Transcript Cluster counting report

Cluster counting report
SuperB L.N.F. meeting
April 6th 2011
Marcello Piccolo
Marcello Piccolo SuperB Meeting LNF
1
The menu
• Few words on expt’l setup
• First look ad the data
– 17 and 30 mm tubes
• Pulse shapes and cluster counting algorithms
– Cluster counting expt’l limitations
•
•
•
•
Statistical accuracies
Analyzing power
Simulation comparisons
Conclusions
Marcello Piccolo SuperB Meeting LNF
2
Expt’l apparatus
• Used a couple of continuous cathode square section devices
30/17 mm. side.
– Better shielding against envt’l noise
– Higher gain at given H.V.
– Easier to handle
• In order to do a quick ( and hopefully) clean job :
– Used a Sr90 source
– Trigger with a scintillation counter ( 4 cm thick)
– Overall efficiency 70% /22%
• Regrets
– Environmental noise quite bad ( rewrapping with copper)
Marcello Piccolo SuperB Meeting LNF
3
A first look at gas mixes
• Started looking at VERY light mixtures
– First attempt with 85% He 15% Methane
• After a couple of days struggle, we gave up
• H.V. gain variation about 5% /V
– Second attempt with 70% He 30% Methane
•
•
•
•
This mix already tested with proto 0
Quite fast
Reasonably stable
Operationally worked with high gas flow ( 1 volume
change in few minutes)
Marcello Piccolo SuperB Meeting LNF
4
Gain vs. HV
(left ) Pulse height vs. H.V. Variation 1.9%/V
(right) Pulse height spectrum with
Sr source @ 2125 V
Marcello Piccolo SuperB Meeting LNF
5
Pulse shape details
Here the challenge is clear: we want to count
spikes without being fooled by the radio stations
broadcasting around Rome.
Marcello Piccolo SuperB Meeting LNF
6
Ways to count spikes
We used a couple of different ways for counting
clusters:
– Giulietto fast discriminator does it hardware: with
a 8 mV threshold , differentiates ( with 8-10 nsec .
Dt) and counts the resulting pulses.
– One can also use software algorithms : we tried a
couple of them. The one that we used more is
based on differences between two adjacent time
bins averages (obtained with three time bins).
Marcello Piccolo SuperB Meeting LNF
7
Let’s count …
clusters (?)
30 mm. tube 90%He 10% Isob
Est. gas gain 2 105
Marcello Piccolo SuperB Meeting LNF
8
Where are the missing clusters
• Garfield (+Magbolz) tell us that for the used
mix one should be able to see ~40 clusters.
– (cfr. Jean Francois talk )
• So one is led to believe that somehow the set
up is not fully efficient in cluster detection.
• We investigated various effects , and finally
put our finger on the bitter spot.
Marcello Piccolo SuperB Meeting LNF
9
Where are the missing clusters (cont.)
The distribution shows the time
difference between two consecutive
clusters.
It is clear the overall “rearming time”
effect.
In the distribution ,both the finite
bandwidth of the F.E and the details
of the C.C. algorithm play a role.
Marcello Piccolo SuperB Meeting LNF
10
More info’s
Timing information of different clusters
does improve the spatial resolution :
Just looking at the edge of the various
cluster timing distribution on might infer
a ~50% improvement.
The last – first difference instead doesn’t
seem to carry much information .
Marcello Piccolo SuperB Meeting LNF
11
Statistical scaling
• Just to be sure that things behave as one expect, we checked
the statistical scaling of the distribution width:
Marcello Piccolo SuperB Meeting LNF
12
How about analyzing power ?
• Now given the efficiency we have to detect clusters, one may
wonder if the gain in resolution is (completely) offset by a
smaller analyzing power.
• To gain a qualitative estimate of the analyzing power, the only
possibility we were able to come up with in our set-up, was
to insert a thin copper absorber between the gas counter and
the trigger counter so that , taking a beating on rate, one
could trigger on relatively stiffer electrons.
• We used 25 mm Copper sheets folded many times
• Assuming min. ion. loss, the energy degradation is 28
KeV/layer.
Marcello Piccolo SuperB Meeting LNF
13
Here is the data
Marcello Piccolo SuperB Meeting LNF
14
Now for the smaller tube
• The time scale of our SCA is a bit marginal for
the 30 mm tube, so we built a smaller (and
longer) tube .
• Gas tightness is fine for this device
• We have just commissioned this second
device and seems to perform well.
Marcello Piccolo SuperB Meeting LNF
15
Here are few plots
Marcello Piccolo SuperB Meeting LNF
16
Some additional tuning
• Still some tuning has to be done on the 17
mm tube:
– The gas gain is different :
• Different detector capacitance
• Different high voltage setting
– The amplitude is different
• The track length is smaller
• Resolution worse
Marcello Piccolo SuperB Meeting LNF
17
A first look at simulation
• Courtesy of Giuseppe and Jean-Francois, we have
simulations of energy loss in our detectors.
• Jean-Francois will discuss details at length in the
next presentation.
• Briefly the Garfield package ( Magboltz and Heed)
was used to generate mixes, ionization and
transport clusters in the electric field generated in
our tube detector(s).
• Output currents were shaped according to a
simplified version of our electronics and analyzed
as if they were real pulses.
Marcello Piccolo SuperB Meeting LNF
18
Few plots for the 30 mm tube
simulation
Marcello Piccolo SuperB Meeting LNF
19
Conclusions
• Our set-up can measure ionization cluster: few statements can be
made from what we have seen up to now:
– A practical detector can measure a fraction of the generated ionization
cluster ( 70…maybe 80 %)
– In order to do that one needs a granular ionization AND a relatively low
drift velocity (e.g. 90%He 10% Isob.)
– Even with an efficiency lower than 100% it seems that cluster counting can
grant of the order of a factor 2 improvement in de/dx resolution.
– We are not able to quantify the timing improvement at this time; an
educated guess would call for a ~ 30-40% improvement.
– We need to tune up simulation so that we can extrapolate small
variations of operating conditions.
• New results will be available shortly, also from the new prototype
that we plan to expose to the linac test beam in the end of May.
Marcello Piccolo SuperB Meeting LNF
20