Reconstruct signal amplitude
Find out how you want to reconstruct the signal amplitude
(and time) from the ADC spectrum. This is done in TPCcalcAmplitude.f95.
There are several routines to do this, you might have
to modify one or write a new one.
ADC spectra can be filled in histograms with e.g. plotPulseEvent
which calls plotPulseZoom or plotPulseFull. Again, you
might have to modify these routines. They are in the
module TPCcalcAmplitude.
Calibration
The calibration was designed for the original exponential
pulses. The determination of the mean pedestal is still
usefull. The code is in the modules TPCcalcAmplitude
and TPCpaw. Use schrbped.f to
write the information from the histogram to file, see
TPCpedestal.txt.
Write DenseData
At this point it makes sense to write out dense data.
In TPCinput.txt set LDData=.TRUE.
You might want to do this with a low threshold minSignal.
Timing
To determine the information needed to calculate Z from
the drifttime
> driftVelocity , timebin(DD=0), ns/timebin,
vDrift (cm/ns)
400.0 5.0 2.3E-3
you need a simple routine like MYanlyzer.f95_t0. From the distribution of 951.time0
you can figure out what the timebin for driftdistance=0 and the driftvelocity
(in cm/ns) is.
Determine Pad Response Function
Set up a robust track fit. Either gaussian profile (in
TPCinput.txt set LPRF=.FALSE.) or PRF. In both cases
make a guess for the width and keep it constant (in TPCinput.txt
have a single number for width**2)
Run MYanalyzer.f95_prf over
2-3 runs (the ntuple can get BIG).
In PAW:
nt/loop 952 nt952.f
exec prf#fit
Start with fitting 4 parameters: amplitude, gam, del, a (b=0 should be OK). Fix
a to its mean, re-run with 3 fit parameters to determine del(Z). Finaly run with
2 fit-parameters to determine gam**2(Z).
E.g. you put the determined value for a in qrat.f instead of a=par(4) and set the error=0 in prf.kumac.
Put the values in TPCinput.txt:
the width (gam) and del can depend on Z; a and b are constant
determine error of amplitudes if using ROWCHIPRF:
exec amplerr
histogram 10101 shows the mean deviation as fkt of amplitude
(should be small)
histogram 10102 shows the RMS, fit this distribution and use parameters for
> AERRpar , amplitude error for chisq fit
1.99 0.0342 0.000105
run TPCanalyzer again, iterate if necesary.
Bias?
With all correct parameters run the trackfit MYanalyzer.f95_reso (you don't need the dwout
part). In PAW:
nt/loop 951 nt951.f
change FUNCTION DX (bottom) if you ran without dwout (and don't forget to change
it back later !!!)
histograms 10000+irow are the bias.
smoothe and write out the smoothed information with
exec nt951#spline irow
and use wrtbias.f (stand alone) to write the body of
the bias file, copy the body into TPCbias.txt.
Resolution
Finaly run the trackfit MYanalyzer.f95_reso. In PAW:
nt/loop 951 nt951.f
Watch out how FUNCTION DX and AMPLCOR (bottom of file)
are defined!
Histogram 10018 gives the (uncorrected) amplitude as fct. of Z; fit this distribution
to get parameters for AMPLCOR.
exec nt951#fit (the offset is for different Z ranges)
exec nt951#fitcd (fit sigma0, CD as function of Z)
Write paper
Sorry no kumac for that.
-- Kirsten
Back |