General description of the Surface-T0 measurements



A 450 Hz pulse is injected into the SWAMP input, the SWAMP TDC and ADC output cables (electrical pulses) or the fibers attached to the patch panels and leading to the ORBs (optical pluses). A trigger pulse with a known time delay triggers the DAQ, a second branch of it is led into a reference channel (usually OM78). Runs are taken string-wise with the HV turned off on all OMs. Histograms showing the differences of leading edges between the test channels and reference channels, respectively, are fitted and finally added to the OMDB. A year-to-year comparison between these values can reveal unintended modifications or changes in the surface electronics of the AMANDA detector



The Used Pulse Shapes:

There was much discussion about the pulse shape, one should use for the measurements. Initially, only rectangular pulses were taken into consideration, until Ignazio pointed out that the SWAMPS won't amplify high frequencies in a correct manner. Test runs showed that indeed, there were some differences between the SWAMP generations in the shapes of the exiting signal. Rise times between 60ns and 90ns were observed.
We found then the pulse generator BNC lying around which was able to imitate OM-pulses (rise time: 20ns, fall time: 100ns). The SWAMPs have been built to correctly amplify these pulses. The problem here is small variations in the rise times (or the amplitudes) can modify the results as different SWAMPs could have different alpha-corrections.
We did not want to make whole ADC fits to the data, which we thought would unnecesserily complicate the analysis. Instead, we decided simply to check different shapes and amplitudes (8 mV and 20 mV) once and choose one best pulse shape afterwards according to the most stable result in the offline analysis. This decision will probably be taken after Christmas.

The Used Frequency:

According to Marc, the DAQ is able to deal with up to 500 Hz if the HV is turned off. Of course, a data rate as high as possible is desired, so we got 450 Hz. Because the LE peaks are very sharp (general widths about 0.8 ns), only some 2000-3000 events per channel are needed. This means that your data taking time is restricted by the speed you are able to plug and unplug the lemo connections :-))) (Advice: A week of practice at home can decrease your surface-t0 obligations by at least four hours!)

Reference Channels:

There is one default reference channel within the detector which belongs to OM 78. It is directly connected to the trigger and serves as the reference. Because of the restrictions in combination with the pulse shapes, we wanted to investigate the effect of the different SWAMP generations as well. For this reason, one adequate (dead OM, no noise, functionning TDC readout) reference channel per SWAMP generation was found additionally. One exception were the SWAMPs used for string 11-13 where no reference channel fulfilling the above requirements could be found. The signal was thus fed simultaneously into the test channel and six additional references: 
OM 39        STR 02         CH 19     yellow SWAMP, rack #2
OM 96        STR 04         CH 18     black SWAMP,  rack #5 (lower crate)
OM 186       STR 07         CH 28     black SWAMP,  rack #5 (upper crate left)
OM 197       STR 08         CH 03     black SWAMP,  rack #5 (upper crate right)
OM 227       STR 08         CH 33     black SWAMP,  rack #2
OM 543       STR 16         CH 31     black SWAMP,  rack #9
The same could be done with the ORBs. We had, however, no additional E/O converters. A possible choice could be: 
OM 399       STR 13         CH 13     ORB,  rack #6
OM 526       STR 16         CH 14     ORB,  rack #8 
OM 577       STR 17         CH 23     ORB,  rack #8
OM 603       STR 18         CH 07     ORB,  rack #9 (DOM)

The Optical Channels:

The light input for the optical channels was accomplished with the help of an E/O converter, provided by Thorsten Schmidt. Best ask Marc Hellwig when you don't know where to find it. It converts electrical signals of up to 2.5 V into optical ones of 1300 nm wavelength (which is also used for signal transmission to the ORBs). We plugged the optical fiber leading from the patch panels to the ORB directly into the converter. The delay caused by the converter, the fiber and the ORB altogether, adds up to 68 ns. Unfortunately, we did not have more coverters. For these measurements, we had therefore only one reference channel, which is the channel belonging to OM 78. It would, however, increase the accuracy if more than one reference could be used, e.g. taking the above reference channels using different converters or splitters.

The Analysis:

Up to now, a combination of steering perl script ("surf.pl") and PAW kumacs looping over fortran functions have been written which book histograms of the time differences out of the raw data files. The perl-script can easily be modified (e.g. to different directories) in the first couple of lines. The histograms will then be fitted and the mean and the width of the distrubtions written to a list. We still have to decide which of the obtained values goes into the OMDB and analysis is going on at the moment. It looks, however, as if the time differences and the widths of the distributions do not exceed 5 ns. (But better wait for the final results!).

Markus Gaug

Last modified: Sun Dec 3 16:38:30 UTC 2000