[d2n-analysis-talk] Scintillator Timewalk

[Brad Sawatzky]

Hi David,

On Mon, 01 Mar 2010, David Flay wrote:

> On Mon, Mar 1, 2010 at 5:39 PM, Brad Sawatzky <brads at jlab.org> wrote:
> 
> > Could you replot with finer binning along the tdc axis?  Place a cut
> > on single tracks, T3 triggers only (no edtm, no T4, etc), and find
> > out what the units are on the TDC axis.  It would be nice to know if
> > the two peaks are ~10ns apart, or 10ps apart...  If it's ps then it
> > is either a bug or a binning artifact.
> 
> The plots I sent before already had one-track and trigger cuts.
> 
> I'm binning at 1000 bins in the TDC axis, but I don't see much of a
> change... going higher (10,000 bins) I get this error message:

It looks like you're changing the binning for both axes.  10000x10000
bins is something like 400MB per histogram -- I wouldn't be surprised
if you're exceeding some limit/sanity-check in the heap allocator.

> Error: Symbol G__exception is not defined in current scope
> scint_timewalk.C:364:
> Error: type G__exception not defined
> FILE:/home/dflay/d2n_analysis/d2n/replay/scripts/LHRS/./scint_timewalk.C
> LINE:364
> *** Interpreter error recovered ***
> 
> which doesn't make sense to me...
> 
> But, in the meantime, here are the finer binned plots -- they look
> slightly better, however I'm not absolutely certain. I changed to the
> 'colz' option to see if I can get any idea of the number of events
> populating the main band w.r.t. that fainter band at the higher TDC
> values -- it seems that the higher TDC value 'band' is much less in
> statistics, if anything.
> 
> http://www.jlab.org/~flay/scint_timewalk/

You can attach PDFs to messages to the list if you want.  JLab has lots
of storage.  Really big files will get bounced by the list software, but
a couple MB once and a while should be OK.

Some blind drag copies in the PDF suggest that the y-axis has a common
factor of x10e-6 particularly obscured by the histo title so the units
on the TDC axis must be seconds.

There's a pretty clear second peak around 0.07.  That would be about
10ns after the main peak.  Any idea what that is coming from?  It also
looks a little like there is a self-timing spike in the right-side TDC
data (which is expected), but also a second blob coming a little bit
later.  That suggests that some other paddle sees a particle and
generates the trigger, but then another paddle sees a second time-
correlated pulse.  Since the S2 paddles don't overlap like S1, and
you've imposed a single-track cut, I'm kind of puzzled by how frequent
the 'doubles' appear to be...  Make some 1D TDC histograms for the data
shown in the plots at your link (log scale, if needed).  That will show
the relative magnitude of the secondary structure(s) quite clearly.  

There is also something that looks a lot like time-walk showing up at
the low ADC bins on the left-side data.  If I'm correct about my units,
that's actually quite a bit for ADC values < 200.  If those low-energy
values correspond to data we actually care about, then we could do
significantly better on the time-walk correction there.  Perhaps it the
fact that the right side PMT always is forced to carrying the trigger
timing biases Vince's calibration script and over-corrects the left
side...

-- Brad

-- 
Brad Sawatzky, PhD <brads at jlab.org>  -<>-  Jefferson Lab / Hall C / C111
  Ph: 757-269-5947 -<>- Pager: 757-584-5947 -<>- Fax: 757-269-7848
The most exciting phrase to hear in science, the one that heralds new
  discoveries, is not "Eureka!" but "That's funny..."   -- Isaac Asimov