Project supported by JCHP-FFE, EC FP7 and Swedish Research Council Development of Pellet Tracking Systems for PANDA and WASA Hans Calén, Kjell Fransson, Carl-Johan Fridén, Björn Gålnander, Elin Hellbeck, Marek Jacewicz and Pascal Scheffels (Erasmus, Bochum) IFA & TSL, Uppsala University Motivation, goals and challenges Uppsala pellet test station, UPTS Pellet Diagnostics with one Camera Camera performance tests Illumination Two Synchronized Cameras Pellet velocity (spread) Time resolution and Camera efficiency Tracking system design and simulations Conclusion and Outlook Results presented in Milestone report* for FP7 HadronPhysics2 WP19 (Futurejet), December 2009. (CM GSI 8/3 2010) *) Design ideas for pellet tracking systems for PANDA and WASA Project supported by JCHP-FFE, EC FP7 and Swedish Research Council

Goal: To know interaction point with a few 100 mm accuracy in 3d(xyz) Method: Track pellets! and synchronize with time of interaction Basis: Pellets F = < 30 mm, v ≈ 70 m/s, Dv/v = few % Pellet stream F<10mm, intensity 10-20k plts/s Possible pellet tracking detector positions 1.5-2 m from int. point Need: Pellet detection accuracy ≤ 50mm (xyz) and < 2ms in time at pellet generator (and maybe detectors also at pellet dump) Idea: - Use laser and line-scan (linear CCD) camera for pellet detection. z1 x1 z2 x2 zip xip Pellet generator Challenges: Detection efficiency Position and Time resolution Handle pellet velocity spread Data processing Calibration, alignment, control Interaction point

Uppsala Pellet Test Station, UPTS Two Cameras Aviiva M2 CL linear CCD - 512 pixels 14 x 14 mm - max line rate ~ 90 kHz? - lens: f=50 mm, work dist. = 240 mm pixel size  40 x 40 mm tot. coverage: +/- 8 mm, DOF: few mm Laser beam with a line profile - type: Lasiris MFL 35 mW, SNF 50 mW - profile: 50 mm(y) x 15 (3) mm(x) at work distance = 185 mm matched to camera with f=50 mm lens Ongoing activities April-June 2010: - Pellet diagnostics with single camera - Understand camera performance - Improve light yield from pellets - Develop synchronized r/o for 2 cameras - Time and position correlation studies - Prepare tracking chamber with two measurement levels Laser LS-camera

UPTS control & monitoring CCD camera monitors Droplet chamber Vacuum injection capillary (VIC) exit At LS-cameras LS-camera ctrl & monitoring

UPTS control Frequency generator and amplifier for the transducer Control of valves and pumps Vacuum display Temperature control Control of hydrogen and carrier gas

Pellet beam diagnostics with one LS camera 0.5 mm Beam profile Time between pellets Pellet “width” Study of pellet beam fluctuations on different time scales 0.1 s bins “vibrations” >0.5 s bins slower movement Used e.g. in optimization of pressure and temperature regulation. Position Fit values 0.5 s 0.1 s ~1 min

Camera tests with thin wire illuminated by LED powered by pulse generator cameras LED wire Used for check of camera performance for different cycle and integration times and different LED pulse lengths, amplitudes and frequencies. Synchronization tests with two cameras

CCD camera cycle specification: - Maximum line rate 98 kHz, found ~90 kHz … Integration time > 5 ms??, can be shorter … - r/o time ~9 ms - Dead time between integration and r/o ~1 ms - Pxl pedestal widths and signal resolution (~10 for 12bit r/o) - Signal stability Linearity vs LED pulse length (ok) and amplitude (?) Splitting of signals between lines and between pxls (x-talk) - Interference effects in camera cycle Improper operation for cycles shorter than ~14 ms (>72 kHz) - “Small” interference effects if integration and r/o overlaps

Camera tests with LED Some pixel signal amplitudes for different cases Max signal peak seen if LED pulse length different from integration time. No peak if they are equal (lower right plot). Spectra are well reproduced by simulations (red) for 72 kHz line rate. Lines between signals ….. LED at 9.2 kHz MC

Pellet illumination Efficient tracking requires that the signals from (all) individual pellets should be clearly separable from background ! Pellets with v ~70 m/s  1 ms light pulse in 50 mm laser beam … difficult to get big enough signals for fully efficient detection. How to get a stronger signal? - Slower (or larger) pellets Maybe not desirable … - Longer light pulses. Decrease camera magnification (use 25 mm lens) and increase laser line width Worsens resolution… - Laser with higher intensity Didn’t find so far… - Optimize laser optics: 5º -> 1º fan angle Too narrow line profile ? - Improve illumination geometry. More lasers ? 135° 45° Spectra very sloping …. Laser light speckle effect ? Laser Cameras 135° 45° Signal amplitude

UPTS November 2009 Two-cameras setup UPTS November 2009 Two-cameras setup. Synchronized measurement of x and z Cameras A&B laser beam exit 135 degrees illumination geometry

Two camera readout and synchronization Camera links 120 Mbytes/s each Framegrabbers: MatrixVision TitanCL PCI-board NIM – Modules to generate synchronization signals for the line sweep of the cameras DAQ Computer Pulse generators: for LEDs (for control and monitoring of camera synchronization) for Camera cycle

Two camera synchronization with pellets 2d profile of the pellet beam Line scans triggered from external pulse generator Frames of 5000 lines each, analyzed in real time Z coordinate X coordinate Lines between signals in camera A & B Time Same pellet seen by both cameras? Frames out of synch

Control and monitoring of camera synchronization. Time reference. LS camera “picture” ( ≈ 500 lines ↔ 10 ms) Use synchronized LEDs providing short light signals with low frequency (≈100 Hz)… to illuminate white screens placed at the vacuum window and at the edge of the field of view of the cameras. (Means also very out of focus). Pellets Synch lines

UPTS March 2010 Misc correlations between infos from the cameras … pellet signal heights measured pellet position position – signal height Laser Cameras A B Expected correlation between measured pellet positions (pixel signals from camera A. vs. camera B.).

Pellet velocity spread WASA pellet generator Pellet velocity spread The pellets have very little spread in velocity and direction in the droplet chamber. A significant spread in velocity and direction arise in connection with the injection into vacuum through the VIC (Vacuum Injection Capillary). At WASA the time distribution of pellets at the interaction region, 2.4 m below the VIC, has turned out to be stochastic and this causes large variations in effective target thickness on a time scale from 10 microseconds to a few milliseconds. At UPTS, already 0.3 m below the VIC, the pellet time distribution looks stochastic. Simulations show that this is expected for a velocity spread of a few percent. For accurate pellet tracking the relative velocities must be measured with a precision better than 10-4 and for high efficiency the time and position measurement values must be correctly assigned to the individual pellets. Simulation studies show how the measurement positions should be arranged for highest efficiency. VIC Fo = 1.4mm

Time between pellets --- pellet velocity spread (1) Gauss upper and Exp lower curve 1. 2. 4. 3. Time between pellets --- pellet velocity spread (1) MC simulation Case: 1) Short dist. / low vel. spread Gauss 2) Longer dist. / higher vel. spread Gauss (trunkated) 3) Longer dist. / higher vel. spread Gauss (tail) / Exp ? 4) Long dist. / high vel. spread Gauss (tail) / Exp ! sv/v=0.3% sv/v=1% 4. sv/v=2.5% 3. 2. 1. sv/v=5% Time between pellets UPTS measurement 300 mm below VIC exit. f≈40kHz, p(H2) ≈ 400mbar, p(DC) ≈ 25mbar (total loss = camera deadtime + illumination ineff. + lost pellets) Curve = Exp w. slope  17 k/s Time between pellets

Time between pellets --- pellet velocity spread (2) At the ion beam region the distribution is stochastic even for a relative velocity spread of 1 ‰ !!! Gauss upper and Exp lower curve Time between pellets sv/v=0.1% sv/v=0.2% sv/v=0.5% 2. 4. 3. MC simulation 4. 3. 2. MC shows that for a pellet rate of 10 k/s and velocity of 60 m/s there is a pellet in the region of a F=4mm ion beam only 50% of the time and that there is exactly one pellet 36% of the time … (in good agreement with expectations from the Poisson distribution).

Time between pellets --- pellets in the ion beam (1) For 10k/s & sv /v = 1% : 50% of the time 70 % of the time 0.7 pellets 1.4 pellets How often are there pellets in the ion beam ? How often are there exactly one pellet candidate in “(hadronic) events” ? What is the average number of pellets in the ion beam ? What is the average number of pellet candidates in “(hadronic) events” ?

Time between pellets --- pellets in ion beam (2) Number of pellets in ion beam vs time (during 5 ms) for different pellet generation frequencies (5-20 k/s): MC results for pellet v=60 m/s, sv /v=1% and ion beam F=4mm. = average over time = average in “events” 0.34 1.41 1.80 0.73 1.04 1.16 1.37 1.54 f = 5 k/s f = 20 k/s f = 15 k/s f = 10 k/s

Time between pellets --- pellets in ion beam (3) Number of pellets in ion beam vs time (during 5 ms) for different pellet generation frequencies (10 & 20 k/s) and sv /v (0.1 & 1%): MC results for pellet v=60 m/s and ion beam F=4mm. = average over time = average in “events” 0.71 1.41 1.80 0.73 1.43 1.23 1.37 1.68 f = 10 k/s sv /v=0.1% f = 20 k/s sv /v=1% f = 20 k/s sv /v=0.1% f = 10 k/s sv /v=1%

Time between pellets --- pellets in ion beam (4) Number of pellets in “ion beam” ( 60-70 ms interval) vs time (during 5 ms) for a pellet rate around 15 k/s: MC result with pellet rate=15k/s, pellet velocity v=60 m/s, sv /v=1% and ion beam F=4mm. UPTS measurement result with a pellet rate ≈ 17 k/s … .. seen by the LS-cameras. 1.04 1.54 MC f = 15 k/s UPTS data Time [20ms]

Synchronized LS cameras at two levels UPTS April/May 2010 Droplet chamber VIC exit Laser Camera Up Synchronized LS cameras at two levels Distance ≈ 30cm Laser Camera Lo

Pellet velocity estimate at UPTS May 2010 Steering of the pellet stream close to the VIC wall introduced a disturbance that broadened the angular distribution. In this distribution one can make cuts …. Synchronized LS cameras at two levels Laser Camera Up At VIC exit Distance ≈ 30 cm Laser Camera Lo 30 cm below Measured pellet position (z) … which is helpful in studies of pellet signal correlations … (in particular when the total pellet rate is high)

… that can be used to get a hook on pellet velocity distributions … Pellet velocity estimate at UPTS May 2010 … that can be used to get a hook on pellet velocity distributions … Pellet generation conditions fdroplet ≈ 50kHz p(H2) ≈ 400mbar, p(droplet.ch.) ≈ 25mbar droplet velocity 25 m/s pellet diameter 25-30 micron (guess ) DT (lower-upper) for all combinations of pellet time signals MC simulation Time difference Time difference A big fraction of the pellets have a velocity v ≈ 80 m/s with velocity spread sv /v ≈ 1%. 60m/s Velocity 100m/s

Time resolution, efficiency & measurement dead time With two specially designed cameras (having 3 ms period time), measuring the same coordinate at the same y-level and synchronized with their cycles shifted half a period time one would have a time bin of 0.5 - 1 ms i.e. s ≈ 0.25 ms which is in the region of the final goal for PANDA. With such an arrangement one would also get rid of inefficiencies due to the camera cycle dead times. With the present camera performance of 14 ms period and 10 ms exposure time one would get s ≈ 1.5 ms which could give an interaction position vertical (y) coordinate s ≈ 1.5 mm i.e. what could be useful for WASA …. A Laser(s) Cameras B Camera A Camera B 1 cycle = 3 µs 0.5 µs 1 µs exp 1 exp 2 … strong / many enough to give full detection possibility.

System design simulations At PANDA & WASA two sections of the target pipe, one at the gene- rator and one at the dump are planned for tracking equipment. The length of the sections are 40 cm (Panda) and 27 cm (Wasa). Simulations are used to determine the optimal use and they are also needed in the development of tracking algorithms. By the simulations one should also be able to investigate the following questions: - What is the number of pellets in the interacting region at a certain time ? How good information can one get from the tracking system on this ? - What are the effectiveness in tracking and the quality of the tracking information (on event by event basis) ? - How accurate is the reconstruction of the interaction position ? The measurement accuracy and efficiency, the alignment of the system and the performance of the tracking algorithm are important factors. PTR section – Interaction region ≈ 2 meters

Pellet tracking Step-by-Step A: Determine mean velocity (v0) & spread in the pellet stream B.1: Determine “roughly” time, position, direction and velocity of a pellet candidate using v0. At the 1st levels in the upper tracking section. B.2: Extrapolate pellet track to the other measurement positions and pick up matching information. Improve velocity determination. It takes ≈100 ms for a pellet to pass all measurement points. B.3: Use all info (including geometrical constraints like point of exit from the VIC) to reconstruct the final track. Determine the time and path for passage through the interaction region. It takes ≈100 ms for a pellet to pass the interaction region. B.4: Store info for all pellets, sorted by the time of passage through the interaction region. Put a time stamp from the experiment clock used for interaction events. A common time scale is needed for matching with the experiment DAQ. C: For each interaction event, process (offline) the pellet info stored and check pellet candidates responsible for the interaction and if possible make use of the position information.

Tracking step B.1: Initial determination of velocity for the individual pellets Efficiency (%) The measurements at the 1st levels are crucial … … for catching a pellet and determine its velocity correctly. For a pellet rate of 10k/s an “efficiency” > 90% is obtained for a velocity spread of 1% and a distance of 50 mm between the measurement points. For 100 mm distance the efficiency has dropped to 85%. Pellet rate (k/s) Pellet rate (k/s)

UPTS tracking chamber in preparation Will be placed below a skimmer. For operation with 2-4 cameras + with check possibilities using the existing upper levels at the pellet generator ….. Two levels for LS-camera measurements separated by 80 mm. Masih Noor, CAI (Center for Accelerator and Instrument Development), Uppsala University UPTS bottom floor (with the old PANDA vacuum test structure)

Design idea: PANDA upper pellet tracking section Four levels for LS-camera measurements - Distance for velocity determination 60 – 260 mm. - Distance for direction determination 200 mm (…internally… one can use VIC exit also). Total height 400 mm. Space requirement radially: rmax = 500 mm. Camera Laser Level for additional monitoring (alignment) Masih Noor, CAI (Center for Accelerator and Instrument Development), Uppsala University

Multi camera readout and synchronization Adapter cards (1 per camera) Camera links 120 Mbytes/s each Optical links 2 Gbytes/s for 16 links Convert data to optical links Synchronize all cameras 16 optical receivers 1.6 Gbits/s 64 Mbyte of DDR ram 8 Mbyte of Flash rom USB, Ethernet VME-64x or LVD –bus readout Pellet identification and storage with time stamp Experiment triggers stored with timestamp Experiment trigger Timestamp to DAQ A few Mbyte/s output data rate for 16 cameras Up to 16 cameras Virtex5-FXT data processing board

Pellet target mode for tracking A pellet position accuracy in the interaction region of s(x), s(z) = 0.15 mm and s (y) ≤ 1.5 mm* should be obtained from the optical pellet tracking system. There should be a high (> 50% **) probability to have exactly one pellet in the accelerator beam region at the same time and there should be useful tracking information available for > 75% of the interaction events. This puts a lower limit on the average distance between pellets and on the acc. beam vertical diameter and an upper limit on the acceptable velocity spread. - The pellet size must be big enough to have efficient pellet detection. Here only pellets of hydrogen are considered. Since pellets of other elements have different optical properties and must have half size for the same target thickness it is not clear how well they can be detected. _____________________________________________________________________ (+) Pellet diameter ≥ 20 mm *** (+) Pellet frequency ≤ 10k plt/s Average pellet velocity ≈ 60 m/s (+) Total spread in pellet relative velocity ≤ +/- 5 % **** (+) Average distance between pellets ≥ 4 mm Effective target thickness ≤ 1.5 × 1015 at./cm2 Pellet stream diameter ≈ 3 mm (+) Accelerator beam vertical diameter ≥ 3.5 mm (s ≥ 1 mm) (+) Average number of pellets in acc. beam ≈ 0.7 _____________________________________________________________________ *) With a dedicated (specially designed) CCD chip also a s (y) around 0.15 mm can be reached. **) Design value. Could be increased by further development of pellet generation. ***) Might go to smaller pellet size if additional improvements in pellet detection will allow. ****) Present design goal.

Status and Conclusions June 2010 Camera performance basically understood …. Synchronized operation with two cameras works ! Illumination/detection conditions good … but should be improved Desired transverse (x,z) position resolution reachable with present cameras. Solution for good time ( y position) resolution and high camera efficiency exists Clear indication that pellet velocity spread sv /v ≤ 1% can be obtained  Effective pellet tracking possible !!! Preparation of a first prototype PTR system for UPTS in progress: - Tracking chamber with two levels of pellet detection - 2-3 LS-cameras with lasers - New readout system Ready for design & preparation of tracking systems with more LS-cameras. Preparing simulations for the design of a full scale system for PANDA

Plan for 2010-2011 Continuation of basic performance studies/optimizations at UPTS. - LS-camera cycle, synchronization, calibration - Illumination conditions - Alignment cameras – lasers - Analysis procedures Continuation of studies of pellet stream properties at UPTS. - Time and position distributions - Correlation studies, velocity measurement procedures - Analysis procedures Continuation of simulation studies connected to basic LS-camera performance and to pellet stream properties. Preparation of a first prototype PTR system at a section to be placed below the pellet stream collimator at UPTS for measurements of pellet target parameters dt, dx, dz - Tracking chamber with two levels of pellet detection - Camera synchronization system (with LED signal monitoring) - Development of a multi-camera readout system prototype (i-face, FPGA) Begin preparations for a simplified tracking system based on at least 4 LS-cameras (that could be tested at WASA). Start simulations for the design of a full scale system (PANDA).

Outlook 2012-2014 Plan for 2012-13: Prepare a (2dim) tracking system based on at least 4 LS-cameras for WASA. Full scale system for PANDA: - Finish full performance simulations of possible tracking system configurations and algorithms using realistic pellet stream parameters. - Finish design of the full scale system (with up to 16 LS-cameras) including mechanics, detectors*, lasers, readout, software etc. *) Try to find faster/dedicated CCD-chip/LS-camera for improved time resolution Plan for 2013-14: Prepare full scale system for PANDA: - System readout and synchronization hardware (interface, FPGA) - Readout software (including FPGA firmware) - LED signal monitoring system - Systems for camera and laser alignment (h-w, s-w) - Control and analysis software - Prepare one, of two planned, tracking sections for PANDA and test it at UPTS.

Resources needed for the pellet tracking systems 2011-2014 (The costs given below are based on more or less rough estimates). UPTS operation: present “total” Cost ≈ 22 k€/y (FP7, SRC grants) PTR UPTS: 2-3 camera system. Cost ≈ 21 k€ + design, tests etc (FFE, FP7, SRC grants/applications) Equipment: additional std diagnostics 2.5k€, laser w holder 3k€, r/o system, adapters+opt.links 5k€, FPGA card 5k€, monitoring system 5k€ PTR for WASA/PANDA prototype: 6 camera system. Cost ≈ 65k€ + design, tests etc (FFE, SRC applications) Equipment: tracking chamber 10k€, cameras w holders 30k€, lasers w holders 9€ r/o system, adapters+opt.links 6k€, FPGA card 5k€, monitoring system 5k€ PTR for PANDA 16 camera system. Cost ≈ 275k€ + design, tests etc (Equipm, FFE, FP7 applications) Equipment: tracking chambers 2x15k€, cameras w holders 20x5.6k€, laser sw holders 20x3k€, synch. LED system 10k€, alignm.system 5k€ r/o system, adapters+opt.links 11k€, FPGA card 5k€, monitoring system 12k€ Misc: system tests at upts 20k€, transport+installation 10k€ el.nics design & tests 2 my, mechanics design 0.5 my. ______________________________________________________________________________________________ Applications SRC 2011-2014. operation UPTS, PTR UPTS, PTR WASA JCHP-FFE 2011-2012. researcher PTR WASA, PTR PANDA FP7 HP3 2012-2014. operation UPTS, PTR UPTS, PTR WASA, PTR PANDA (pers., consum., travel)