BTeV RICH Testbeam 2005

The DAQ system

A proton travels through C₄F₈O gas in the RICH tank. It produces Cherenkov lights along its trajectory. Cherenkov photons are focused by a mirror downstream onto a MAPMT array, and produce analog electronic signals. The MAPMT tubes are mounted on baseboards which connect to front end hybrids(FEH). The analog signals from MAPMTs are converted into digital signal on FEHs. The digital hit signals are shipped to a DAQ computer via multiplexer (MUX) boards, sets of short and long cables and feed through (FT) boards. The DAQ computer in counting room houses 6 pairs of PCI based PTA (PCI Test Adaptor) and PMC (Programmable Mezzanine Card) boards. The PTA/PMC pairs process the hit signals, add unique numbers to identify different front end boards (plane ID). An incremental counting (BCO ID) of 16MHz clock is also added before hit signals on each PTA/PMC pair so that one can group hits from different PTA/PMC into events (event alignment). Part of cable connections and part IDs can be found in tb05_connection_1.gif and tb05_connection_2.gif.

The MAPMT array consists of 53 MAPMT tubes (A 54th tube was populated by not used). Each MAPMT tube has 16 cells (pixels) in a 4×4 array. Each baseboard can connect up to 16 MAPMT tubes. The mapping of cells and tubes is explained later. The baseboards also provides HV power to MAPMT tubes. The tubes are gain matched into three groups, connected to different power supplies with different voltages. For the grouping and cable connection between baseboards and FEHs please see Steve's plot. The final configuration is slightly different and will be explained later.

Each FEH processes analog signals from up to 128 channels, thus can serve up to 8 MAPMT tubes. Each hybrid connects to one multiplexer (MUX) board which also provides low voltage power. In total 12 FEHs are mounted and 10 of them are really used. Each PTA/PMC pair connects to 2 FEHs.

We have two DAQ computers: one PC with window OS, and the other with Linux. The window PC runs a Labview based program mainly for firmware development and debug. For normal data taking we use Linux machine runs a program coded in C++. The BCO header and hit data direct from PTA memory buffer, with addition of several control words, are written in a "raw data file" on hard disk for further study. The BCO header and hit data words are further processed offline, with hits from same physics event grouped together using the BCO header and other information. The event alignment also happened online for quality monitor, and the aligned events were written to a file for quick offline studies. The C++ based program is modified from "Pomone" package developed by Milano group mainly for BTeV pixel testbeam.

Channel mapping

In total 12 FEHs are used to read out all MAPMT tubes. Each FEH is assigned a plane ID from 1 to 12. In the data file only the plane ID and channel ID of FEH can be reconstructed. Thus a clear mapping between FEH channels and MAPMT cell locations is necessary. In order to clearly identify segments we denote each baseboard, α to δ from top row to bottom row and 1 to 4 from left to right looking from mirror. The connection between FEHs and the base boards are as listed in (tb05_connection_2.gif). Note that base board δ-1 has two FEHs connected to it. The bottom one corresponds to plane 9, which is not used. The other one that is not used is plane 1.

Each FEH has 128 channels with ID from 1 to 128. It can process analog signal from up to 8 MAPMT tubes. Thus a MAPMT is given an octet ID, shown as the second number in plot map_mapmt.gif. The first number corresponds to plane ID. The location, octet ID and plane ID (FEH ID) of each tube are listed in map_feh.dat file. Also listed is the orientation of the tube with respect to the channel ID. With this map file a more detailed FEH-channel to cell location map is easily generated as map_cell.dat (plot in map_cell.ps), which is used by offline analysis code.

Data format

When a PMC receives a trigger signal it generates a 32-bit trigger header including a 26-bit synchronized counting of 16 MHz clock (extended BCO count) which is used to align events between different PTA/PMC pairs. The PMC also passes trigger signal to FEHs.

If a FEH has hit data latched when it receives the trigger signal it ships out 12 14-bit words to PMC. The two least significant bits (LSB) of each word indicate data valid and hit data. The rest of first word is FEH BCO count. Each of the following 11 words contains hits from 12 channels, one bit to one channel. If there is no hit data latched no word is shipped, not even the FEH BCO count.

The PMC re-pack the data words from FEH into 11 hit data word, with FEH BCO count in each hit data word. Information also packed are the plane IDs indicating which FEH the data is from and word IDs indicating which 12 channels. Even if there is no hit from both FEHs connected to the PMC, it ships the trigger header.

The trigger header and hit data words are stored in one of the two EEPROMs on each PTA card. When one of 6 PTAs reports memory full all PTAs receive memory toggle commands and each dumps the memory into a sub-buffer. All 6 sub-buffers together form a superblock with a serial ID. Each sub-buffer is written to a raw data file as one block led by a superblock ID word and length of the block. The superblock ID and the trigger header is to be used to align events.

The raw data file is processed offline to generate a more sophisticated final data file for study of each event. The file contains a run header block at the beginning, run trailer block at the end, both contains general information of the run. In between there are event blocks. The endian issue is taken care of. A file of same format is also created by online program for monitor and quick offline study purpose.

Different studies and list of runs

The setting of the electronics, HV, and trigger windows changes for different tests. The MAPMTs are gain matched into 3 groups and applied different HV. They are listed as HV1 (planes 1, 2, 3, 5), HV2 (planes 4, 6, 8, 9, 10) and HV3 (planes 7, 11, 12). The nominal HV settings are HV1=800, HV2=750 and HV3=700V respectively. The trigger window is set to be 1000 ns or 500 ns for most tests. The nominal values of electronic bias DACs are THR(threshold)=121, Vfp=161, Vfs=163 and Vrc=213. If not specifically stated the bias DACs are set to nominal values.

1. Light leakage test: The light tightness of the the RICH enclosure is much better than last beamtest. The new DAQ system is also less sensitive to random light leakage due to the triggering scheme. The two runs recorded in this test are 823 with lights in the beam hall on, and 824 with most of the lights switched off. The settings are: All HV=1000V, THR=123, Vfp=158, Vfd=165 and trigger window of 1000 ns.
2. Observation of the first ring: While there is beam and the scintillation coincidence trigger was still under testing we took a short run, 836, with an 1K Hz pulser source as trigger to the DAQ. We observed the first ring ( first_ring.jpg , first_ring.ps ). In this run there are two Cherenkov ring events with 43 and 49 hits respectively. The settings are All HV=700V, THR=123, Vfp=158, Vfs=165 and trigger window of 1000 ns.
3. HV scan 1 with beam: This is to determine the nominal operating voltages. We scanned HV between 500 - 950 V at steps of 50V. The runs (847 - 854) were taken with THR=123, trigger window of 1000 ns. They were analyzed briefly to determine plateau as shown in
   1. hv_11_nhits.jpg , hv_11_nhits.ps : average number of raw pixel hits per event.
   2. hv_11_nclus.jpg , hv_11_nclus.ps : average number of cluster hits per event.
   3. hv_11_csize.jpg , hv_11_csize.ps : average size of clusters.
   We use cluster hits instead of pixel hits to rid of cross-talk effect. Any adjacent channels that have hits are grouped in one cluster. A cluster hit is treated as one photon. Due that the geometry location of each MAPMT cell/pixel is correlated with channel ID, this treatment is not totally correct. On the top and bottom of the ring, ie planes 2, 3, 10 and 11, a cluster may be from two or more photon. We also count average number of clusters without background rejection. For determination of nominal setting, however, this treatment is precise enough. From number of clusters and cluster size plots we decided that HV1=800, HV2=750 and HV3=700 is an ideal setting.
4. Threshold scan 1 with beam: With nominal HV setting and trigger window of 1000 ns we scanned threshold from 103 - 126 step of 2. We also took runs without beam for couple of thresholds. Similar to that of HV scan runs we checked number of clusters vs threshold and decided to use 121 as nominal THR (threshold DAC) setting. But for most of rest studies we also took data at THR=105.
   1. th_61_nhits.jpg , th_61_nhits.ps : average number of raw pixel hits per event.
   2. th_61_nclus.jpg , th_61_nclus.ps : average number of cluster hits per event.
   3. th_61_csize.jpg , th_61_csize.ps : average size of clusters.
   4. th_61_cs_feh.jpg , th_61_cs_feh.ps : average size of clusters on each FEH.
5. Trigger window study: This is to study how the width of trigger window affects the data taking. We took data with and without beam and varied trigger window width between 25 ns to 1250 ns.
6. Trigger delay study: This is to test different delays in the Firmware of FEH. The delay was tested between 1 to 6 clocks in the beam with trigger window 500 ns. The nominal value of the delay is 6 clock.
7. Firmware test: Run ID is 963, inside beam with 500 ns trigger window. The purpose is unknown to me.
8. HV scan 2 with beam: This is a refined HV scan with step 25 V and large amount of data at each points. The threshold is set to be 121 and 105, trigger window 500 ns. At three points including the nominal HV setting we took data at different time to study stability.
9. HV scan 3, anti-beam: This scan has same setting as HV scan 2. But the trigger uses anti-coincidence with scintillation counters while beam is on. It is to study the background in beam condition. But I would not be surprised to see Cherenkov rings.
10. Beam intensity: This is to test how the beam intensity affects the data taking. The trigger window is set to 500 ns. The intensity was changed from ~6K/spill to ~70K/spill.
11. HV scan 4, no beam: This scan has same setting as HV scan 2. It is to study the background without beam. We use a noise source at about 1 KHz to trigger the DAQ.
12. HV scan 5, pulser: This scan has same setting as HV scan 2. But there is no beam, instead the LED pulser is enabled by 1 KHz pulse generator. The same pulse was sent to trigger the DAQ. This is to study the cross talk. Since there is no Cherenkov rings it is much easier especially for planes 2, 3, 10 and 11 which are on top or bottom of the configuration ring.
13. Threshold scan 2, pulser: This scan has same setting as threshold scan 1. There is no beam but LED pulser. It is to study cross talk effect.
14. Mirror adjustment: The mirror was adjusted at the beginning of the testbeam runs so that the MAPMT array could cover Cherenkov ring better. With improved gas purity the mirror needs to be adjusted further. The runs are taken with different thresholds and HV settings to have better Cherenkov ring plots ( accum_1112.jpg , accum_1112.ps ).

Issues in data

1. Login to hepsu03 and change default (cd) to location where you want your working directory be.
2. Un-tar tb_050307.tar file. This will create a directory tb_anal.

    tar -xvf /cdat04/btev_rich/tb05/WORK/tb_050307.tar

3. Compile the source code

   cd tb_anal/src
   gmake
   cd ..
   

4. Run the program and display first two events.

   ./rich_tb
   display
   help
   next
   display
   event
   hits
   quit
   

To look into details of each event, one can set parameter ncycle=1 and dump_opt=0. The program prompts for command. The commands and explanations are listed in the table. Abbreviation is accepted providing there is no ambiguity, eg, hi, not h, for hit. One integer parameter following the command may be used in some cases.

The source codes in tb_anal/src directory are mostly written in FORTRAN with interface to C functions. Below is a list of files in this directory. The most interesting one to user is hit_hfill.F where the information are saved in histograms and ntuple.