I succeeded in getting the history of the toroid string encoders from the online production database, after numerous false starts. For the aficionado, I have the data as an ntuple suitable for PAW. Readin works with
ntuple/create 1 'encoder positions' 6 ! ! time nw sw ne se run nt/read 1 joint.listwhere the time is seconds since Jan 1970; nw, sw, ne, and se are the encoder values in an-yet uncalibrated units (assuming it is linear), and run is the run number (0 or -1 if between runs).
It is pretty easy to tell when a toroid quadrant is opened, and from this I get a list of times each was opened. Of course, this isn't as useful as run number for most analysis, but it is handier for this sort of study of the behavior of the toroids.
In a much larger file, compare the encoder values with time for each quadrant. There's also a closeup of the NorthEast.
As a back-of-the-envelope estimate, the range of the encoder values for the NorthEast is 65.68, and the toroid moves about 3 meters. IFF the system is linear, the difference in encoder values from 59.15 to 59.45 maps into a difference in distance of about 1.4 cm. That size of number isn't going to make much difference in the track/stub matching, but it will have some impact on the fiducial estimates.
In increasing resolution, look at two time regions, a little bit finer resolution, and then resolution fine enough to estimate the relaxation time of the sensors. In the left hand plots the sensor doesn't settle down to its final (albeit still drifting) value for something between 3 and 5 hours! In the right hand plot, it only takes about 1.4 hours to settle down to its final (steady) value. Thus, when we attempt to calibrate the toroids, we're going to have to pull each back an inch, measure where they are, and then go have lunch. (I'd vote for measuring at full forward, an inch back, a foot back, full back, and full forward again. That's at least two days worth of measurements, assuming the sensors aren't drifting.)
For your amusement, look at the time interval between sensor measurements. It can be as low as a second, but is usually around 10.5 minutes, or multiples of the same (or multiples of a minute).
Modified 6-June-2003 at 10:00
http://hep.physics.wisc.edu/~jnb/imu/11June2003