CDF Run2 Calibration Database Primer

Introduction

The best way of introducing the Calibration database is to describe a typical sequence of events concerning calibration which happen during the normal course of CDF operations. Important terms will be highlighted on the way through. We will use the context of the SiChipPed table as an example.

Before we get started, here is a list of the important terms described in this page which are bookmarked to the places where they occur.

CID
ValidSet
UsedSet

SiChipPed is a table in the calibration database which contains information about the pedestal values for the Silicon detectors. The pedestal value for a single channel (usually 1 strip) in the silicon detector tells you what the mean output voltage is when there is no beam in the machine. These are then reference values to compare to when reading actual data. The standard deviation on this pedestal is then called the "noise" in this context. As a further refinement, it is possible to get a more accurate estimate of the noise for a given channel by subtracting the noise for one channel from its neighbour's noise. This is called DNOISE (delta-noise) and is a better estimate for one single channel as opposed to noise characteristics shared by all the channels in the chip.

In the database, this information is stored in a table called SiChipPed. If you fire up sqlplus on fcdflnx3 by doing:

 
source ~cdfsoft/cdf2.cshrc setup
cdfsoft2 development 
sqlplus cdf_reader/reader@cdfofread

and execute the sql command:

 
desc SiChipPed

this gives you the description of the SiChipPed table in the database (ie the colum names and properties). The output is:

 

SQL> desc SiChipPed
 Name					   Null?    Type
 ----------------------------------------- -------- ---------------------------- 
CID					   NOT NULL NUMBER(38)
 CHIPKEY				   NOT NULL NUMBER(38)
 PED1						    RAW(128)
 PED2						    RAW(128)
 PED3						    RAW(128)
 PED4						    RAW(128)
 NOISE1 					    RAW(128)
 NOISE2 					    RAW(128)
 NOISE3 					    RAW(128)
 NOISE4 					    RAW(128)
 DNOISE 					    RAW(128)

Here is what the column names mean:

CID : This stands for Calibration ID (identity). This is a very fundamental concept in the calibration database, and is used as the "key" (unique identifier) for many tables. Each time a set of calibration constants is generated for a sub-detector a CID is assigned.
CHIPKEY : This identifies a particular readout chip in the silicon detector system (L00, SVX, or ISL).
PED1 : There are 128 channels for each readout chip. This column of the table stores the pedestal values for each of these channels concatenated together as a single field 128 bytes long. PED2,PED3,PED4 are exactly the same thing except for different readout states of the chip (not used as of sep03).
NOISE1 : These are noise levels (standard deviation on the pedestal) for each channel. NOISE2,3,4 correspond to different readout states as with PED.
DNOISE : This is the differential noise for each channel (a better estimate of the noise for a single channel).

This is what a silicon-guru does to generate a new set of calibration-constants to be put into the SiChipPed table:

Uses Run-Control to make the detector take two different runs consisting of 5000 events each with no beam in the machine.
Executes a script which takes the raw data from these runs and generate the values of PED, NOISE and DNOISE for each channel on each chip.
Runs a-2-db to convert the ascii output of this procedure into a database readable version ready to be inserted into the calibration database.
Inserts the set of rows (appx 7000) into the SiChipPed (one particular CID assigned).

There are other calibration constants which need to be calculated for the silicon detector in addition to pedestals. Once these have been calculated a ValidSet can be created for the silicon detector. The concept of a ValidSet is very often encountered in the calibration database. A ValidSet consists of all the calibration data necessary to completely calibrate a sub-detector. Several sub-detectors can be grouped into a single valid-set.

Once a ValidSet has been created, a UsedSet can be created. A UsedSet is the combination of a ValidSet and the run number (or more usually range of run numbers) for which this calibration is useful for. There are many combinations of Calibrations and the run ranges they are useful for going all the way back to the dawn of data-taking.

Once the UsedSet has been created, the offline Calibration folks can create a Calibration Production "Pass" which consists of all the Calibration data to enable someone to run production with the most recent Calibrations. This consists of all the UsedSets for all the run-ranges there are (apart from data which is deemed to be useless).

Here is a fairly irrelavent feynmann diagram to chew out a bit of space so that the advertised bookmarks at the top of the page actually take your browser to where the words appear in the text. That is unless you have your browser in teeny tiny text mode on an enormous screen.

Matthew Martin

Last modified: Fri Sep 26 10:05:01 CDT 2003