The CsI, like most detectors, needs to be calibrated and monitored. It should be calibrated at least once per running period, but probably should be done twice. An initial calibration should be done at startup. Another one can be done towards the end of the run to check for changes in the calibration over the running period. So far, we haven't seen many changes during the 1994 run, but they could become more significant as the detector gets older.
The calibration procedure has several parts: taking the halo data, analyzing it and putting the information into the Map. The information on taking the data is in the trigger section. Analyzing the data is broken into two parts, event selection and fitting. Putting the data into the map is covered in the next section. Finally, some documentation on previous calibrations is provided.
The trigger setup is described as was used for the calibrations of the 1995 run. Some of the description is dependent on the halo counters only. Another part is dependent on the interface with the full trigger. If this part is changed by the trigger experts, the interface will need to be changed.
A halo trigger diagram is provided in the CsI log book and page 190. Some additional description is given on surrounding pages. The six halo signals are sent into a 4413 discriminator. These signals go to both halves of a 4508 MLU. One half of the MLU simply passes the signals through to be sent to latches and scalers. The other half does the logic to determine if a halo trigger should be issued. The halo logic is:
Before using the halo trigger, the MLU logic must be loaded properly. This should be possible through the Trigger program, but it is unclear if this has been tested. The process which was used for the 1995 run was to run the program load_halo_4508. This program is in SCCS in the CsI subsystem. If the MLU slot changes this code should be modified appropriately.
The main trigger should have all level 2 triggers turned off. There is a coincidence unit in Rack , Crate , Slot , Section 4 which needs to be changed. It should require a coincidence of 4 for normal running and 3 for halo running. This needs to be turned back to 4 when done. That shuts off the halo triggers.
There are a few other things to check with the halo trigger. The timing of the trigger needs to be checked. The gate should be the same as for the normal triggers. The gate should arrive at about the peak of the CsI signal. Probably about 2.2 microseconds after trigger. The voltages on the halo counters need to be set. They probably won't plateau, but they should be set so that the signals are high enough above noise, but don't send out hugh (5 V) signals. Check previous settings for a reference point. The beam should be tuned properly. The tune which seemed to work is a normal beam tune with the collimators closed fairly tight and the MPS magnet turned off. A low rate through the CsI gives the cleanest halo events.
The halo data should be monitorred online (with csidisplay) to determine the quality of the trigger and adjustments made if necessary.
After the data is taken, it needs to be processed and histogrammed in preparation for fitting. This will probably be an interative process. There is a script called NDCalibrate which can be used to guide the actions. A minor problem with this script will be noted further on. If NDCalibrate is not used the programs can be run individually. All of the code for this should be availible in the CsI subsystem of SCCS.
The first thing which is needed is to load the data to disk. It will need to be read several times probably and this will make it much easier. A good selection criteria for data written out is between 8 and 17 hits in the CsI. This should throw out junk events which won't be used in the calibration. The programs CsIWrite or NDWrite may be useful for this.
After the data is availible it must be histogrammed. This can be done with NDHist or NDMon. A histogram is produced for every crystal. Cuts can be made to select the best data for these histograms. The order of the cuts appear on the command line is the order in which they are applied to the data. Not all cuts will have the same effect if the order is changed. This is currently a failing of NDCalibrate. It applies the cuts in numerical order rather than the order entered. This may not always be desired and NDHist should then be run directly from the command line.
Here is a list of possible cuts to be applied in with a ranking of importance. This ranking may or may not apply to every case.
The object of these cuts is to get good minimum ionizing peaks in all channels with as little noise near pedestal as possible. The less noise the easier the fit. It may be best to get a very clean set of data with tight cuts to do most of the channels and another set with looser cuts for low statistics channels.
Once the proper histograms are produced, they need to be fitted. This is done to determine the peak position of the minimum ionizing particles. These peak positions are then turned into calibration constants using the pedestal value as energy = 0 and a linear calibration to place the peak at the proper energy. More information of the fits is availible in the document on analysis of the CsI.
The program which does the fitting is called CsICalib. The output gives information on the fits. An HBOOK file is created with the fit results for viewing. Another file is created with the calibration constants. If all fits are satisfactory the fitting is complete. However, if some fits are not good or impossible another set of histograms may be needed to fit those channels. The calibration constant file can be editted by hand to combine constants found from different sets of histograms. One file of calibration constants for all CsI crystals should be created.
Once the calibration constants are determined they must be loaded into the Map. This is done with CsIFillmap. It takes a file from CsICalib and loads it into the map under a selected run number. After the constants are loaded in properly, the original data file should be rerun with energy calibration to make certain the minimum ionizing peaks come out in the correct spot. If all of them check out, the calibration procedure is complete.
Two documents are availible which describe some results of previous calibrations.
Original author: Todd Adams 5-12-95