| |
Unpacking Data
First off, the data from
the x-ray run must be transferred to Idun, this is done
using WinSCP. Connect to Idun and transfer the files
over (typically stored in /files2/data/micromegas/Currentdate_or_month/).
Copy the files unpack and unpackall from any of the folders
into your new folder. From here, edit unpackall and put
in all of the .dat file names. Then run unpackall. This
calls unpack, which calls unpackdatanew and gemread from
the ~/gemanal/ folders. This *should* give you .out files
for all of the .dat files.
*Note: IF any of the scope channels were labelled as
anything other than an integer (ie. Pad: Sum 1) then
the unpack procedure will fail. Simply edit all of the
.dat files and change the Pad numbers to integers, the
pad numbers are listed at the top of the .dat files and
are easy to manipulate. Also keep in mind that they must
all have the same field of width so keep the correct
number of spaces after any and all integers.
Bias Curves and Residuals/Resolution
Plots (~gmd/xray_analysis/micromegas_oct_04/)
The first
method is using a bias curve analysis. The
data is analyzed one pad at a time. For each
pad, biascurveanal.f (which is the cleaned out program,
called eventanal.f in the micromegas_oct_04 folders)
is used. The data files which hold the event information
for each pad (one data file per position on the pad -
at least 8 or 9 files per pad) should be in a file called
data.txt
There are two hardcoded variables which need to be
changed when running this program. The first is
the S variable, this indicates which pad is being bombarded
with x-rays. In the array of 6 pads, if the x-rays
are pointed at the second pad, S=1. The bias curve
really only works when there is at least one pad on either
side of it. If data is only recorded from 6 pads,
only 4 can be examined. The second variable which
needs to be modified for each pad is after the event
data file is read in. A value is subtracted from
XMEASURED, this value should be the middle location of
the pad in question (the location value given on the
scope data acquisition program). This is to centre
the readings about the centre of the pad.
The program is first run with calibration set to '1'. This
runs through the first 400 events of each data file. The
position of each even is calculated using the center
of gravity for the top 3 pad readings. This calculated
position is then checked with the actual position (which
is stored in the data files and dealt/scaled properly
in the program). The differences are averaged and
a bias curve can be formed for that particular pad. The
outputs are in xmea_xcalc_resid_reso_uncor.txt. There
is one line per data file used. The outputs are
the xmeasured, xcalculated, residual and resolution.
When the program is run for the
second time, calibration should be set to '0'. The second 400 events from
each data file are now used. This time the uncorrected
txt file is used (this is where the bias curve information
is stored). A spline interpolation is done with
the on the data points (xmeasured vs. xcalculated) and
this is used to determine the actual position. Again
the outputs are printed to a txt file (*_cor.txt), as
well as an rzhist file. This corrected rzhist
file and text file have the final residuals and resolution
plots for the analysis.
To Run Bias analysis
-Must be run individually
on each pad
From folder gmd/xray_analysis/micromegas_oct_04/pad19
>>compile
eventanal.f
>>eventanal.run
>>1 (calibrate)
>>eventanal.run
>>0 (analyze)
-> Open
xmea_xcalc_resid_reso_cor.txt in a spreadsheet, columns
in file are as described above.
Previous directions/info (might have something
I left out from above)…
To create the bias curves and the residual/resolution
plots, the program eventanal.f can be used. The most
current version is located in ~gmd/xray_analysis/micromegas_oct_04/pad16/.
The program is currently set up to read 400 events, create
the bias curve information, then read the following 400
events and use that information to create the residuals/resolution
information. The info is stored in uncorrected and corrected
.txt and .rzhist files. The bias information can be found
in the uncorrected .txt file (first two columns) and
the res. info can be found in the corrected .txt file
(3, 4th column I believe). Plots of the centroid vs.
position can be found in the .rzhist files.
*Note: When reading the files in from data.txt, they
must be in descending order (pad 18.0, 17.9, 17.8,...)
for the spline to work properly.
The file is also set to ignore channels 1+5 for the
data processing, and use them only to discern bad data.
This can be changed fairly easily at the point in the
program where the events are read in. Channel 1 is overwritten,
and channel 5 is stored in array position 7. Thus with
8 channels read in, the Upper Bound is set to NSCOPE-2,
so that it only uses channels 1-6. This was done because
channels 1 and 5 were the Sum channels on the oscilloscope.
Pad Response Functions (~gmd/xray_analysis/PRF/)
This
method is still under development. Ideally
the first 400 events from all of the files will be used
to create PRF(s). The created PRF(s) are used either
in discrete data points or by fitting a gaussian to the
discrete points and using that function. The second
400 events should then each be fitted to the PRF (either
by interpolating between the data points or simply using
the determined function).
This has been done before, although I am not sure how
it was done, or in which files. Alasdair has given
some clues in Al_analysis.txt. The files should
be in the folder gmd/mmcalib/.
The problem of different events having different charges
is hoped to be found by normalizing all of the readings
by dividing them by the total charge (sum) on all the
pads.
Some PRF histograms can be seen by running gmd/xray_analysis/PRF/totalcharge.f
. The plot.kumac plots the total charge (two plots,
each for 400 events) and the 6 found prfs. This
program also examines prf differences between events
with a large energy and low energy events. diff.kumac
plots 12 prfs, two for each pad. The first 400
events are used to determine the average total energy,
and the second 400 events are split into larger or smaller. The
histograms are then normalized to area 1, and plotted.
The most recent prf fitting program can be found in
/gmd/xray_analysis/PRF/prfuse.f . There have been
many different versions of this program, however this
is the most recent one. Old data will be placed
in /PRF/old/. The main differences were how the
peak of each event was found. a) minimum value,
b) value averaged around the peak c) re-binned and then
minimum value.
prfuse.f uses prfdata.txt as the pad response data
(the curve that the data is fit to). This is currently
created manually using Excel (or open office spreadsheet),
and the data found in prfanal.f.
ALL From the gmd/xray_analysis/PRF ---->
To create the prf
>>compile prfanal.f
>>prfanal.run
>>47 (data files)
>>800 (events)
-This creates prf.txt use this file and Excel, shift
the columns down by 40,32,24,16,8,0 respectively, this
puts all the PRFs ontop of each other. Average
the rows, into a column of 87 rows. copy and paste
this column into a text file, with the number 87 at the
top, call it prfdata.txt.
To Run the Analysis:
>>compile prfuse.f
>>prfuse.run
This will perform the analysis using the prfdata. To
see the data
>>paw++
paw++>>h/file 1 prf.rzhist
paw++>>h/pl 900
paw++>>h/pl 901
Previous PRF info
To create the pad response
functions, one may use ~gmd/xray_analysis/PRF/prfanal.f
. This is a simplified version of eventanal.f, which
simply reads in the events from all of the listed files,
and records the average max at each position. Thus for
each of the 6 pads we used, and for each of the 47 positions
(250 microns apart) it created one point. They were then
outputted to prf.txt in columns by padnumber. This file
can be quickly imported into Excel and used to create
the pad response functions. To place all of the prfs
on top of each other, simply create an X-axis offset
each pad by 2mm. This file was made slightly more user
friendly by asking the user to input most of the variables.
It is still however assumed that there are 2 sum channels,
on chan 1 + 5. To make any other large changes one must
alter the code itself.
That is all for data analysis, I hope it helps!
Last modified December 17th, 2004 by Dan Burke
Back
|
|