This blog relates to QA plots made regarding the pedestals and status for each strip.  The macros to produce these plots can run in a few minutes and use a DAQ file for input.  Note: by fast offline, I mean running on a DAQ file directly, generally without using a BFC.  It is also assumed that the runtime of the code is relatively small.  The pedestals and strip/APV status bits are computed via a fast offline script.  These QA plots use the results of the computed pedestals & status, not the DB values, and are intended to be generated when the pedestals are computed.  These macros are nearly, but not completely, finalized.

Note: RMS, sigma, and st. dev. are used somewhat interchangably in discussing the width of the pedestal distribution per strip.

Reading the DAQ

The usual RTS code is used to read the DAQ file, just as is done with cosmic data and for the full processing in the BFC.  There are thus three interfaces between the RTS code and the StFgtCollection in StEvent:

1. The StFgtRawMaker, to be used in a BFC, which assumes some other code has opened the DAQ file.  This maker finds access to the RTS code as it inherits from RTSbaseMaker
2. The StFgtCosmicMaker, which instanciates a daqReader, to read the DAQ file.
3. The StFgtRawDaqReader, which can replace the StFgtCosmicMaker.  The raw daq reader works on both DAQ and cosmic ray files.  The only difference in the two cases is the mapping to be used between electronic coordinates (rdo/arm/apv/channel) and physical coordinates (disc/quadrant/layer/strip).

Note: any offline QA software designed to use the Fgt classes can then be used, without any modification, in a bfc or for fast offline (i.e. running on a DAQ file w/o b)

Computing the Pedestals

Two algorithms exist, a simple method (in StFgtPedMaker) that just computes the mean and st. dev. from the sum and sum of squares of the ADC values.  A more robust method (in StFgtRobustPedMaker) makes a histogram for each strip.  Currently, the robust method is as follows, though it is still under development:

1. Make a histogram for each channel of each APV for each selected timebin.
2. Optionally smooth the histogram
3. Determine the most probable value--set this to the initial estimate of the pedestal
4. Compute the full width at half max, from which compute the initial estimate of the st. dev.: FWHM = 2 sqrt( ln( 2 ) ).
5. Restrict the histogram to +/- 3 sigma of the pedestal value
6. Set the sigma (RMS) to the RMS of the histogram

Values to Plot

In addition to plotting the pedestals and RMS (histogrammed per APV), additionally it is useful to plot the percentage of ADC values that fall within 1 RMS of the pedestal value ("near fraction").  This will assist in detecting mishapen ADC distributions which might indicate failed strips.

Strip Status

The strip status bits are currently under development, but will include cuts on the pedestal, RMS, and near fraction.  Since the status bits are still being developed, the following plots for percentage of dead strips are blank.  This will be completed soon.

The Pedestal QA Plots

The plots are included in a 29 page pdf.  The first page is a summary page, and then the following 28 show plots for each APV at a time:  (5 APVs per page) / (10 APVs x 14 discs ) = 28.

Page 1

Note: the x axis is identical for all plots on page 1, and the z-axis is identical for all 2D plots on page 1.

• Top plot: 2D histogram of pedestal values. The x-axis is an enumeration of the APVs, the y-axis is ADC bins for the pedestal value, and the z-axis is the number of strips in the given APV with pedestals falling in the given ADC bin.
• Second plot: 2D histogram of RMS values.  The y-axis is the RMS in units of ADCs
• Third plot: 2D histogram of fraction of strips near pedestal (defined above in Section "Values to Plot").  The y-axis is the fraction.
• Fourth plot: histogram of the fraction of dead strips per APV.  Currently blank, as the strip status is not yet computed.

Page 2-29

Note: the x axis is identical for all plots on pages 2-29.

• Rows indicate a given APV.  The APVs are identified by RDO, ARM, APV, as well as the "bin".  Bin referes to the enumeration on the histograms on the first page.  Thus, if the first page indicates APV 137 is questionable, one finds the corresponing plot for APV with bin 137, and can then see the corresponding RDO, ARM and APV coordinates.
• Columns are pedestal value, RMS value, and fraction of strips within +/- the RMS of the pedestal value.
• All data from the plots on pages 2-29 are included in the first three histograms on page 1.  The only information missing is the RDO, ARM, APV to APV bin mapping which is included in the titles of the histograms on page 2-29.  However, certain features are sometimes easier to see on a 1D plot rather than looking at a column on a 2D plot.

Attached Files

Attached files include

• A text file with all the pedestal information.  The columns are:
  
  • arm rdo apv channel | disc quad layer strip | fracNear | geoId timebin ped RMS
• The QA pdf
• A Root file with all the histograms included in the QA plot

The attached files were made with a DAQ file taken in Dec. 2012, specifically run 12348038.

Note: the above is subject to change, as there may be a few iterations until final covergence.  A new blog will be made with the final iteration, and link will be added in the comments for this blog.