Drift Chamber FAQ's

Q. I'm going to deal with dc calibration for Bonus experiment. The only tutorial/description of the system on how to do this that I found on JLab webpage had a note of being "terribly out of date". Could you tell me where I should look to find some current documentation on this procedure and tools (f.e. I've heard of some web interface) to do that.

A. The dc3 manual is not terribly out of date. You can access it from cvs; under docs/dc3. The command 'make' will produce a postscript document, dc3.ps, which you should read. There is also a web-site which can help you and your group to do the calibrations: http://www/~davidl/CLAS/tools/dc/index.php.
Q. I successfully went through the cooking stages of the manual you pointed to me, but encountered a problem trying to run the dc3 program itself. I found it in /work/clas/calib/dc_cal/dc3/PRO/Linux/dc3. When I try to run it I get the following message:
Incorrectly built binary which accesses errno or h_errno directly. Needs to be fixed. /work/clas/calib/dc_cal/dc3/PRO/Linux/dc3: error while loading shared libraries: libstdc++.so.3: cannot open shared object file: No such file or directory.
Could you, please, tell me wether this is the right version of the executable that I found and if it is, how to fix the problem (I tried to find libstdc++.so.3 in jlab directories, but the only thing with this name that came up was a few broken links in other users' directories).

A. Right now it looks like the /work/clas/calib/dc_cal/dc3/PRO/Linux/ directory points to a Red Hat 7 build, while if you are on an ifarm machine it should be running Red Hat Enterprise. Try running the executable /work/clas/calib/dc_cal/dc3/PRO/LinuxRHEL3/dc. and let us know what happens. This should work OK.
Q. Regarding, finding T0: in the subwindow dedicated to T0 there is only "Mac" method available, although the manual talks about 2: leading edge and iterative. By looking at the program I would guess that the method there is iterative, although in the manual doing the leading edge first is recommended.

A. The "Mac" method is the leading edge method. You first change the max and min values under the plot limits until you get a detailed view of the time distribution's leading edge, from zero counts all the way until it hits its maximum value and starts going down. You then change the max and min values of the fit limits until you have highlighted the linear part of the leading edge; from roughly 10% to 90% of the maximum number of counts. The window shows you a linear fit to this portion of the leading edge; when you hit "fit one" it will calculate and set the value of T0 (you can see T0 change in the small window to the right of the "Set Tzero" window. You can also use the slide bar to set T0 by hand. The alternative method is to change T0 by hand and then hit "Plot DOCA", and repeat until the DOCA plot is as flat as possible; i.e. when the central spike is about the same height as the rest of the distribution. The two methods should give the same answer within a couple of nanoseconds. If they don't; I would trust the leading edge method because it is simpler.
Q. In the "Fit" window I get absolutely lost, because the first thing the manual wants me to do is to select the type of fit I want and I don't see where I can do it.

A. By default you will fit using the XvsT method; the options appear after you hit "Expert Mode". The XvsT method is a simple fit to the Fitdoca vs. Corrected Time scatterplot. The "Resi" or residual method is a fit of the means of the Fitdoca residuals to the DIFFERENCE between the new function (with new parameters) and the old function (with the previous parameters). It is recommended to use the "Resi" method only after your first iteration with the "XvsT" method.
Q. By the way, when I hit quit, the program always gives me a segmentation fault upon quitting, but this is just a side note.

A. We don't know the answer here; we'll work on it.
Q. Having described my problems, I might add that I'm really confused by this experience.

A. Boy, you must be dumb! Just kidding; our manual is one of the most poorly written ones that I have ever seen.
Q. I want to copy parameters from the calibrated run into other runs. Having looked into the database system, I got the impression, that, besides DC_DOCA, I would also need to copy DC_TDLY, DC_TIMEDIST. Is that right and also what other systems would I need to copy if any?

A. The data-base systems used by the trk and dc programs are doca, geom, status, tdly and trans. DC_SWAP and DC_TIMEDIST are not used by recsis or a1c. You have only changed doca and geom, so I think these are the only constants which need to be moved.
Q. Some dc3 problems I noticed: 1) tmax "find all" doesn't work; A. it does now 2) on "Quit" one more often than not gets a segmantation fault; A. intermittent problem 3) very rarely: when initializing, dc3 suddenly stops and quits. A. intermittent problem
Q. Additional suggestions: 1) If the graphics window had a save as postscript option (like the summary window), it would be nice. A. good suggestion; we'll work on it. 2) I have no clue what "write to database" option in caldb window is about. Since it's dealing (supposedly) with changing run index, I wasn't really willing to experiment with it. The manual recommends using "write to map" for writing to the database, which I successfully used, but I don't think it discusses the "write to database" option. A. There is no "map" anymore; we'll change the notation. 3) We alreardy talked about this, but just to complete the picture: making cuts sticky through the whole calibration procedure would sometimes help. A. For others; if you change the definition of "Tmax" from 97% to 98% you might want to keep that your "sticky" default for the whole run period. We may not want individual calibrators to do this; but we'll talk about it.
Q. We are getting close to pass0 cooking and I was asked about parameters needed to be monitored for drift chambers. I would like total widths for different sectors/superlayers and their averages over superlayers to be monitored (please, let me know if you disagree). My problem is I'm not sure which system/subsystem/item/parameter it would be, my guess is: system DC_DOCA then subsystems xvst_params, xvst_par_sec2 and so on till sector 6, and items will be the superlayer numbers. Problem: there is a bunch of parameters in there. Could you, please, tell me which ones would have to do with "total width"?

A. You are referring to calibration constants (syst/subsyst/item); i.e. the constants used by the track-fitting program. We don't keep the results of fitting in the calibration data base. Each run group has handled this in a different way. I believe there are various scripts which do the double Gaussian fit to the residual distribution, but I don't know where they are located.
Q. I have a question. What are subsystems RMS, Sigma, timewalk, tmax in the system DC_DOCA? I would expect all parameters to be in xvst_param# for each sector.

A. RMS is the rms of the FITDOCA residual distribution. It is NOT USED. Likewise, Sigma is NOT USED. The timewalk parameters are used to describe the beta-dependent changes to the drift time. It is in general NOT CHANGED by dc3. The parameter, tmax, is the the max. drift time as determined by dc3 for each superlayer and sector (36 total) for the 3rd layer in each superlayer. It takes the individual sector,superlayer value of TMAX and extrapolates it to a value appropriate for each individual layer in the superlayer; typically TMAX changes by about 1% from one layer to the next because the cell sizes are getting larger.
Q. And, by the way, do you know of a place where I can look up what is what in the offline database?

A. http://clasweb.jlab.org/cgi-bin/caldb/cq.pl