The article about this yesterday mentioned that three disks were used to load a control program onto a MS-DOS based system, the control system maintaining safe spacing of MUNI trams inside the tunnel. This system is to be replaced by a Hitachi train control system controlling the entire network at the bargain (?) price of $200 million plus.

Replacing a floppy drive is a no-brainer. It might be a bit fiddly in a very old MS-DOS based system because of the lack of USB support but this is San Francisco, right?** There should be no shortage of local talent to figure this out. MS-DOS might be a bit old but its only acting as a loader and at least we know that its bug-free. I suppose sales people couldn't conceive of a Raspberry Pi Zero control system (equal to half a dozen or more vintage PCs), they have to get something that's over designed -- and likely overpriced -- for the job.

(**An interim solution that would be really easy to implement would be to replace the 5.25" drive with a 3.5" one. It will just plug in. One of these disks will carry the three disks' worth of data.)

Or a Gotek. "Looks like" a floppy disk drive to the controller, of any geometry. Holds as many images as you can wave a stuff on a USB stick.

My understanding is that the $200 million also includes maintenance of the control system for ten years, which makes it less outrageous.

Cobbling things together with Raspberry Pis and other hobbyist gear is not something you do with systems that are life safety critical. Anything used for rail signaling has to be failsafe*, which is one reason why these systems tend to be technologically a bit behind. (Just switching from relay logic using failsafe "vital signal" relays to PLCs took a long time.)

* "Failsafe" in this case being literal -- it's not that it can't fail, it's that it has to predictably fail in the safest condition, which usually means stopping the trains. A trivial example is the old semaphore signals, which were weighted such that a broken operating rod would cause the signal to fall into the "danger" position.

Industrially 'hardened' computers rarely are anything special. Stuxnet targeting a specific PLC only had to go for Windows and X86...

The clever stuff is in the architecture rather than the individual devices.

And I've lost count of the number of pi pico cored industrial devices out there.

That's true, but there's more to it than buying a Pi off the shelf and slapping it in there.

"Industrially 'hardened' computers rarely are anything special."

Agree. A friend of a friend has a raspberry pi running flawlessly since 2015.

The youngsters in San Francisco need to get off my lawn. Why, it was just a few years ago that the mighty United States Air Farce.. ah, that is, 'Air Force' was finally forced to retire 8-inch floppies for no particularly good reason other than it was getting hard to source spares. Bah, humbug. https://www.extremetech.com/defense/300543-after-decades-the-air-force-retires-8-inch-floppies-from-missile-control-systems Several electric utilities in the Caribbean, especially including the Jamaica Public Service Company, Ltd., have finally put to rest their old British-built oil-fired mechanical analog controlled steam turbo-generators, some as newly built as 1951, merely because the companies which built them have been out of business for two to four decades and the only place where you could get parts was JPSCo's workshop, and the last of the old workshop men there who knew the arcane art of keeping 1950s British steam tech alive are about to retire. 1998? That stuff is still new! It just needs to be buffed up a bit.

No respect for age and tradition. None at all. I despair for the future.

(Why, yes, I do still have late 1990s computer equipment; among other items, a beige PowerMac G3 and a Compaq Presario. The G3 has a 3.5" floppy drive that still works, as of earlier this year and a Zip drive which doesn't work and hasn't for years; the Compaq has both a 3.5 and a 5.25, but I'm not sure if they work. Respect the elderly!)

Don't worry, the new system is using the latest Japanese technology, the 3" 1/2 hard diskette.

Unfortunately, no fax was available for communication between systems.

5.25" disks.... in 1998???? I was using 3.5" disks in 1985.

Oh tiny little disks ... my 8 inch drives still work fine. If you think that's a rubbish comment then try buying a MITS ALTAIR 8" drive on EBay.

Given the scope of the project and government contract rules, I would be willing to bet the specs were written no later than 1988.

"Modern" sounds great. Sure, let's get rid of the slow and old induction loops and replace them with cellular networks. That's great until you have your first cellular outage that ties up all the trains. It gets even worse when the cell company discontinues support for the cell standard.

In most US cities just getting a large crowd of people together in one place is enough to crash the cell network. Stadiums here often have WiFi because 50,000 people in one place is more than cell phones can cope with.

The cell phones can cope with 50 000 people at the same place, unless it is in a sardine can.

The cellular network, on the other hand...

Scene: a US high school, opened on 6 Jan 1996, freshly built with all new networks and shiny computers in the main lab running Windows 95, plus a handful of Macs. Yet one computer in the main office was still running only DOS [1], and had to have 24/7 uptime... because it ran the fancy 16 X 80 incandescent Daktronics sign out front, the first for any school in the district.

Said sign turned lots of 'leccy into lots of heat, plus sufficient light to be distracting to the commuters headed for the nearest major metro (almost blinding at night). It wasn't the best Big D had to offer, but on the district's budget it was the best they could do. And the master software [2] ran in DOS -- I personally ran it my senior year (12th grade). I could manually run sequences or load a full month's worth (or more) into the schedule, including on/bright/dim/off commands. So while the sign was off, the DOS box still had to run all night to wake it up again in the morning and command the appropriate sequence(s) for the next day.

It mostly ran on a (small) hard drive, but it did have a 3-1/2" drive (not 5-1/4), and loaded other software just to see what the machine could handle... without messing up the current operations. [3]

Good times...

[1. Admittedly not actually that out of date, yet. If it's still there now, that would be a different story, but I assume they've upgrading the sign and the computer since then, probably full-color, full-video LED.]

[2. Actual sign ops were in a separate box screwed to the wall, connected by serial. This ran only the current sequence from its own memory and needed the PC to give it the next instruction/sequence.]

[3. Because of #2, the PC *was* allowed to reboot -- and the control software called up from autoexec.bat -- but you had to manually reload the Scheduler. This was definitely NOT preferred and usually only happened if power was lost, in which case the sign rebooted blank, which was (marginally) better than some default pattern or re-running something out-of-date.]

Of all places! Now, say some place like Gary, Indiana, and all is forgiven.

everything that is wrong with Municipal IT orgs, concentrated into one hyper-dysfunctional smelly lump of procurement abuse.

Remember when they got hit with Ransomware a few years back, in an easily preventable scenario that completely could be laid at the feet of their Storage lead and his unwillingness to deploy any solution that did not rhyme with Yell, Smell, and Hell?

Same team...

To be fair, how many times do you think the various layers of local bureaucracy were told by City Hall (or whoever) that there was no money to upgrade the IT, because the trains were just about running, so it must be working OK?

Historically railways have tended to keep rolling stock for four, five, six or more decades, and the same parsimony by the purse-string holders often extends to infrastructure and IT.

This approach will of course not work in future, because they won't be permitted to run trains whose controls involve unsupported software, and the complex control and sensor systems of modern trains will be subject to far quicker parts obsolescence than older stuff you could fix with a hammer. Case in point, Britain's railways scrapped most of their class 321 units at about the same time they scrapped their final class 483 units. The difference being that the class 321 was about 42 years old, the class 483 was 83 years old.

A/C because no one wants to be labelled a closet train spotter.

< "It's unlikely the floppy disks won't just vanish overnight. ... In fact, the whole upgrade could even drag on for the better part of a decade​."

How about some proofreading? This looks like something I'd post on social media after composing and then editing it on my phone while half awake.

"London, England" and "Vancouver, Canada". Who knew? Explaining where these cities are was vital to understanding this story. It would have been so different if it was the Vancouver in Russia that was using floppy disks in its metro.

If it's working. Leave it. Just cos it's old doesn't mean it's shit.

Sounds like there's several other problems that need attention first.

I guess you didn't read the article?

Go back. Try again