Cross-talk study with Flat- Panel: a step forward

Transcription

1 Cross-talk study with Flat- Panel: a step forward INFN-Milano-Bicocca Syracuse University RICH-UP, 03/06/09 1

2 Summary Progress with the H9500 study; Report on the meeting with Hamamtsu at the Isola d Elba Conference; Microchannel Plate from Photonic (not reported). RICH-UP, 03/06/09 2

3 HAMAMATSU H9500 H9500 is a 16 x 16 pixels (256 in total) having about 3 x 3 mm 2 active area per pixel. Electrical connections are on the back of the PMT. 4 Samtec (QTE F T A) connectors with 80 contacts in 2 rows are present (16 GND + 64 anodes) RICH-UP, 03/06/09 3

4 Front-End (I) The readout chip and the acquisition system are from Syracuse, originally intended for BTeV. The main features of the VA64MaPMTv0r6 are: 64 channels, 0.35-CMOS from Syracuse Uni. - Ideas; 0.32 V/Mel gain with about 70 ns CR-RC shaping time; Adjustable trigger threshold level, minimum at 10 fc (62 Kel); Noise maximum at 2200 el. NIMA V 553 p NIMA V 558 p VA64MaPMTv0r6.pdf ( RICH-UP, 03/06/09 4

5 Front-End (II) X 64 Signal is analog at its start and ends digital at the DAQ. External Threshold level adj. added The only way we have to study the signal is through the setting of the threshold, that is common within each chip and to the 4 chips of the board. Glue board and PS Mezzanine board and PTA Acquisition system RICH-UP, 03/06/09 5

6 Measurement set-up for cross-talk study (I) To study the cross talk we concentrate on only a few number of pixels: the 8 around an illuminated pixel We concentrate only on 9 pixels at a time. RICH-UP, 03/06/09 6

7 Measurement set-up for cross-talk study (II) Pixel j Pixel j+1 Z i Z i C F C j C j+5 We connected one channel every 5 suppressing completely the cross talk from the small, necessary, short flat cable. Pixel j+n Connections well separated Z i C j+5k RICH-UP, 03/06/09 7

8 Measurement set-up for cross-talk study (III) Connections of the pixels to the electronics has been made with mini coaxial cable or twisted cables. In both cases well distant and separated. Flat Panel To the Electronic Coaxial connecting cables well apart RICH-UP, 03/06/09 8

9 Flat Panel (I) Here 2 examples of a pair of standard, plastic fibers optic on 2pixels, all the other darkened. RICH-UP, 03/06/09 9

10 Flat Panel (II) The fiber optics has been illuminated with a commercial blue led at 470 nm wave length. 1KΩ LEDs has been biased just around Pulser Led diodes threshold and tiny coupled to the optical fiber in order to send to the pixel, at random, a photon RICH-UP, 03/06/09 every second. 10

11 Set-up (I) The box with the flat panel has been housed inside a small Faraday cage. Faraday cage Box with flat panel Erica RICH-UP, 03/06/09 11

12 Set-up (II) The final look of the box connected to the electronic. assembled very closed to minimize stray capacitances. A black cloth covered the box inside the cage. RICH-UP, 03/06/09 12

13 Cross-talk interpretation Hamamatsu style Measurement conditions for cross talk at the factory: The level of cross-talk quoted is adequately small, but it is measured at large signals: signals are generated by many photons that hit the one pixel at the same time. Statistical interpretation: N F =incident photons; K e = electrons generated per photon at the first dynode ( a few electrons, say 4); E F =energy per photon; p=probability of cross talk per electron (p=5 %). E T =N F E F =total input energy; K et =K e N F =electrons generated per signal; E e =E F /K e =energy per electron; N cr =pk et =pk e N F >> 1 that means: E cr =pk et E e =pe T or 5% of the input energy RICH-UP, 03/06/09 13

14 Cross-talk interpretation Single Photon signal The single Photon statistic must be interpreted in a different way. Statistical interpretation: K e = electrons generated from the incoming photon at the first dynode ( a few electrons, say 4); E F =energy from the photon; p=probability of cross talk per electron (p=5 %). E e =E F /K e =energy per electron; N cr =pk et =pk e N F 0.2 << 1: since the electron can not be split we expect a cross talk signal every few single photon signals. LP Unfortunately, when this happens the energy lost in the side pixel is: E cr =E F /K e 0.25E F in case 4 electrons are generated. LP CP LP This figure is even worse, by a factor close to 4, if we consider that 5% is defined, according to Hamamatsu, the average cross talk signal on the 4 pixels around the fired one. RICH-UP, 03/06/09 14 LP

15 Simulation of cross-talk from single-photon signal (I) In the simulation and measurements we have considered that: the noise and the single photon signal response have both a normal distribution; There is a dependence of the number of electrons generated at the first dynode with the drop out voltage applied to it. HV: 900 V RICH-UP, 03/06/09 15

16 Simulation of cross-talk from single-photon signal (II) Number of Hits Central pixel Drop-out at the first Dynode= 50 V Simulation confirms that the singlephoton response is as expected Electron Numbers after 3 dynodes to single photon response 500 At 50 V of drop out at the first anode the cross talk signal extend to the % of the maximum. Number of Hits Side pixel Drop-out at the first Dynode= 50 V Number of cross-talk Electrons RICH-UP, 03/06/09 16

17 Simulation of cross-talk from single-photon signal (III) Number of Hits Central pixel Drop-out at the first Dynode= 110 V Electron Numbers after 3 dynodes to single photon response 500 At 110 V of drop out at the first dynode the cross talk signal drops to %. Number of Hits Side pixel Drop-out at the first Dynode= 110 V Number of cross-talk Electrons RICH-UP, 03/06/09 17

18 Simulation of cross-talk from single-photon signal (IV) Cross-talk (%) Single photon response vs the number of electrons generated at the first anode: the limit is 20 % Electron Numbers after 3 dynodes to single photon response The response to multi photon signal has the 5 % limit expected. Cross-talk (%) Electron Numbers after 3 dynodes to 100-photon rsignal RICH-UP, 03/06/09 18

19 Simulation and measurements correlation (I) Results from simulation has revealed that we quoted a larger cross talk level at the last meeting. The reason was due to the actual range of the electronic, limited to about 3.5 Mel, that we did not considered correctly. N. Elettroni 7,0E+06 6,0E+06 5,0E+06 4,0E+06 3,0E+06 2,0E+06 1,0E+06 0,0E Vth For instance: at 900 V of bias we measured 1.1 Mel of cross talk signal. At the previous meeting we quoted about 2.2 Mel for the maximum of the signal, giving about 50 % of cross talk level. Considering the correct range (see the method described next) we obtained about 4.5 Mel of signal with a cross talk ratio now corrected to about 30 %. RICH-UP, 03/06/09 19

20 Simulation and measurements correlation (II) We have characterized the flat panel at different bias voltage, from 750 V to 1000 V. Results seem to confirm our prediction although we prefer to post pone to show numbers since not yet fully verified: Cross talk seems to be closed to 40 % at small bias to become slightly above 20 % at the larger biases. It is interesting to consider the description of the measurement method adopted to reduce the effect of the preamplifier dynamic to the signal analysis. RICH-UP, 03/06/09 20

21 Measurement method (I) We introduced an integral method in the measurement. We take data changing the threshold of the preamplifier system at every measurement. This way we construct the function: TH V 2 TH ( V V p ) σ F(V ) = N(V)dv = N(V)dv N(V)dv N(V) e VTH 0 0 Considering that, around the peak V p : We have: 2 ( V V ) ( ) 2 p V Vp σ N(V) e 1 F(V ) = N(V)dv = N(V)dv V + σ 2 ( V ) 3 TH Vp TH TH 2 VTH 0 3σ We expect a linear behavior of the measured function around the peak. RICH-UP, 03/06/09 21

22 Measurement method (II) 1 slope This is an example result of a characterization taken at a bias of 1000 V. Although the preamplifier saturates above about 3.5 Mel we were able to predict the response to a single photon. In addition the noise is also taken into account in the final fit. Our next steps are: To apply the same method to the cross talk signal; To make the measurement set up automatic and programmable. RICH-UP, 03/06/09 22

23 Hamamamtsu meeting at Isola d Elba (I) We met: ICHIROU OHTSU, manager of the Electron Tube Division YuJI HOTTA, Application Engineering of the Electron Tube Division. The argument of discussion was technical, about single photon response, and on new possible devices targeted to a RICH detector. RICH-UP, 03/06/09 23

24 Hamamamtsu meeting at Isola d Elba (II): cross-talk According to their simulations electrons can not escape from the path between dynode to dynode. They claim that cross talk is due occasionally from one of the photo electron that jump in the lateral pixel. They never tested the cross talk at the single photon level and remained surprised from our results. RICH-UP, 03/06/09 24

25 Hamamamtsu meeting at Isola d Elba (III): cross-talk At the moment their interpretation of our results is as follow: In the H9500 a pixel is composed of many grids (6x6) surmounted by a mesh. They suppose that sometime a photoelectron can hit the peripheral mesh generating the possible cross talk signal. It is not clear why this should happen with 5 % of probability. They will report to Hamamtsu and feedback to us their interpretation. Mesh RICH-UP, 03/06/09 25

26 Hamamamtsu meeting at Isola d Elba (IV): devices for RICH In the opinion of Hamamatsu the H9500 is not the best candidate for single photonelectron study since it is not optimized at the first dynode. Their present candidate for single photon electron detector is the R7600, that is 8x8 with about 2.3 x 2.3 mm 2 area per pixel. Unfortunately the R7600 has a large border of about 2.4 mm, a large unusable area. RICH-UP, 03/06/09 26

27 Hamamamtsu meeting at Isola d Elba (V): devices for RICH Nevertheless the R7600 (or R7546, its encapsulated version) are well optimized for single photon and also for the cross talk. At the moment we are considering to study this device, because RICH-UP, 03/06/09 27

28 Hamamamtsu meeting at Isola d Elba (VI): devices for RICH But: Hamamatsu is developing a new tube the R11265 that should have characteristics very closed to those of the R7600, but with a smaller border wildness. The R11265 will be a 8 x 8 tube with a border wildness closed to about 1 mm. This will be possible thanks to a increased area of the single pixel to about 2.8 (2.9) x 2.8 (2.9) mm 2. The R11265 will be available by the end of this year (tentatively). RICH-UP, 03/06/09 28

29 Conclusions The work on the characterization of the cross talk at the single photon level for the H9500 flat panel is progressing with both simulation and measurements. As soon as available a new (parallel) study will be done on the R7600 flat panel, precursor of the R11265, optimized for single photon event and small cross talk. RICH-UP, 03/06/09 29