Last modified: 7-OCT-1999
TPC pedestals are evaluated individually for each of 512 time buckets per channel. Time bucket 512 shows excessive variability and will probably not be used. On some channels, time buckets 1-3 also show excessive variation (but this is not too serious, since the gating grid already causes significant noise in the early time bins.
In general, the TPC pedestals should fall within 50-255; pedestals greater than 255 are illegal, since single bytes are used for on-line pedestals. Illegal pedestal values are seen for shorted pads.
The first pedestal event of each run may have to be ignored, since pedestals drift upwards by several ADC counts if the system has been idle for several minutes; this problem is understood and may be fixed now.
Pedestal means are calculated on-line to a precision of 1/8 th ADC count. These should be checked against off-line values, and values should be verified to fall into the valid range.
Pedestal rms is calculated on-line to a precision of 1/8 th ADC count. These should be checked against off-line values.
Pedestal stability is calculated on-line with poor precision, since old pedestals are stored as integer values. Stability was studied off-line for 20 FEE cards without temperature control, but only a rough check has been done for the current system.
The pedestal temperature coefficient is not accurately known; bench tests by Fred Bieser and Chinh Vu found a value that is 1/3 that estimated (by R. Bossingham) from on-sector data taken during summer/fall 1996.
For a moderate humidity, the pedestals seem to show only a weak humidity dependence. The value is not well known. However, Howard Wieman found that a condensing atmosphere has a large effect (i.e., blowing gently on an FEE card).
To date, pedestals have been calculated as a simple average over values taken without anode-wire gain (except that, off-line, the first pedestal event is usually rejected). More elaborate algorithms (e.g., truncated mean) are supported on-line, but have not been studied to date.
Not studied; FEE card-to-card variations may hide small, systematic effects on the sector (due to variations in socket resistance, for example).
Systematic variation of pedestal means with time bucket has been suppressed to a reasonable level by now. Variations with FEE card channel have not been studied, but this is probably unimportant.