Tile SiPM production and test status

1
Tile SiPM production and test status

• Test of photodetectors - procedure and results
• Status of detector manufacturing
• Test of assembled tiles - procedure and results
• Future plans

HCAL main meeting, DESY, July 7, 2005

M.Danilov ITEP,Moscow
2

• Test of photodetectors
• Test bench
• 15 detectors are illuminated simultaneously with
  monitored light produced by UV LED and re-emitted
  by bundle of WLS fibers
• The test set-up includes
• CAMAC random and LED trigger generator with LED
  driver
• 16 channel power supply to feed SiPMs
• PMT for LED light monitoring
• computer read-out digital voltmeter to monitor
  SiPM voltage and current, temperature of LED,
  SiPMs and PMT during test

HCAL main meeting, DESY, July 7, 2005

M.Danilov ITEP,Moscow
3

• Test procedure
• Tune HV in order to have SiPM response
• equal to 15 pixels/MIP in 3x3 cm2 tiles.
• At chosen HV the following SiPM parameters
• are controlled
• gain
• cross talk
• noise
• noise at 1/2 MIP threshold
• width of single photoelectron peak
• response to light in the range 0.1-200 MIP
• SiPM average current and its stability
• The set-up operation is based at measurements of
  tiles with Sr90 b-source.
• First for 10 SiPMs HV was tuned to have
  response to b-particles in 3x3 cm2 tile ?15
  pixels.
• Then these SiPMs were used in test bench to
  calibrate LED light.
• The amount of LED light is monitored by PMT and
  its stability implies the accuracy of SiPM
  calibration.
• So, the special care was taken to PMT gain
  stability which is deduced from the single
  photoelectron peak position in PMT spectrum.

HCAL main meeting, DESY, July 7, 2005

M.Danilov ITEP,Moscow
4
SiPM selection Ped RMS lt 50 ADC channels Gain gt
4105 or 1 pixel gt 26 ADC ch., corresponds to 1
pC/MIP Cross talk lt 0.35 At HV adjusted Npix/MIP
15 lt 0.75 Noise frequency at zero threshold lt
2.5 MHz Noise frequency at ½ MIP threshold lt 500
Hz Single photoelectron peak width to gain ratio
lt 0.20 Mean value of SiPM current lt 2 mA RMS of
SiPM current during test lt 20 nA Number of pixels
at maximal light (200 MIP) gt 900
HCAL main meeting, DESY, July 7, 2005

M.Danilov ITEP,Moscow
5
SiPM test results new samples ( gt 1254 )
HCAL main meeting, DESY, July 7, 2005

M.Danilov ITEP,Moscow
6
Summary of SiPM test for 1172 SiPMs tested in May
June 2005
Rejection factor for this batch is big because
last 280 SiPMs showed very high noise at zero
as well as 1/2 MIP level
HCAL main meeting, DESY, July 7 2005

M.Danilov ITEP,Moscow
7
Spectra below show the difference between SiPMs
with acceptable and high noise The left SiPM has
noise frequency 1.5 MHz The right one has noise
frequency 2.8 MHz Note the difference in
photoelectron peaks separation and in the slopes
of the tail of random trigger spectrum.
HCAL main meeting, DESY, July 7, 2005

M.Danilov ITEP,Moscow
8
Tile production
Tile size 3x3 cm2
6x6 cm2 12x12 cm2 To be produced
3500 4000
1000 Mold
3500 4000 1000 Edge mated
3500 4000
1000 Groove milled 3500
3000 800 Shipped to DESY
600 600 120
HCAL main meeting, DESY, July 7, 2005

M.Danilov ITEP,Moscow
9
WLS fiber production Tile size
3x3 cm2 6x6 cm2 12x12 cm2 To be
produced 3500 4000
1000 Cut 3500
4000 1000 Installed to tiles
3500 3000 800 Shipped to
DESY 1200 1200 120
HCAL main meeting, DESY, Julyl 7, 2005

M.Danilov ITEP,Moscow
10
Tile test

• Up to16 tiles are loaded in the same time into
  box
• Computer driven carriage carries the b-source and
  the trigger counter to a tested tile. Source
  position is controlled by the computer readout
  sensor
• SiPM HV is set according to SiPM test and
  temperature measurement
• Measurement of MIP response with help of
  triggered b-particles
• LED spectrum gives information about SiPM gain
• So, one may deduce the number of pixels in MIP
  peak

HCAL main meeting, DESY, July 7, 2005

M.Danilov ITEP,Moscow
11
Light yield for tiles for cassettes 1 - 6
30x30 mm2 60x60 mm2
120x120 mm2
We had overestimated value of light yield due to
wrong determination of SiPM gain in the
measurement of the 1st cassette
HCAL main meeting, DESY, July 7, 2005

M.Danilov ITEP,Moscow
12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
2 questions after looking on the Light Yield
results from the first cassettes
1. High spread in the Light Yield inside one
cassette for each tile size (rms of order of
12-14 percent). 2. The non-stability of mean
value of the Light Yield especially for 3030 mm2
tiles, even reduction of this value.
16
Possible reasons for light yield spread in tested
tiles - scintillator brightness - spot check of
tiles after molding showed RMS 3.2 - quality
of edge coating - measurements during tiles
manufacturing give RMS 2 - fiber quality
and length - possible fiber non-uniformity
(elasticity, cladding peeling, diameter
variation) may contribute to light yield
reduction but these are difficult to estimate.
Variation in fiber length results in alternate
gap between fiber and SiPM. Monte Carlo
estimates give 10 light yield variation for 100
m SiPM-fiber gap variation. We have measured
several tens of fibers for 30x30 mm2 tiles and
found the RMS 100 m, this leads to 10
contribution in light yield spread
experimentally we see the influence of fiber
changing of about 9 for low statistic, see
picture below - possible displacement of SiPM
according to fiber axis gives 5-10 light yield
reduction per 100 m displacement at fiber-SiPM
gap 200 m - 300 m, correspondingly. Measurements
of SiPM position at ceramic support give spread
of 50 m RMS, so the contribution to light yield
spread is about 3. - we estimate the error
during HV tuning as big as 5 . This value may be
deduced from multiple measurement of the same
SiPMs. Summing all contributions one gets 12.2
for total light yield variation.This value
should be compared with 13.5 - RMS value
averaged over all cassettes data
HCAL main meeting, DESY, July 7, 2005

M.Danilov ITEP,Moscow
17
(No Transcript)
18
2. Non-stability of the tile Light Yield. The
main question - what happens with 3030 mm2
tiles. We obtain big difference in the light
yield from one cassette to other and even
decreasing of mean value. We have checked all
possibilities scintillator brightness, stability
of groove quality and detector position, edge
reflection stability,controlling set-up,
including the interference between SiPMs and
tiles set-ups. Looking at behavior of the light
yield for 30 mm (non-stable, decreasing) and 60
mm tiles (much more stable, constant) we may
focus at the question what principle difference
between them? Answer - fiber preparation.
19
We have tried to provide check of possible fiber
aging - tiles with smallest light yield have
been remeasured with new fibers, sometimes it
has helped, sometimes - not. Today - only after
looking at complete results from 6 cassettes -
we have reach an understanding that this item
should be checked much more carefully and as soon
as possible.
20
Future plans

• The tile-fiber assembly will be completed by
  beginning of the fall
• SiPM testing rate is 500 pieces per week and does
  not delay test of assembled detectors
• Starting September we hope to resume the 1-11/2
  cassettes per week production rate and maintain
  it up to end of year
• This schedule may be realized provided that SiPMs
  delivery does not delay it

HCAL main meeting, DESY, July 7, 2005

M.Danilov ITEP,Moscow