Date: Fri, 30 Jun 1995   00:34 -0800 (PST)
From: HFWS2@SLACVM
Subject: Monitoring
 
Calibration and Monitoring of Silicon Detectors with Binary Readout
 
The performance of the silicon tracking system is monitored in two ways:
        a) calibration of the electronics chain
        b) measurement of efficiency and noise occupancy with beams
 
a) the electronics chain is monitored with calibrations, which :
       - are taken usually during beam-off time;
       - are as exhaustive as one wishes during the time of the
         calibration of the calorimeter, muon system, straws etc,
         ranging from a full response curve to a single gain point;
       - result in a list of channels which are dead or abnormal
         in terms of gain and noise;
       - result possibly in an adjustement of the threshold voltage
         for part of the STC.
 
(Note that we have some experience with a binary system at ZEUS:
A graduate student did a calibration every few month, proving that
the system was stable)
 
b) The performance of the silicon detectors
 is monitored through determination of occupancy and efficiency,
 (this is equivalent to the more abstract signal and noise).
 
 Overall degradation is a slow proces. Thus to check it, occasional
detailed threshold scans with beams might be useful to measure the
median pulse height, as often as perhaps once a month and after
major changes occur - for example warming up the system for repairs.
Other than this the performance numbers involve the occupancy and
efficiency (the resolution is automatically ok if the efficiency is high
and doesn't require additional monitoring online).
They can be determined during data collection and require a rudimentary
tracking program.
The efficiency is taken from on-track hits, and the noise occupancy from
off-track hits. In fact, the occupancy can be determined by looking at
channel maps or module occupancy numbers INCLUDING beam related hits.
The efficency can be monitored in the same way as the level two trigger
works and can perhaps be incorporated into the trigger software as
information stored out of the trigger. The idea is that the trigger cuts on
high pt patterns. For these patterns which will require only a
fraction of the hits possible (for example three out of four) we could
keep the fraction of misses as the efficiency monitor.
This could be done per channel but it would be adequate to do it per
detector module - with the channel maps giving the channel
to channel effects.
Averaging over detector module allows more rapid accumulation of data.
Note the information does not require that the event was eventually
kept by the second level trigger, but rather that a track was present
at the first level. The track requirement eliminates noise
so that we are monitoring the real efficiency.
 
 
The binary data lends itself ideally to this task ON-LINE, because
no data correction are needed (and are even possible!).  This procedure
was used in the KEK beam tests during offline and should be transfered
to on-line for H8.
From our ZEUS experience, we learned that an efficiency plot is extremely
useful. With a trivial "tracking program", the efficiency of every
plane is determined. This was done off-line, but as we all experienced
during prop chamber days, can be implemented easily on-line.
An inefficiency plot from ZEUS shows how the operation of the detectors
can be monitored: inefficiencies were generally on the 0.1% level,
but increased to 10% in cases where the bias voltages were set too low.
If done on-line, one can get the information during a run, and
correct the operating conditions, if needed.
 
An important aspect of H8 and BSP testing should be the demonstration of
the monitoring tasks and their relative ease or complexity with which
they allow on-line diagnostics.
 
.