p-p 2006 Pedestal Fitting

The PP 2006 data (from the Indiana files), has been run and pedestal fits made

The pedestal fitting routine works by finding the maximum and then the RMS from the half hight from a plot of the Preshower Channel

This preliminary estimate is then used to constrain a Gaussian fit using ROOT

The fit is made over a range of 4sigma to the left of the peak and 2 sigma to the right, to aviod fitting into a region with physics

If the fit is too wide or fits on the first bin only, it is discarded

The fitting sequence is run over all towers the had a status bit == 1 in the BEMC Pedestal database

This is an example of a typical "good" fit

Comparison with the BEMC pedestal database

Mathew made an inspection of 25 preshowers using this method vs. the BEMC database values by hand.

It was discussed that the pedestal values in the DB were not to be trusted and in the RMS values some clear miss-matched events can be seen. However, for "Good" channels, the location of the mean between out values and the database are very close

Fitting Plots

The fitting program produced and saves a plot for each fit. By loading them into a browser it was possible to scan the quality of the fits "by eye". This allowed a preliminary "by eye" list of the distributions to be made. A majority were good fits: (Thick Black line our fit, thin line data)

Note: All plot fits can be found online here, format ped"chan id".png:

There seemed to be about around 80 channels with bad fits:

Too wide:

3073 3074 3076 3077 3078 3079 3094 3096 3100

Ones with "Double peaks" Start at 3144 and seems to finish after 3022 A nice example of a clear double peak is 3049

Wide, with incorrect slope distribution

1975, 1974, 1956, 1955, 1954

Some have a sort of spiked distribution, there are a few! so I will give a range:

1306-1336 (1331 has a spike at zero) and 601, 612, 618, 620, 821 to 828, 832, 842, 843, 847, 848

Some have their distribution with an ADC value starting at zero,

379, 371, 821 to 857 and 1093, 1092, 1081, 1079

These distridution had the ADC value of the peak was too low with noise:

343, 344

Status

• The preshower pedestal RMS and Mean values from the pp 2006 data have been genereated
• This list has some bad channels removed by the code and "by eye"
• This pedestal list is now re-generating the pedestal subtraction and fitting sequence on the pp data
• From previous inspections, bad distributions give bad slopes with this stage. Slopes that deviate more than a few sigma will be cut and the channels added to the list giving a preshower status table, this stage is also very quick (cut from a plot and save channels to text file) and I will email the results of the output as soon as the runs complete