Review of the Super-Bigbite project Reply to Questions of the Committee

Transcription

1 Review of the Super-Bigbite project Reply to Questions of the Committee December 14, 2008

2 Chapter 1 Reply to question #1 What is the occupancy in the detector elements? This requires completing the Monte Carlo modeling of target chamber and beam dump line to include secondary interactions of the outgoing beam. The simulation should also address the sensitivity of the experiment to residual magnetic fields on the beam line. 1.1 Introduction The MC study of the SBS apparatus was performed within the framework of the GEANT3 code. Figure 1.1 shows a view of the GEANT model, which includes: a target, including Al side walls and end-caps; a scattering chamber with a vacuum window in front of the magnetic field clamp; the magnet, including a field clamp, an aperture with magnetic field, and the opening for the beam line; the beam line lead shielding behind the magnet; a region with magnetic field on the beam line; a detector presented by a sensitive volume at 325 cm from the target (not shown in the Figure); a diffuser in the beam dump at 28 m from the target (not shown); The field on the beam line was included in the model as a 70 cm long region with 70 G field oriented transverse to the beam and 8 kg along the beam obtained from 3-d magnetic field calculations. The MC was also done with a 10 times larger transverse field in this region. 1.2 Magnet calculation The magnet 3-d calculation was done with the MERMAID code [1]. Figure 1.2 shows the cross section of the magnet in the horizontal plane. Calculations were done with a current 1

3 of 2 ka in the coils. The cross section of the coils is reduced compared to the original BNL coils, which were used with 4 ka excitation. A view of the magnet model used in the calculation is shown in Fig The resulting field on the central trajectory is shown in Fig The total field integral is about 2.5 Tm, as was used in the design of experiments with SBS. The value of the field in the target area is below 2 Gauss, 200 cm from the magnet. Fig. 1.5 shows the field along the beam, B s, and transverse to the beam, B t. The longitudinal field reaches 8 kg in the area between the clamp and the yoke. The field map (Figs. 1.6,1.7) shows that a solenoid wound around the pipe with 1 ka/cm current density is able to de-gauss the iron pipe. As a result the pipe provides shielding against the transverse field. The remaining field has an integral of 70 G 70 cm. The effect of this field on the beam is small and the effect on the background rate in the detector was analyzed using the GEANT MC (see next section). 1.3 Parameters of the model elements Cryogenic target The target is a 40 cm long, 2 cm diameter cylinder filled with liquid hydrogen. The Al side wall is 0.10 mm thick. The end-caps are 50 µm thick. Standard practice calls for an 0.18 mm thick side wall, which leads to 12% higher background. We expect that 0.10 mm could be achieved for the proposed cell diameter Scattering Chamber The diameter and thickness of the scattering chamber are taken as they are in the existing Hall A chamber. A 30 cone around beam line is removed from the chamber front wall (see Fig. 1.1). A custom extension was added with flat Al walls at the field clamp. This snout will be attached to the scattering chamber via a flat bellows. The window needed for the spectrometer aperture (16 inches wide) is made of 0.5 mm thick Kapton film (0.25% radiation length). The beam-line vacuum pipe is welded to the same flange on the snout Beam line lead shielding The beam line vacuum pipe has a ±25 mr opening which leads to 10 5 probability for an electron hitting the front end of the pipe. Due to the significant length of the target there is a probability that electrons of the beam scattered by the same 25 mr will hit different places on the beam line vacuum pipe. As result, for background suppression, a lead cover must be added all along the beam line where it passes in front of or near the detector. This shielding consists of a 12-inch lead tube around the beam line behind the magnet and additional lead between the front clamp and the yoke. 2

4 Table 1.1: Counting rate in the First Tracker in khz/cm 2 for different configurations of the MC simulation. The statistical error of the results is 2%. Configuration Counting rate, khz/cm 2 Comments Liquid hydrogen 244 due to photons Target (T) = LH + Al cell 288 due to photons T + Scattering Chamber 306 due to photons T + Magnet (including the clamp, Pb) 362 all particles Final = T + Magnet + Diffuser(1 ) 367 all particles Beam Dump The vacuum pipe of the beam line has a 25 mr opening ending 5 m inside the beam dump tunnel. The beam passes through a 12 diameter Be window and a diffuser, which present the main source of background. We represented the diffuser as a 1 thick Al plate located 28 m from the pivot. 1.4 Detector counting rate The rate at the detector was calculated from the number of the photons with energy above 100 kev and an average detection efficiency of photons in the GEM chamber. The later was found from an independent MC to be 0.33%. Figure 1.8 shows an example of 10 events. The table 1.1 presents the resulting hit rates in the front GEM chamber for several conditions. Configuration Target has only the target and the detector. The rates of photons and electrons are presented separately. Configuration T + Scattering Chamber has the target and the scattering chamber. Configuration T + Magnet has in addition the 48D48 Magnet with the clamp and the beam line magnetic shielding (the solenoid and the iron pipe). Configuration T + Magnet + Diffuser includes the diffuser in the beam dump tunnel. A 75 µa beam current and 40 cm target was assumed. The spectrometer central angle is 14. Distance from the pivot to the magnet yoke is 65 inches. As one can see the beam dump contribution is very small. The beam line related background is totally suppressed. The observed final rate is close to the direct flux from the target. 1.5 Summary We have developed a configuration of field clamps and beam line shielding, which provides a reduction of the transverse field integral along the beam line down to IB t =0.005 Tm. The effect of such a field on the detector rate was investigated with MC simulations and has been found to be very small even at a 10 times larger value of IB t. We investigated the effect of the beam dump (diffuser) on the detector rate and found that effect is less than 2% without any shielding. The MC simulation of the full configuration has been performed and 3

5 results indicate a hit rate in the Front Tracker below 400 khz/cm 2. At such a hit rate the occupancy in the GEN Tracker is about 8%. 4

6 Clamp (iron) Magnet Snout Pb B=14 kg cell Pb Window Beam line field region Bz = 8, Bx = 0.07 Figure 1.1: Section of the GEANT model (the detector and diffuser are not shown). The boxes with numbers indicate magnetic field volumes with the field given in kg. 5

7 . Iron pipe Beam line Figure 1.2: Section of the magnet in the horizontal plane at beam line level as used in the magnetic field calculations.. 6

8 Figure 1.3: View of the upper part of the magnet in the MERMAID model. 7

9 Figure 1.4: The field on the central trajectory of the SBS magnet, B x - transverse, big component, B y - normal, small component. 8

10 Figure 1.5: The field on the beam line. B s is along and B t is transverse to the beam direction. 9

11 Figure 1.6: The field map in the area of the beam line inside the yoke. The numbers in the field give the field value in kg. 10

12 Figure 1.7: The field on the beam line with additional shielding. 11

13 Figure 1.8: 10 MC events. 12

14 Chapter 2 Reply to question #2 What is the relationship between occupancy in the GEM detector elements and track finding efficiency and false-track rejection? For the GEP5 experiment, on the Front Tracker we expect to have hits at a rate of about 500kHz/cm 2 produced randomly by soft photons. The occupancy of the cm 2 strips is 10% assuming 100 ns integration time. At such high rates tracking (with the proposed readout configuration) is possible only because the information from the electron arm imposes constraints on the elastic proton track in the SBS. Similarly, we use the hadron calorimeter information to find the proton track on the rear trackers. The track reconstruction procedure consists of: Finding the range of the elastic proton track parameters on the front chamber Reconstructing the track in the Front Tracker Track reconstruction in the Third Tracker using the coordinate information from the hadron calorimeter Tracking in the Second Tracker with the help the already reconstructed tracks in the Front and Third Trackers. 2.1 Proton track parameters on the front chamber The elastic electron will be identified in the electro-magnetic calorimeter at 3.5m from the target with a 7mm coordinate resolution. Using the position of the electron, one can determine the range of possible electron scattering angles. In the vertical plane the range is ±6 mrad, assuming the actual electron hit position is within ±3σ from the measured hit. The range in the horizontal plane is defined mainly by the 40cm length of the target, which results in ± 33 mrad (1.9 ), while the electron position uncertainty adds an additional 6 mrad. Using elastic kinematics one can find the ranges of the momentum, p p, and scattering angles, θ p of the elastic proton. Furthermore, from the SBS optics, one can determine the corresponding ranges of the track parameters at the Front Tracker. 13

15 z tar, cm θ e, deg p p, GeV/c δ θ p, deg y targ, mrad y targ, cm x targ, mrad y = 0.995y y targ, cm y = 0.05y targ y targ, mrad x = 0.330x targ + 8.2δ, cm x = 1.00x targ + 81δ, mrad Table 2.1: Correlated contributions to the ranges of the elastic proton track parameters at the first chamber: x/x and y/y coordinates/angles in the dispersive and non-dispersive planes, obtained from the corresponding target parameters, y targ, y targ, δ, using the first order optical matrix. The uncertainty in the interaction point results in a correlated effect on the different kinematic parameters as illustrated in Table 2.1 for the central kinematics. The resulting region on the front chamber, where the elastic proton is expected to be, is cm (vertical horizontal). The corresponding angular ranges are 5 27 mrad. The effect of the calorimeter coordinate resolution has to be added to the above result. The range of ± 6 mrad for the electron angle in both vertical and horizontal direction results in a ± 2 mrad range for the elastic proton angle and ± 0.7 cm for the proton hit position on the front chamber. Summing the two contributions, we expect to have the proton in a cm 2 region and within a 9 31 mrad 2 angular range. The results for this central kinematics as well as for the two extreme kinematics at the ends of the acceptance, are shown in Table 2.2. Q 2, GeV 2 p p GeV/c θ e, deg θ p, deg x, cm y, cm x, mrad y, mrad Table 2.2: Range widths of the elastic proton track parameters at the first chamber: x/x and y/y coordinates/angles in the dispersive and non-dispersive planes for different kinematics covering the SBS acceptance. For the analyses presented here we will use the widths for the central kinematics. One can see from Table 2.2 that at higher Q 2 values one can better predict the proton trajectory from the electron angle, which is explained by the corresponding change of the Jacobian. Therefore, from this point of view at high Q 2 the tracking will be easier than at low Q 2. 14

16 2.2 Tracking in the First Tracker The Front tracker consists of five chamber planes, each equipped with two sets of strips at 90 one to another, with a 400µ pitch. A pair of chamber planes, separated by 4cm is placed at the front. Another pair is positioned at the end of the tracker, 70cm from the the front pair. The last chamber plane is in the middle of the tracker. With the help of the track constraints, one starts with the hit reconstruction at the front pair. Using the reconstructed hit candidates and the angular constraints, one proceeds with the hit reconstruction at the last chamber pair. Having the hit positions at the front and at the end of the tracker, one reconstructs the possible tracks and, for each of them, checks if there is a corresponding hit in the middle chamber plane. Finally, having the trajectories in the tracker, one can reconstruct the target parameters and, using elastic correlations, further reduce the number of the pseudo tracks. The cm 2 area corresponds to strips at the front chamber, which at 10% occupancy gives 246 pseudo hits 1 per event. The real hits produce correlated signals on the x and y strips. As explained in the Front Tracker section of the CDR, using the correlation between the analog strip signals one can reduce the number of possible hits by a factor of three. This result is based on the analysis of real GEM data from the COMPASS experiment. Thus, we end up with 82 pseudo hits on the front chamber plane. Due to the small, 4cm, distance to the second chamber plane, one can significantly reduce this number. Taking into account the 9 31 mrad 2 proton angular range, for each hit at the front plane we expect to have a hit in the second plane within a crossing area between 3 horizontal and 5 vertical strips. The probability for a pseudo hit in this area is 0.15 which is reduced to 0.05 by applying strip correlations. As a result we expect 4.1 pseudo hits per event at the front pair. We repeat the same procedure with the rear pair of chamber planes, this time using the hit information from the front pair. Using again the proton angular range, for each hit at the front, one expects to find the proton in a cm 2 area at the rear chambers. With the help of the two rear chamber planes and taking into account the strip signal correlation, this results in possible random hits per event. Combining this number with 4.1 pseudo hits at the front chamber we have 0.60 possible pseudo tracks. For each of these tracks, defined with 70 µ resolution at the ends, we check if there is a corresponding hit in the middle chamber plane within the 3 3 strips crossing area. The probability for a pseudo hit there is 0.03 and we end up with pseudo tracks per event. Having the track parameters at the Front Tracker, we reconstruct the target quantities: y targ, y targ, and δ with resolutions of 0.30 mrad, 1.08 mm, and 0.54%, respectively (see the SBS Optics Section in the CDR). Using the reconstructed y targ, the electron angle in the horizontal plane can now be determined with 2 mrad resolution. Based on this, from the elastic kinematics one can predict y targ and δ with resolutions of 0.85 mrad and 0.18% respectively. Combining this result with the resolutions from the track reconstruction, one can expect the elastic proton within 6σ intervals of 5.4 mrad and 3.4%, respectively. The pseudo tracks will be uniformly distributed within the possible intervals for y targ and δ of 1 Actually, in a 39 cm 2 area we expect two of the hits to be real, but their effect in the analysis here is negligible. 15

17 31 mrad, and 6.1% respectively, as explained in the previous subsection. Thus we expect a factor of 0.1 reduction of the pseudo tracks which will bring them down to per event. The pseudo hit/track reduction factors of the procedure are summarized in Table 2.3. One can scale this result for different hit rates. Because we have used 10 strip planes, the number of the pseudo tracks is proportional to the rate 1 0. For example, if the rate was 1MHz/cm 2 instead of 500kHz/cm 2, we would have 1.8 pseudo tracks per event. In this case one would have to consider adding an additional chamber plane in the middle, that would reduce the pseudo tracks by a factor of 0.12 (see Table 2.3, bottom block). 2.3 Tracking in the Third Tracker The Third Tracker has three detector planes, a pair of planes at the rear 4 cm apart and a front plane 25cm away. Each plane is equipped with three fiber layers in the x,u,v directions, at 60 from one another, The fibers are positioned with a 1 mm pitch and will have a maximum length of 1m. For the expected rates of 65kHz/cm 2 and 50 ns integration time, we have 3.3% fiber occupancy. The hadron calorimeter determines the proton hit position with a 1.6 cm resolution in both directions. Thus, one expects the elastic proton on the rear plane to be within a 4.8 cm radius circle. This area covers 96 fibers in each direction. Assuming only two directions we expect to have 7.9 pseudo hits. For each of these we have at most two fibers covering the hit in the third direction. Requiring a coincidence with these fibers reduces the pseudo hit number to As will be explained in the next subsection, we will consider events with a 10 maximum total scattering angle in the two analyzers. Since we know the incident proton angle from the first tracker, the range for the proton angle is ±10. For each hit found in the rear plane, we expect the proton to be within a 0.7 cm radius circle on the middle plane, which is just 4 cm away. Applying the same reasoning, we will have random hits in the middle plane. Combining the hits in the two rear planes, we have pseudo tracks. Using these two planes, the tracks are reconstructed with 12 mrad resolution. Projecting again a 3σ cone on the front plane, we define an 0.3 cm radius circle there, in which we have only pseudo hits. Combined with the tracks from the two rear planes, this gives pseudo tracks per event. Assuming again as an example that the rate were 2 times higher, the expected number of pseudo tracks would be 512 times higher, or 0.18%. 2.4 Tracking in the Second Tracker The Second Tracker will use the same type of chambers as the first one, except that each 5 strips will be bridged into one, resulting in a 2 mm strip pitch. With the expected 130kHz/cm 2 rates we will have 13% occupancy. The Tracker will consist of two pairs of chamber planes, 50 cm apart. The polarimeter used in the GEP experiments in Hall C has the same geometry and analyzer thickness. Based on the experience with it and on the observed scaling of the 16

18 analyzing power with the transverse component of the scattered proton momentum, the useful range of proton scattering angle in each of the analyzers is expected to be We also know that about 60% of the events used in the polarimetry have scattered only in the first analyzer, 20% only in the second analyzer, and 20% have scatterred in both. Thus, the idea for the tracking in the Second Tracker is to check, first, for protons that have not scattered in the first or second analyzer. In this case, using the tracks already reconstructed in the Front or Third Tracker, one can predict with high precision the proton track in the Second Tracker. In 20% of the usefull events we will not find such tracks and this case will be considered separately. The multiple scattering angle in the analyzer (50 cm CH 2 ) has an r.m.s. value of 1.6 mrad. The angular resolution of the Third Tracker is 1.7 mrad. If the proton didn t scatter in the second analyzer, we would expect it within a +/- 7 mrad (0.4 ) opening cone of +/- 7 mrad, or within an 0.35 cm radius circle on the rear plane pair of the Second Tracker. Applying similar analysis as for the Front Tracker, the probability to have a pseudo hit there is 0.49%. The distance between the two pairs of chamber planes is 50 cm, the same as the thickness of the analyzer. Therefore, we will also have 0.49% pseudo hits per event in the front pair. This results in a pseudo tracks per event. If the proton didn t scatter in the front analyzer, the pseudo track probability would be even smaller because of the better angular resolution of the Front Tracker. Thus, in 80% of the usefull events we will have less than pseudo tracks per event. If we double the rates, this number will increase by a factor of 256 and will still be negligible. In 20% of the events used for polarimetry, we will have scattering in both analyzers. In this case we will consider a maximum scattering angle of 5 in each of them. According to the results from the GEP experiments, this will reduce the figure of merit by about 10%, or 2% from the total number of the events. The tracks in the First and Third Trackers define two cones with a ±5 opening angle. We expect the proton track in the Second Tracker to be within the overlapping volume of the two cones. In the worst case the axes of the cones will overlap. In this case we will have two circles on the front and rear detector pairs with a 2.2 cm radius where the proton is expected. In each of the circles we have 0.19 pseudo hits, which results in pseudo tracks per event. As a fraction of the total number of event, we will have 0.76% pseudo tracks. If the rates were two times higher, we expect 1.9 pseudo tracks per event. In this case the events with scattering in both analyzers are not useful. 2.5 Summary Based on the analysis presented here, using the rates predicted from the simulations (see the rate section in the CDR report), for the Front Tracker we expect 0.18% pseudo tracks per event. If the rates were two times higher, we would have to include an additional (sixth) detector plane in which case the pseudo tracks would be 22%. In the analysis we have always assumed ±3σ cuts, which means that the track finding efficiency is close to 100%. In case of rates higher than expected, one can also strengthen the cuts, in order to optimize the balance between the tracking efficiency and pseudo-track probability. For the Third Tracker, the tracking efficiency is close to 100% and the pseudo-track 17

19 probability is negligible even one doubles the expected rates. At 98% track reconstruction efficiency for the Second Tracker we will have 0.76% pseudo tracks. For double the rates, we will reduce the track reconstruction efficiency to 80% and will have 0.61% pseudo tracks per event. 18

20 procedure number of hits/tracks reduction factor before the procedure initial elastic correlations front chamber x,y strip correlation coincidence in second chamber second chamber x,y strip correlation final hits 4.1 in front pair coincidence in fourth chamber fourth chamber x,y strip correlation coincidence in fifth chamber fifth chamber x,y strip correlation final hits 0.14 in rear pair initial tracks from 0.60 front and rear hits coincidence in third chamber third chamber x,y strip correlation fine elastic correlations final pseudo tracks tracks per event Table 2.3: Number of pseudo hits or tracks per event in the Front Tracker and corresponding reduction factors when applying different procedures as explained in the text. 19

21 Chapter 3 Reply to question #3 What will the DAQ look like? The DAQ chapter should address in some detail the challenges that the novel electronics associated with the GEM and VLPC detectors will present. The section should contain simple block diagrams of the GEM APV25/FADC boards, as well as a discussion of how the GEM and VLPC electronics will interface with the trigger and the CODA system. (Note that this will be the first time at JLab that ADC chips will be mounted directly on the detector.) 3.1 Data Acquisition System for the SBS The main challenge for the design of the DAQ system for the GEM and scintillator-fiber trackers of the SBS is to build a system that reads out a large number of channels, distributed over the various detector systems operating at high event rates (see, for example, Appendix C, Table C.2 for parameters of the first tracker), and with a deadtime as low as possible. The proposed SBS DAQ for the GEM trackers will follow the scheme developed for the COMPASS experiment at CERN [2], and the one for the scintillator-fiber tracker will be similar to the one for the DØ Central Fiber Tracker at Fermilab [3]. The two schemes are very similar in that both use customized ASIC chips for the front-end readout electronics and intermediate VME modules for buffering of the data. To cope with the expected high data rates, the readout system has a pipelined architecture and the signals are digitized and concentrated as close to the detectors as possible, which not only simplifies the processing and transport of the detector data, but additionally lowers the readout noise. Most of the data links utilize optical fibers, which provide high data bandwidths, are insensitive to electromagnetic noise, and simplify the electrical grounding of the apparatus. Moreover, the system has a modular design in which the modules, that are connected by a tree-like hierarchical network, merge the digital data streams of the various equipments at different levels. DAQ for the GEM trackers: The current plan for processing the signals from the GEM strips (for both the first- and 20

22 second SBS trackers) is to store them in an analog pipeline. The best candidate is the APV25-S1 chip [4], currently in use for the GEM trackers in the COMPASS experiment at CERN and silicon microstrip detectors in the CMS tracker at the LHC. The chip samples 128 channels in parallel at 40 MHz and stores 192 analog samples per channel. This allows 4.8 µs to make the trigger decision which will use a coincidence between the electron arm and the proton arm for the GEP5 experiment. Note that the GEM and scintillator-fiber trackers are not part of the trigger. Once the trigger is made, the one most interesting single sample of each channel can be read, or possibly an average of three samples around the one corresponding to the trigger time (the multi-mode operation of the chip). This permits extracting accurate timing information of the signal, a feature very important to reduce pileup in high-intensity beams. All 128 channels of a single chip have to be read out, which can be run at 40 MHz as well (3.2 µs to read the chip). The read out samples are fed into a fast ADC (25 ns conversion time or better). At this point, zero-suppression becomes important to eliminate inactive channels. The sparsified data will be buffered, either within the ADC or in a subsequent FIFO. The experiment trigger will provide the timing and initiate readout of the buffered data which are cached in VME modules for delivery to the CODA system. At 10 % occupancy for the GEM strips for the first tracker, there will be about 13 active channels per ADC, and there will be about 512 ADCs for a total of 64,000 channels. Hence, the event size at 10 % occupancy is about 2 bytes/channel 13 channels/chip 512 chips = 13 kbytes. Assuming that the front-end electronics can take a data rate of 60 MBytes/s (realistic with today s technology if using gigabit networking) the maximum trigger rate would be about (60 MBytes/s) / (13 kbytes/event) 4600 events/s. These event rates are well within the capabilities of the planned 12 GeV DAQ and trigger updates for Hall A [5]. DAQ for the scintillator-fiber tracker: The current proposal for the third tracker of the SBS is to use Visible Light Photon Counters (VLPCs) to detect the light output from the scintillator-fiber trackers coupled to clear-fiber waveguides. The front-end readout electronics, buffering of the data samples in VME modules, and delivery of the data to the CODA system are very similar to the methods described above for the GEM trackers. The ASIC chip used for the readout of the VLPCs is SVXIIe which is currently in use in the Central Fiber Tracker of the DØ experiment. The incident rate for the third tracker is 100 khz/cm 2, whereas for the first tracker it is 500 khz/cm 2. Consequently, the event rate for the VLPC readout will be much reduced compared to the GEMs. Technical details of the COMPASS and DØ front-end electronics and block diagrams are provided (for reference) in sections 3.1 and 3.2, respectively. Needed expertise for the readout electronics: Albeit, the GEMs and the VLPCs require very sophisticated readout electronics, they are part of a working technology. The APV25-S1 and the SVXIIe chips are currently in use at CERN and at Fermilab. The INFN, Rome group has the technical expertise and support at their home laboratory to implement the APV25-S1 chip into the SBS GEM readout. For the VLPCs, the group at Florida International University will lead the effort to integrate the 21

23 VLPC of the Central Fiber Tracker in the DØ experiment to the planned use in the third tracker of the SBS in JLab environment COMPASS GEM readout electronics and DAQ The DAQ for the GEM trackers in the COMPASS experiment consists of several subcomponents: The front-end electronics processes and digitizes the analog detector signals. The digital data are then accumulated in concentrator modules. From these modules, the data are transferred via optical links to the DAQ computers. These links follow the S-Link standard developed at CERN and offer data bandwidths of MBytes/s. The data received via S-Link are buffered on PCI cards called ReadOut Buffers (ROBs). The computers housing the ROBs transmit the data via Gigabit Ethernet to event builder computers, which combine the data from sub-detectors that belong to the same trigger to complete event data blocks. The event data blocks are then transferred to the event recorder computers of the DAQ system. The high performance of the DAQ system is achieved by a cost-effective combination of custom made modules with standard off-the-shelf components. The custom modules are based on reconfigurable logic devices, so-called Field Programmable Gate Arrays (FPGA). The readout electronics makes use of the APV25-S1 chip, a 128-channel analog pipeline chip. Each channel comprises a low noise amplifier-shaper ASIC, a 192-cell deep analog pipeline and a deconvolution readout circuit. Output data from the chip are transmitted on a single differential current output via an analog multiplexer. The chip is fabricated in a 0.25 µm CMOS process to take advantage of the radiation tolerance, low noise and power consumption intrinsic to the technology, and high circuit density. Input signals are pre-amplified and shaped by a CR-RC filter with 50 ns peaking time. Only the fast electron charge in the GEM foils is detected on the readout strips, without the slow ion components, a unique property of the GEM concept that is essential for good pileup rejection. The amplifier outputs are sampled at a rate of 40 MHz and stored in the pipeline allowing for a maximum of 4.8 µs for external trigger selections. When a trigger is received, data in the corresponding pipeline columns are read out to a FIFO buffer. The samples are then read out separately through an analog multiplexer operating at 20 MHz, its output being converted into a differential signal by the output buffer of the chip. After 60 cm of high-density flat cable, the multiplexed data from each chip are transmitted by a differential voltage amplifier through 2 m of twisted pair flat cable to the ADC. A 10-bit ADC samples the incoming, multiplexed data stream of one chip at 20 MHz. After digitization, pedestal values are subtracted and a possible shift common to all channels in one chip, due to pickup noise in the detector, is corrected using one FPGA. The FPGA also performs zero-suppression by sending out only those channels with amplitudes higher than a programmable multiple of the respective noise figure. Data from the ADC module are then sent via a bi-directional optical fiber using the Hotlink protocol to a 9U VME concentrator module, GeSiCA (GEM and Silicon Control and Acquisition). There, the data from several APV25 chips are buffered and merged before being sent, again via optical fibers using the S-Link protocol, to the DAQ computers. GeSiCA also encodes the trigger and clock signals received from the Trigger Control System (TCS) 22

24 as well as slow control sequences for the ADC and APV25 chips, and sends them through the same Hotlink fibers to the front-end modules. The readout chain for GEM detectors in the COMPASS experiment is shown schematically in Fig. 3.1 Figure 3.1: Schematic block diagram of the readout electronics for the COMPASS GEM detectors [2]. The concentrator modules for the GEM DAQ are a central part of the readout chain. They process the digital data of the detector front-ends and provide the interface to the trigger control system which supplies the trigger signal, the reference clock, and the event labels. The clock is distributed to the connected front-ends, so that they operate synchronously and may be used as a reference for time measurements. In the same way the trigger signal is forwarded to initiate the readout in the front-end electronics. The concentrator modules buffer the data, that arrive after each trigger from the frontends, and packages them into well-defined blocks. The blocks are labeled with additional header information, which include the unique event identifier provided by the TCS, and are sent via S-Link optical fibers to the readout buffer computers. Furthermore, the concentrator modules provide a common interface to load firmware and configuration data into the various types of front-end modules. 23

25 3.1.2 DØ VLPC readout electronics and DAQ A schematic view of single channel of the DØ central fiber tracker is shown in Fig. 3.2 A Single Fiber Channel Mirror Scintillating fiber Optical connector VLPC cassette Clear Fiber Waveguide Electronics Cryostat Figure 3.2: Schematic view of a single scintillating fiber channel of the third SBS tracker. At the readout end, the scintillating fiber is mated to an 8.0 m long clear-fiber waveguide via an optical connector. A mirror at the non-readout end of the scintillating fiber reflects light back towards the clear-fiber waveguide. The waveguide fiber is in turn mated to a second 50 cm long clear fiber which is located inside a cryogenic container known as a cassette. At the bottom of the cassette, the end of the 50 cm long fiber illuminates one pixel of the VLPC photodetector. The cassette is housed in a liquid helium cryostat which maintains the VLPC at its operating temperature of 6.5 K. The VLPC cassette contains 1024 channels of VLPC readout and is divided into 8 modules of 128 channels each. The cassette is distinguished as having a cold end, that portion of the cassette which lies within the cryostat, and a warm end, the portion of the cassette which emerges from the cryostat and is at room temperature. The readout electronics boards, used as interface between the cassettes and the data acquisition system, are considered to be part of the warm end structure with regards to mechanical layout. The tracker readout electronics are contained in custom printed analog front-end boards (the AFEs) which digitize the signals and form the trigger tracks. These boards are mounted directly on the VLPC cassettes. The Port Card Boards (the Sequencer) read out the digitized values and transmit them via fast optical link to the third set of boards, the VME Readout 24

26 Buffers (the VRBs). An overall layout of the readout system is shown in Fig. 3.3 VLPC Readout Overview Data to DAQ G-Link Cable VRB VME Transition Copper Cable AFE boards Sequencer LVDS cable to L1 AFE backplane Figure 3.3: Schematic block diagram of the VLPC SVXIIe readout for the DØ experiment central fiber tracker front-end [6] The AFE is a large and complex board that performs a number of functions. It has charge-sensitive amplifiers to deal with the very small signals from the VLPCs and it is part of the slow control and monitoring system. It also controls the bias and temperature of the VLPCs. This functionality is embedded in the AFE because it is the only piece of electronics that interfaces to the VLPCs. The analog read out is carried out using the 128-channel SVXIIe chip developed for the Central Fiber Tracker in the DØ experiment at Fermilab. The chip includes preamplifier, analog delay, digitization, and data sparsification. Input charge is integrated on the preamplifier and it is delivered to a 32-cell analog pipeline. Upon a experiment trigger accept, this analog information is fed to a parallel set of ADCs. Digitization utilizes both edges of a 53 MHz clock, providing 8 bits of analog information in 2.4 µs. The readout of the SVXIIe chips mounted on the AFE boards is coordinated via signals 25

27 from the sequencers and sequencer controllers. The SVXIIe sequencer provides timing and control signals for the SVXIIe chips. These signals are regenerated by interface boards that also control power and bias for the SVXIIe chips and provide interfaces to the monitoring systems. The sparsified data are transferred from the sequencers to VME transition modules (VTM) via fiber optics cables. The VTMs transfer the data to buffers in VRBs in VME crates. A VRB controller mounted in the same crate controls the assignment of buffers. Upon a trigger accept, the data from the VRBs are sent to the readout DAQ computers for eventual building of event blocks and data storage. 26

28 Chapter 4 Reply to question #4 Does the collaboration have sufficient strength and expertise to deal with the large number of new technologies? Specifically, is the third tracker (presently requiring a new technology for financial reasons) required at turn-on? Dec. 13, 2008, data collected by C.F. Perdrisat Personnel, commitments and funding We discuss the planned organization of the work involved for each subsystem: 1) dipole with auxiliary vacuum chambers, shielding, support structure 2) electromagnetic calorimeter 3) first tracker 4) second tracker or first polarimeter 5) third tracker or second polarimeter 6) hadron calorimeter 7) trigger electronics Dipole with auxiliary equipment The project leader for this part is B. Wojsekhowski; in addition the group involves J. LeRose, both JLab. They will design and construct the shielding in the beam hole through the yoke, target vacuum chamber extension, front shielding and magnet support. Design and construction work associated with the magnet are supported by Hall A. Other major contributors are J. Annand and D. Hamilton, University of Glasgow. Their main responsibility is field calculation and system design, including field clamps and beam hole. The Glasgow group plan to include funding for 12 GeV related physics at JLab with their next funding request, next spring, including this project. Electromagnetic calorimeter The existing BigCal calorimeter will require modifications to the readout electronics associated with formation of a trigger signal, as well as for refining the radiation damage healing procedure. The project leader for this part is L. Pentchev; in addition, C.F. Perdrisat and students will be responsible for the timely preparation of BigCal. Funds required for 27

29 purchase of additional linear summers will come from the trigger electronics funding (see 1.7) First tracker Design, construction and financing of the first tracker is the responsibility of the INFN, Rome group with the support of other INFN groups. It is expected that the required detectors, with associated readout chips (and associated electronics), as detailed in this CDR, will be delivered to JLab complete. The group includes E. Cisbani (leader), F. Cusanno, F. Garibaldi, S. Frullani and G.M. Urciuoli; in addition the group has technical support at their home laboratories. The project has long range support, with one year funding already given; remaining funding to follow on a year-by-year basis. The Glasgow group has provided a number of GEM foils which included in prototypes currently being tested with beam in Hall A. Second tracker The second tracker is the detector component of the first polarimeter; it is the responsibility of groups at UVa (N. Liyanage (leader), R. Lindgren, D. Day), NSU (V. Punjabi, M. Khandaker) and W&M (C.F. Perdrisat, L. Pentchev). Other major contributors are J. Annand and D. Hamilton, University of Glasgow. The UVa group has obtained UVa financial support at the level of $160k. An MRI of approximately $900k is to be submitted in January It will provide for the establishment of a clean room for the assembly of the GEM detectors at UVa, for a technician and students stipends, and secondary work facilities at NSU and W&M for prototyping, test of components and assembly of components. Third tracker The third tracker is the detector component of the second polarimeter. The group consists of A. Sarty (leader) at Saint Mary s U. (N.S., Canada), D. Hornidge at Mount Allison U. (N.B., Canada), M. Khandaker at NSU, W. Boeglin and P. Markowitz at Florida International U (FIU). A. Sarty has requested from NSERC (Canadian funding agency) small initial funding for in order to initiate prototyping; a request for $150,000 will then be submitted in October 2009 through NSERCs RTI program (Research Tools and Instrumentation) in order to fund construction of the scintillating fiber detectors (including purchase of the fibers). The group will collaborate closely with the Detector Group at TRIUMF national lab (B.C., Canada) for the coupling of fiber-optic light-guides to the scintillating fibers. The Visible Light Photon Counters (VLPC) from D0 s Central Fiber Tracker are to be installed at JLab, and thus the connection of the light-guides to the VLPCs will also be done at JLab. The FIU will assume the task of integrating the VLPC system into the third tracker. Hadron calorimeter The primary movers for this component essential for the formation of a trigger are G.B. Franklin (leader) and B. Quinn, CMU, C.F. Perdrisat, and L. Pentchev, W&M, V. Punjabi and M. Khandaker, NSU. The current plan is to have our Dubna collaborators (I. Savin et al., Lab. for High Energy, Dubna) develop a proposal to JINR, requesting partial support of the construction of the required modules on a design similar to the COMPASS HCAL1 design. There are plans to secure the remaining financial support from the CMU collaboration; a direct contribution from CMU is also expected. Trigger electronics 28

30 The group consists of J. Calarco (leader), U. of New Hampshire, and R. Gilman, Rutgers U. Calarco has requested funding in his DOE grant to contribute $50,000 per year for 5 years. This will cover the cost of the new electronics required for the trigger, including the two calorimeters. Appendix A second polarimeter is required for the GEp(5) experiment. Based on data obtained with the similar double-polarimeter setup built in Hall C and installed in the HMS, recently used for GEp(3), we know that the probabilities for single scattering in the first and in the second analyzers are typically 38 and 26%, respectively. This corresponds to an increase of the combined probability to scatter a proton in either polarimeter by a factor of 1.7, compared to a single analyzer. This factor does not vary significantly when the selection criteria are changed. The analyzing power in the second polarimeter is similar to that of the first. The consequence of not having the second polarimeter, for a fixed total beam on target time, is an increase of the absolute statistical error bar on the G Ep /G Mp ratio by a factor of , or 30%. Form factor experiments are typically statistically limited, and thus the consequence of not having the second polarimeter is an increase of the final uncertainty of 30%. To achieve the proposal error bar with only one polarimeter requires 1.7 times as much beam time. With a production time of 1344 hours (of beam on target) approved for this experiment, this would correspond to a direct increase of beam time required of 940 hours or 39 days (at least 78 calendar days). Building the second polarimeter is economically strongly preferable, in fact required to attain the proposal goals. To conclude, the physicists committed to the realization of the SBS setup have experience in building subsystems of various sizes at JLab, in Hall A and Hall C. Although some of the technology required is new at JLab, it has been developed and has been proven to work at CERN, Fermilab and elsewhere. The team involved will benefit from participation in the RD51 group at CERN, which is dedicated to the development of microgas detectors like GEM and Micromega. Also, members of the team will participate in the D0 experiment to acquire the necessary expertise. There is no showstopper we can think of at this time! 29

31 Bibliography [1] A.N. Dubrovin, Budker Institute of Nuclear Physics, Novosibirsk, Russia. [2] L. Schmitt et al., The DAQ of the COMPASS Experiment, IEEE Trans. Nucl. Sci. 51, (2004) ; C. Altunbas et al., Nucl. Instr. and Meth. A 490 (2002) [3] V.M. Abazov et al., Nucl. Instr. and Meth. A 565 (2006) ; A. Bross et al., Nucl. Instr. and Meth. A 477 (2002) [4] M.J. French et al., Nucl. Instr. and Meth. A 466 (2001) [5] R. Michaels and D. Abbott, Hall A Status Report , ( [6] The DØ Upgrade Central Fiber Tracker, Technical Design Report, ( home.html). 30