The Display System

Have you ever wondered how the display system of the SL subway works? After having spent countless hours waiting for trains that were not on time, so have I. One thing is clear - it's a mess. The conceptual model of the system is at best messy, at worst faulty. Here's what I've found out staring for too long at those displays.

Select:   Display Mode 1  |  Display Mode 2  |  Display Modes 3,4,...  |  Congestions   :tceleS

The display system is a two line red LED array system with 26 characters on each line. The upper line displays fixed text while the lower one is usually scrolling its text at various speeds. One of the displays at the southbound platform here in Kista, for example, is very fast and scrolls at uneven speeds. Also, every time some information is updated, which might occur three times a minute, the display must restart its scrolling instead of just updating the most often counted down minute forecast. This looks quite ugly and is a design flaw in the system.

The signalling system for the Green line was installed during March and April 1999. One year later, April 2000, the display system was being being tested, not with real data but with a test string on row two. The displays often carried the text "SL provar skyltsystemet" (SL is testing the display system). On rare occasions, time estimates (mode 1 as described below) were spotted. Not real ones, though. After a very long testing time, the system finally became operational.

-----

Display Mode 1

There are two main operational modes of the display. In mode 1, there is no train at the station. The upper line displays the ETA (estimated time until arrival) for the next train, and the lower line scrolls the ETA for the 2nd and 3rd (sometimes only the 2nd) closest trains on track or scheduled. The mode 1 format is

  nn end_station    mm min#

where 'nn' is the line number, e.g. 11 for Akalla - Kungsträdgården, 'end_station' is the end station for the particular train in that direction (which is not necessary the end of the track), and 'mm' is the number of minutes ETA. (Actually, if ETA > 30 then the time in format HH:MM is displayed instead.) The # is a character that explains the source of the ETA estimate. This is a major bug in the system. If # is a dot, then the ETA is fetched from the scheduled timetable, and then "3 min." means 3-4 minutes. If # is a space, then the ETA is calculated based on sensors up the track, and now "3 min " means 2-3 minutes!!! The latter is because someone discovered that it is unwise to display "0 min " for a train with ETA 0-1 minute, but that person did not adjust the definition for timetable calculated ETAs. The result is that all running trains look one minute late even if they are on time! There is a third option, when # is an apostrophe. The meaning of this is unclear, but seems to have to do with guesswork or approximations of the timetable.

If the system loses track of the next train, it will display stars '*' to notify of its confusion. It starts with an array of three (sometimes four, occasionally five) stars shown where the minutes should be. It seems each star combination is shown for about 30 seconds, with the exception of the single star which stays for up to 45 seconds. There is a countdown through three-two-one star, during which time period the train hopefully either is detected up the track (and then the new ETA is shown) or arrives at the station (then mode 2 below is displayed). Unfortunately, often none of these two things occur. Then the confused system does one of two things. Either (i) it guesses that the train vanished from earth, promoting the second train to be the 'next' even if that one has not yet started, or (ii) it shuts down the display, leaving everyone in mystery. Case (i) is often combined with shutting down line two of the display (actually mode 1B), giving as little information as possible.

Needless to say, the display shuts down frequently, for reasons unclear to most viewers and maybe to the designers as well.

-----

Display Mode 2

When a train arrives at a station, the display enters mode 2. For some stations, this is done more than 30 sec's before the train arrives, and for others long after the train is visibly and audibly detected by those waiting for it. The format of mode 2 is

  nn  end_station mvagnar

where 'nn' is again the line number, e.g. 10 for Hjulsta - Kungsträdgården, 'end_station' is the end station for the particular train in that direction (which is not necessary the end of the track), and 'm' is this time the number of cars in the train set, usually 4, 6, or 8. Notice that for some obscure reason, there is no space between the number and the word 'vagnar'. At first, this seems like an effort to squeeze longer end station names onto the 26-character line, (remember that line one does not scroll). But instead there are two spaces between train number and end station, and besides, some end station names, like Hässelby Strand, are shorted anyway, so this seems only to be a clueless layout. Sometimes, 'kort tåg' ('short train') is shown instead of '6vagnar'. There is no hint to what the difference might indicate and besides, who defines what a short train is anyway?

Sometimes, the display starts showing the completely wrong information. A train number 11 to Akalla arriving at a station might be shown as '10  Hjulsta    6vagnar', causing those on their way to Akalla to skip the train and those not on their way there to take it instead. Big surprise if you are short on time. Sometimes the right number but wrong direction is shown, which is a bit more harmless, at least if you are an accomplished commuter.

-----

Display Modes 3, 4, ...

As an exception to the rule, the message 'INGEN PÅSTIGNING' (no admittance) is for some reason shown on line two instead with line one shut down and in caps unlike other messages. Maybe to create a surprise effect to wake sleepy commuters up.

There are actually more display modes, but those seem not to be directed to passengers of SL. For example, the message 'MSG # 913' on line one with a shut down line two is not very intelligible to the average viewer and maybe not to SL personnel either.

For the Green line, which is the oldest and longest - and thus the one with most delays and in most need of delay information - mode 1 is not in operation. The only thing the displays do is showing the end station and currently not even the number of cars in the train set when the train arrives. This is due to the now infamous signalling system, due to be replaced in its entirety in 1998 but severely delayed because of enormous problems encountered during installations. The computer system (Siemens) appears to have contained serious design flaws. The end result has been many more hours of waiting for trains to arrive.

Update note April 2000: The old signalling system was finally replaced in March and April 1999, with traffic discontinued during the Easter. The SEK 280 million Siemens signalling system is, however, still not delivering, not even a year after launch at a project cost now estimated to be well above SEK 450 million. No wonder the other bids for the signalling systems were all at or above SEK 400 million. The story continues...

-----

Congestions

Astonishingly, the display system seems not to be able to handle congestions in the traffic at all. Even if a delay is announced through the old speaker system (and you are lucky enough to be on a station where the speakers are not broken), the display system keeps guessing that a lot of trains still run on time according to the timetable. Then, suddenly, the train disappears from the display and the next one is promoted to closest ETA train, not reflecting much of reality.

If the traffic comes to a complete halt, the display never shows this. There are no messages saying the traffic is halted and no information on its duration - this is handled exclusively through the speaker system. If you are lucky, the speaker system is operational. If you are not so lucky, you are left staring at a shut down display system when you need it the most. Almost anyone can read a timetable to find out the scheduled ETA of the next train. Many commuters learn their trains' respective departure times by heart, but no-one can forecast delays or havocs.

-----

Disclaimer: This is of course not the official standpoint of either SL or DSV. In fact, it's not a standpoint at all.

-----

[Back] Back to DSV's home page.

-----

Generated by SpiderWEB. Latest update 2000-04-23. Comments to <mad@dsv.su.se>