1. PSA: ADAPTIVE GRID SEARCH The Method Experimental Results Optimization aspects Roberto Venturelli (INFN Padova - IPSIA “Giorgi” Verona) SACLAY, 05-may-06

2. REFERENCE BASE MARS detector D=72mm L=90mm; 25 segments Point distance:  x,  y,  z  1.5 mm Time resolution: 10ns About 120600 points per crystal More than 5000 points per segm. At each point corresponds a set of pulses (up to 10): net charge at the point’s segment Induced in the neighbors The other 15 are compatible with 0 The REFERENCE BASE consists of more than 120000 sets of pulses

3. AGS: the principle SEARCH THE BEST  2 FOR PULSE SHAPES IN THE REFERENCE BASE Two possibilities: SEARCH 1 POINT IN THE SEGMENT Hypothesis: in case of multiple hits in segment the energetic barycentre is considered r-phi-z cycle in a coarse grid inside the real segment r-phi-z cycle with the fine grid in a small volume identified SEARCH 2 POINTS IN THE SEGMENT Hypothesis: in case of multiple hits in segment at most two points are considered e energy division cycle (r-phi-z)1 (r-phi-z)2 cycles in a coarse grid (r-phi-z)1 (r-phi-z)2 cycles in a small volume with the fine grid r  z r1r1 11 z1z1 r2r2 22 z2z2 e1e1 e2e2

4. AGS: test with simulated signals The hypothesis about the energetic barycentre have been verified ! CHOOSE 2 OR MORE HITS PER SEGMENT ( evaluation of energetic barycentre ) Results after 40kevents analyzed : 0.36 cm   : 0.10 cm 2 : 0.27 cm   : 0.19 cm 2 : 0.73 cm   : 0.39 cm 2 : 8% : 32% SEARCH TWO POINTS SEARCH ONE POINT

5. AGS: test with simulated signals After tests with simulated and measured data the SEARCH ONE POINT was preferred to the SEARCH TWO POINTS For these motivations: When two points have similar energies there’s high probability to make a mistake in position assignment (position exchange) When two points have very different energies, the one with higher energy is well identified in position but the other can be everywhere (high uncertainty) The analysis have similar performances for our experimental data with the exception of processing time

6. Possibility to start with a simplified PSA approach? Peak FWHMPhotopeak efficiency 1 MeV photons, M  = 1. In the “pack” case, only one interaction per segment is assumed. E. FARNEA G4 AGATA Demonstrator Simulations

7. Beam 56 Fe @ 240 MeV Target: 2.5mg/cm 2 208 Pb MARS detector at 135º with respect the beam direction  MARS  22º Excited beam cross section at 60º is  250 mbarn/sr 15 particle detectors at  90º with respect the MARS detector  p.d.  2.6º Event rate of  detected in coincidence with the scattered ion detected from the particle detectors  2 Hz Excited recoils have a velocity of 0.08c IN-BEAM EXPERIMENT

8. GENETIC ALGORITHMSADAPTIVE GRID SEARCH 56 Fe : 2 +  0 + emitted at  =8% E  = 847 keV SPECTRA Raw F = 1 F = 2 F = 3 16 keV Raw F = 1 F = 2 F = 3 R. Venturelli et al. INFN-LNL (Rep) 204/2005, p.220

9. IN-BEAM EXPERIMENT GENETIC ALGORITHMSADAPTIVE GRID SEARCH 5.32 4.8ALL 5.63 4.51 FWHM (keV) FOLD (segments) 5.42 4.8ALL 5.73 4.21 FWHM (keV) FOLD (segments) s/evP. TIME (5d12h) ms/evP. TIME (6’) 56 Fe : 2 +  0 + emitted at  =8% E  = 847 keV RESOLUTIONS

10. Performance Indicators FWHM of the Doppler corrected peak Event processing time Analyzed aspects : Response function of electronics Chi 2 evaluation Timing Sampling rate

11. A B C D 5 4 6 3 1 2 F A B C D F Impulse response function for a column REFERENCE BASE WITH RESPONSE FUNCTION OF ELECTRONICS FWHM of the Doppler corrected peak searching in reference bases calculated with different response function

12. AGS optimizations WEIGHTS OF THE NET CHARGE SIGNALS FOR  2 Better results have been obtained weighting in different way net charge signals with respect transient ones Similar FWHM are obtained without consider net charge signal Induced signals could be sufficient for position identification at low fold Amplified Attenuated

13. AGS optimizations TIME MATCHING t Better results have been obtained allowing time adjustment between measured signals and reference base signals Better TIMING can be achieved using base pulses with time steps of the order of few ns

14. Which sampling rate ? Signals acquired at 200 MS/s Downsampling to 100 MS/s gaining a fraction of bit in dynamic range  2 evaluation with time decimation but good triggering Less than 10% Bandwidth cutoff of Eurisys preamplifiers  15 MHz 50ns 40ns 30ns 20ns 10ns

15. OUTLOOK Test the method with new experimental data (i.e. Liverpool scans, Cologne experiment) What is needed ? A reference base with good time and position resolution (i.e.:  t = 1ns,  x = 1mm) for the AGATA symmetric capsules To limit the base dimension an Adapted Grid Base should be desirable (AGS in a AGB with signal values above electronic noise ) Experimental data preprocessed for triggering (a first time alignment), normalization, recalibration and zero suppression For future on-line PSA start to think to a standard input format for data to avoid conversions

16. FINE