hit which doesn’t involve “wakeboards”)
7 Likes
How about switching over to Sneaker Net, using thumb drive couriers, riding the trains from station to station, to move data around the system? 
18 Likes
Ancient technology that just works hangs on in transport industries for a long time. I’m reminded of the scenes in The Taking of Pelham 123 (1974) where one of the NYC subway sub-control stations is still using a candlestick telephone on an accordion arm mount to communicate with the central tower. I don’t know if those sequences were filmed on location, but either way it’s a nice touch.
12 Likes
Thank you!
6 Likes
I do not know how is the deal with Hitachi, but slow bandwidth per-se is not a problem. Resilience and data integrity are real problems in those type of systems. If the cable is fragile, OK, may be, but you can always make a better cable. That said, one way or the other, seems to a modernization is needed.
4 Likes
sort of
at the abandoned Court Street station in Brooklyn. Closed to the public in 1946, it became a filming location
that may explain the stick telephone.
matthau about the location (link jumps to wikipedia entry - filming locations);
“There are bacteria down there that haven’t been discovered yet. And bugs. Big ugly bugs from the planet Uranus. They all settled in the New York subway tunnels. I saw one bug mug a guy. I wasn’t down there a long time—but long enough to develop the strangest cold I ever had. It stayed in my nose for five days, then went to my throat. Finally I woke up one morning with no voice at all, and they had to shut down for the day.”
4 Likes
my first guess as to “what is a loop cable system” would “a system of cables that are looping around”
I imagine they’re run between custom-made 50 year old modems that communicate in some rudimentary way any electrical engineering student could recreate from first principles, which is why it costs $212m to replace them. But there’s a good chance that the real answer will be something like “oh nothing so fancy, it’s just mechanical relays and 300 miles of asbestos-sheathed lamp wire” and that’s why it costs $212m to replace them.
14 Likes
The SFMTA plans for Hitachi to replace the loop cables by 2028 with a comms system that uses Wi-Fi and cellular signals for tracking trains, assuming the SFMTA’s board of supervisors approves it.
and that’s why it costs $212m to replace them.
7 Likes
It’s probably an RS-485 communication loop. IBM was fond of these in the 1970s through the 1990s. And RS-485 was used in a lot of industrial applications.
RS-485 has pretty good distance performance. Depending on the quality of the wiring it can travel up to 4000 feet without a repeater or short haul modem. That’s 10 times farther than Ethernet, and a pair of simple wires was a whole lot cheaper than fancy network cable. They’d run a pair of wires from the main loop controller computer to the first node, from the first to the second, and so on, eventually closing the loop by running the last pair back to the loop controller.
Each box had an address from 1-254. Communication was done by having the controller send a token to the first box, with a chain of messages attached that were addressed to whichever boxes needed the message. As the token was passed from box to box around the loop, the destinations would remove any messages sent to them, append any new messages they needed to send, then forward the token and rest of the chain along.
It was a precursor to IBM’s Token Ring, only at much lower speeds. We ran our first loops at 2400 or 4800 baud, but we replaced them in the 80’s with loops that ran at an astounding 28800 baud. (That’s a not-so-blazing 0.0000288 gigabit speed, for those of you comparing performance to your in-home fiber or broadband, or even a 5G cell phone.)
By the turn of the millennium, we had already replaced our loops with Ethernet, and our registers with PC based computers. It was clear that the aging loop technology was increasingly expensive compared to $10 Ethernet cards stamped out by the millions.
18 Likes
Would this be the kind of data wiring that was susceptible to a surge going through a big-ass power line right next to it?
6 Likes
2 Likes
Is any electrically based signalling system immune to that? Yes fiber isn’t but the media converters might be.
1 Like
Not as much as you’d think. We found that our main noise problems came in via the power lines, caused by not having filters on the big industrial motors running the HVAC systems. (Identifying and solving that was a sleepless week!)
However, TFA says they blame some of the problems on the janky-ass wiring, so it’s certainly a possibility.
7 Likes
Freight elevator. Which was used only now and then, and with varying loads. Basically a random event. Took a while to nail that down.
8 Likes
It still is, surprisingly often. It pops up whenever the cable length has to be longer than 100 meters (too long for copper Ethernet) and has to be cheap (so fibre transceivers are out). In the large-acreage world of solar farms, this happens pretty often.
RS-232 is also weirdly still popular in and between electrical substations, and the 25 m range limitation is overcome by a copper-to-fibre serial converter at each end. For long haul data transmission, the fibre is often optical ground wires - fibre optic cores wrapped in steel cable that runs along the top of high voltage transmission pylons. Why don’t they use fibre optic Ethernet for long haul data? Because a) they’ve got a lot of legacy devices and the new devices are backward compatible, and b) the cybersecurity is hard to beat. If you hack into the serial link, you only get access to the endpoint box; you can’t use it as a “jump box” into the remote network.
7 Likes
I used RS-485 in 1988 for LED display signs along a GM assembly line. It didn’t use token ring, just collision detection. It was a boss/peon setup and the signs would only originate traffic in response to queries from the controller.
I don’t know if they ever had problems with power surges. Shielded pair might help with that.
As I recall, we got that contract after some company called Research In Motion (RIM) flubbed it.
7 Likes
We were a week from the grand opening and nothing was working. An oscilloscope only showed the more nodes we attached, the noisier the signal. One node looked sort-of OK, but at our full complement of devices it was like looking at zebras in tall grass. We thought it was interference from a nearby 50KW AM transmitter tower but none of the RF countermeasures we tried had any effect. RF chokes on every wire. We replaced every wire in the entire loop system with plenum rated shielded 2 pair (the most expensive wire we could have used.) We had the electricians rewire every device outlet to use an isolated ground. All the money, and nothing.
At precisely 1AM I was watching the scope and the extremely noisy waveform I was looking at suddenly turned into a picture perfect square wave. I excitedly radioed the IBM guy out working on the wires and asked “what did you do?” He said he had done nothing, but noted the building had gone silent as the air conditioning had just shut off. After talking to the energy management guy at 3AM, who told us it was a scheduled nightly 15 minute shutdown, we discovered the electricians had omitted the filters on the motors because they were too expensive (~30K$) and not needed.
Filters were installed the next day and we opened on time.
10 Likes
I’ll show YOU an aging floppy!
…
Wait, I just hurt my own feelings
8 Likes
