Test Suite for the LHCb Muon Chambers Quality Control

IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 55, NO. 2, APRIL 2008 723 Test Suite for the LHCb Muon Chambers Quality Control V. Bocci, G. Carboni, A. Massafferri, R. Nobrega, E. Santovetti, and A. Sarti Abstract—This paper describes the apparatus and the procedures implemented to test the Front-End (FE) electronics of the LHCb Muon Detector Multi Wire Proportional Chambers (MWPC). Aim of the test procedure is to diagnose every FE channel of a given chamber by performing an analysis of the noise rate versus threshold and of the performances at the operational thresholds. Measurements of the key noise parameters, obtained while performing quality tests on the MWPC chambers before the installation on the experiment, are presented. The Test Suite proved to be an automatic, fast and user-friendly system for mass production tests of chambers. It provided the electronic identification of every chamber and FE board, and the storage and bookkeeping of test results that will be made available to the Experiment Control System during data taking. Index Terms—Electronics, gas detectors, software quality, wire chambers. I. INTRODUCTION HE LHCb experiment is devoted to the precision measurements of CP violation and rare decays in the system at LHC. The LHCb Muon Detector [1] consists of five stations (M1-M5) of rectangular shape, equipped with Multi Wire Proportional Chambers (MWPC) and Gas Electron Multiplier (GEM) detectors, located along the beam axis and interspersed with iron filters. Each station is divided into four regions (R1-R4) with increasing distance from the beam axis with chambers having different dimensions and logical channel granularity, for a total of 1380 chambers. The geometry is projective and optimized in order to identify muons in the zero resolution of 20%. trigger level (L0) with a The L0 Muon trigger is designed to have an efficiency of at least 95% within a 25 ns window. This requirement is achieved by using MWPCs with two double gas gaps [1] in logical OR having a time resolution lower than four ns rms. The readout circuit peaking time and the operational threshold must be low enough to minimize the time slewing effect and not to degrade the chamber time resolution. The test of each chamber is done in three main steps: T • The identification of dead electronic channels, open channels given by bad connectivity and short-circuits inside chambers. Manuscript received October 18, 2007; revised December 18, 2007. V. Bocci and R. Nobrega are with the Università La Sapienza, 00185 Rome, Italy and also with Sezione INFN, 00185 Rome, Italy. G. Carboni, A. Massafferri, and E. Santovetti are with the Università Tor Vergata, 00133 Rome, Italy and also with Sezione INFN, 00185 Rome, Italy. A. Sarti is with INFN Sezione LNF, 00044 Frascati, Italy. Digital Object Identifier 10.1109/TNS.2008.918522 Those problems are identified measuring the noise versus threshold spectra. • The quality checks of the readout and control cables, verified using a pulse injection procedure. • The verification of the noise level of each channel at the operational threshold. An upper limit of one kHz on the noise rate for each Front-End (FE) channel has been set considering that the L0 trigger takes as input the OR of up to six input FE channels and tolerate a rate up to 10 kHz without a significant performance degradation [2]. This limit imposes a minimum operational threshold for the chamber signal detection circuit. For this reason, it is crucial to evaluate the noise level of equipped chambers before their final installation on the detector. FE The test procedure has been applied to all the boards used to readout the FE channels of the Muon System. The noise versus threshold analysis results are used to set different thresholds on each FE channel. Such threshold fine tuning is needed when trying to minimize the HV operational value, and hence the aging processes, while keeping a good efficiency. The rate spectra are also used, for the first time in a mass production test, to yield an indirect measurement of the capacitance that allows to diagnose the whole chain of the system (FE electronics plus chamber) and not just the FE channel. The chambers, FE characteristics and Test Suite setup details are described in Sections II and III, while the experimental methods and the tests results are presented in Sections IV–VI. II. CHAMBERS AND FRONT-END CHARACTERISTICS Due to the different dimensions and design details of the twenty types of chambers of the Muon System [1], the input capacitance, as seen by the FE electronics, can vary from 20 to 220 pF. Two types of chamber readout are possible, anode or cathode, with different polarities. A dedicated 8-channel amplifier, shaper and discriminator rad-hard chip [3] (CARIOCA) with different preamplifiers in the input stage and selectable input polarity has been developed to collect signals from either readout. The on-detector circuitry is composed of two boards: SparkProtection and CARDIAC board. The former board makes use of passive components while the latter processes and digitalizes chamber signals. A CARDIAC board accommodates two CARIOCAs and one DIALOG chip [4] which has been developed to control and readout the CARIOCA chip. The DIALOG chip has 16 DAC channels to individually set CARIOCA thresholds, 16 24-bit internal counters and allows for timing adjustments of the signal. 0018-9499/$25.00 © 2008 IEEE 724 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 55, NO. 2, APRIL 2008 The CARIOCA discriminator makes use of Differential Threshold Voltage (DTV) technique which provides a differential threshold to the amplifier from a single polarity adjustable reference threshold voltage. As a consequence, by varying the reference threshold we obtain symmetric values of the threshold applied to the amplifier. The reference threshold where the minimum threshold value is found is named OffSet and is a characteristic of the each channel. The pattern formed by the set of the 16 OffSet values of a given FE is unique and has been used for board identification. The threshold value ( ) needed for noise studies (see Section IV-B) is calculated by subtracting the channel OffSet and adding the average minimum threshold value (40 mV). The main characteristics of CARIOCA are: • Charge gain ( ): from 8 to 16 mV/fC, depending on the input capacitance. • Offset: • Equivalent Noise Charge (ENC): from 0.3 to 2 fC, depending on the input capacitance • Input impedance: 50 Ohm where the quoted errors are indication of the spread of the chip characteristics after the production. III. TEST SUITE SETUP The Test Suite setup consists of a MWPC equipped with CARDIAC boards and the control and readout circuitry, all supervised by a PC. The FE electronics control is performed by means of a Service Board (SB) [5] that can access in parallel all the CARDIAC boards under test and can be used to inject signals and to control internal and external counters. The SB is accessed using the CANopen protocol [6] with a commercial adapter [7] while the SB-CARDIAC communication is given by a custom 3 wire serial link with LVDS signals as physical layer. The readout of the CARDIAC boards is performed using the DIALOG internal counters as well as a 64-channels ACQ board [8] which is controlled by a VME v1718 CAEN module. A Visual C++ ROOT [9] integrated software has been developed to control all circuits, perform the analysis followed by the diagnostics of the chambers and visualize the results. The Test Suite is also used to identify the chambers and the CARDIAC bar-codes, read with a bar-code reader, and to store the test results in a database that will be used by the Experiment Control System during data taking. Fig. 1. Noise rate as a function of the threshold applied. The methods employed to measure those parameters are described below. A. Offset Fig. 1 shows the raw data obtained performing a threshold scan: the noise counts are shown as a function of the threshold applied. The peaked spectrum of the noise rate versus threshold, shown in Fig. 1 is due to characteristics of the DTV circuit described in Section II. The Offset is measured from this spectrum: its value is equal to the threshold where the maximum rate is observed. To evaluate the error of this Offset measurement procedure 16 channels of one CARDIAC board have been used: they have been measured 100 times with four different input capacitances (56, 100, 180 and 220 pF). Results have shown that the Offset measurement is not correlated to the input capacitance. The correspondent error is dominated by the DIALOG DAC resolution and is small enough to be safely neglected. B. Equivalent Noise The noise measured in each CARDIAC channel is proportional to the preamplifier integration time and to the input chamber capacitance. The Test Suite performs an indirect capacitance measurement to diagnose possible problems. The translation of the electronic noise in terms of the capacitance was shown to be adequate since the latter can be easily compared with a direct capacitance measurement. Considering a charge sensitive amplifier, the noise affects the amplitude measurement resolution and consequently, the minimum detectable charge value. In the presence of Gaussian noise the probability distribution for the amplifier response is given by: IV. EXPERIMENTAL METHOD There are three key parameters that are used by the Test Suite to perform a quality check of each FE board: Offset, Equivalent Noise and zero threshold crossing rate. They are measured using a threshold versus noise rate analysis and they provide important diagnostic informations on the CARIOCA preamplifier. The Offset and Equivalent Noise are used for the chamber diagnostics while the zero threshold crossing rate is available as an extra information during the test for consistency checks. The Offset information is also used to set the different thresholds on each FE board channel. (1) where is the Equivalent Noise Voltage (ENV). as the zero threshold ( ) crossing rate, Defining the noise rate as a function of the threshold to noise ratio ( ), taking into account the circuit dead time, can be described by [10]: (2) BOCCI et al.: TEST SUITE FOR THE LHCb MUON CHAMBERS QUALITY CONTROL Fig. 2. Example of the noise curve measurement method. Both the right and left shapes of the noise spectrum are shown. The result of the fit to the average is superimposed as a dotted line. 725 Fig. 3. ENC parameter of a single channel as a function of the input capacitance (56, 100, 180 and 220 pF). This equation has been used to fit the data collected during the threshold scan test of the chambers (see for example Fig. 2) values for each CARDIAC channel. and measure the Since DTV presents a symmetrical behavior with respect to the Offset we can extract the of each CARDIAC channel by taking the average of the slopes of the left and right shapes of the noise spectrum. Fig. 2 shows an example of such measurement: left and right slopes are shown, as well as the fit to the average. is The chamber capacitance as a function of the measured evaluated using the following equation: Fig. 4. ENC error measurement of channels of different FE boards connected to an input capacitance of 47, 100 and 220 pF. (3) where is the expected gain of the amplifier for a given detector capacitance, as obtained from the detector simulations, and are the ones used in the following and the parameters equation that parametrizes the Equivalent Noise Charge (ENC) : as a function of (4) and for the negative with amplifier and and for the and have been obtained using a charge inpositive one. jection technique. The error on the ENC measurement has been estimated using the same setup and method employed in the estimation of the offset measurement error (see Section IV-A). The results are shown in Fig. 3: a precision better than 0.05 fC is obtained. The error evaluation study has been done on different channels of different FE boards. Fig. 4 shows the measurement of five different FE boards connected to input capacitances of 47, 100 and 220 pF. The resolution worsening is explained by taking into account the spread of channels characteristics among a given FE board. The resolution found, about 0.1 fC, is good enough to diagnose the equipped chambers. C. Zero Threshold Crossing Rate The zero threshold crossing rate ( ) at the output of an ideal band-pass filter with lower and upper cutoff frequencies and , considering only positive excursions as done in [11], is given by: (5) This rate can be also directly extracted from (2) with a large uncertainty due to the logarithmic relation and the large dependence of the rate on the minimum threshold value. This uncertainty can be reduced using the invariance of the crossing rate with respect to the input capacitance. Fig. 5 shows the determiusing the intersection point of three different fits nation of to datasets obtained using different input capacitances. This method can be used when measuring the different FE channels of a given chamber since each channel has different characteristics and capacitance according to its physical position in the chamber. V. PROCEDURES AND DIAGNOSTICS A. Cable Checking and Auto-Injection The readout cable configuration is checked using a train of one hundred pulses injected in the first channel of each FE board. The signal readout in a different FE board is an indication of the wrong cabling of the FE. After that the connectivity test is performed by the simultaneous injection of five hundred pulses separately in the even and odd channels of all the FE 726 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 55, NO. 2, APRIL 2008 Fig. 5. Example of the crossing point method to estimate f . The same channel has been measured using three different input capacitances: 47, 100 and 220 pF. The noise rate distributions are shown using threshold values in 10 MHz is which the offset has been subtracted. A crossing point at f found. ' boards. In these two steps the threshold is set to the maximum value to suppress the noise. The readout of the CARDIAC internal and ACQ external counters is then compared to the number of injected pulses: agreement within 1% tolerance is required. Any discrepancy can be used to identify bad connectivity or chip counters failures with very high efficiency. This procedure has been used successfully in the experimental area for a reliable and fast check of the cable ) of stations M2 to configuration of all the chambers ( M5. B. Threshold Scan The Noise Rate versus threshold spectrum (see Fig. 1) is obtained by counting in a 200 ms gate time. Several checks are performed using these data. The first one verifies that the number of measured points obtained during the scan is sufficient to fit the data. Channels not showing any noise are identified as dead channels. The second one checks that the measured Offset values are within the limits given by the CARIOCA characteristics. In addition, the Offset pattern relative to the 16 channels of each FE is compared with the results obtained in the mass production quality test, to check if the FE boards on the chamber have been correctly identified. Finally a finer noise analysis is performed. The Equivalent Noise parameter in Volts units is obtained by fitting the experimental data using (2). The ENV and the expected charge gain value of the chamber under test are then used to estimate using (3). Open channels and short circuit inside the chambers can be identified by comparing the expected capacitance and . The LHCb chambers have capacitances in the range 20 to 220 pF: a measured value below 15 pF (above 250 pF) is a clear indication of open channel (short circuit). Short circuit are of particularly difficult diagnosis, since the noise enters exclusively through the Spark-Protection Board which presents a capacitance of about 2 nF, outside the range of the FE band-width. The signature of about 250 pF is due the saturation of the FE working point that decreases dramatically the capacitance seen by the system. Fig. 6. Offset measured for all the various types of chambers. TABLE I DETECTOR CAPACITANCE COMPUTED AND MEASURED VALUES (C ). As an additional check the average of the measured capacitances for all the channels is required to be within from the expected capacitance. The diagnostic routine subsequently value of every channel is inside a three rms verifies if the window with respect to the average. The software identify channels that are out of range, issuing an alarm message, and possible sources of the observed anomalous behaviour are indicated. C. Noise Rate at the Nominal Threshold The LHCb Muon TDR [1] requires a maximum noise rate of 100 Hz per channel for MWPCs. Recent studies of LHCb Muon Detector have shown that channels with a noise rate up to a few kHz are not critical [2]. The diagnostic tool is able to measure the electronics noise rate at three different thresholds (equal to 6, 7 and 8 fC, for cathode readout chambers, and 10, 12 and 14 fC, for anode readout chambers). Specific errors or warning messages are issued if rates larger than 1 kHz are measured. VI. RESULTS About 400 MWPCs from two different production sites have been tested. Those chambers have been installed in stations M2 to M5 in different regions (R2-R4). The Test Suite was able to diagnose chambers of largely different characteristics and readout (anode in R4 region, cathode in R2 and R3). Fig. 6 shows the measured offset for all the FE channels of all the measured chambers together. Detector capacitance values have been obtained for all the different types of chambers and are shown in Table I together with the values expected from the detector simulation. The meadistributions (shown in sured values are extracted from the BOCCI et al.: TEST SUITE FOR THE LHCb MUON CHAMBERS QUALITY CONTROL Fig. 7. C values measured from CARDIACS of M5R4, M5R3, M5R2 and M4R4 chambers. 727 Fig. 9. Noise rate values measured from CARDIACS of M5R4, M5R3, M5R2 and M4R4 chambers. Fig. 8. C values measured from CARDIACS of M4R3, M4R2, M3R3 and M2R3 chambers. Fig. 10. Noise rate values measured from CARDIACS of M4R3, M4R2, M3R3 and M2R3 chambers. Figs. 7 and 8) of all the different FE channels for each different chamber type. The systematic difference between the computed and measured values is due to the missing calculation of the printed circuit capacitance, that has been measured to be nearly 25 pF (depending on the channel position and chamber type). The measured values, obtained using an indirect method, are in good agreement with the ones expected for the various types of chambers when properly taking into account this shift. The rate analysis result (see Figs. 9 and 10) shows that no channel with rate larger than 1 kHz has been found: all the chambers sent to the experimental area for installation did satisfy the quality requirements. The Test Suite capability to detect FE malfunctions has been cross checked by testing several chambers with a different system that makes use of cosmic rays. The cosmics acquisition 728 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 55, NO. 2, APRIL 2008 data analysis confirmed the faulty channels diagnosis obtained by using the Test Suite. The low efficiency channels identified using the Test Suite and the cosmics acquisition were hardware checked by opening some chambers: the short-circuit diagnosis was verified in all cases. VII. CONCLUSIONS The goal of this project was to provide an efficient, fast, reliable and user friendly system able to perform the acquisition, control and analysis of all the FE boards of all the MWPC chambers of the LHCb Muon Detector. The Test Suite described in this paper has been used to test nearly four hundred chambers in the production sites, in the CERN laboratory and in the experimental area. The test procedure that has been adopted uses the results of the noise analysis. The threshold scan fit parameters are translated in terms of the capacitance of each FE channel easing the problem identification and providing an excellent diagnostic tool. This approach can be easily adopted by other collaborations that need to test a large number of electronics channels with the largest efficiency for problems detection before any final cross-check based on cosmics acquisition. The Test Suite was able to correctly identify faulty CARDIAC boards that have been dismounted and replaced, or fixed whenever possible. The Test Suite is also being currently used for the commissioning of all the MWPC Muon chambers in the experimental area. Results have shown that the proposed noise analysis technique provides an accurate and efficient monitoring of the LHCb muon apparatus quality. APPENDIX Figs. 7 and 8 show the detector capacitance measured values for M5R4, M5R3, M5R2, M4R4, M4R3, M4R2, M3R3 and M2R3 chambers respectively. Figs. 9 and 10 show the noise measured values for M5R4, M5R3, M5R2, M4R4, M4R3, M4R2, M3R3 and M2R3 chambers respectively. ACKNOWLEDGMENT The authors would like to thank the INFN for the support to this project, A. Kashchuk for many useful discussions, and A. Rossi and G.Paoluzzi for their technical support. REFERENCES [1] LHCb Muon System TDR LHCb Collaboration, 2001, CERN/LHCC 2001-010. [2] E. Aslanides et al., LHCb Public Note, 2002, LHCb 2002-041. [3] D. Moraes et al., LHCb Public Note, 2000, LHCb 2000-093. [4] S. Cadeddu et al., Nucl. Instrum. Methods Phys. Res. A, vol. A518, p. 486, 2004. [5] V. Bocci et al., “The services boards system for the LHCb muon detector,” in Proc. Nuclear Science Symp. Conf. Rec., Honolulu, HI, Oct./ Nov. 2007, vol. 3, pp. 2134–2140. [6] [Online]. Available: http://www.can-cia.org/canopen/protocol [7] [Online]. Available: http://www.kvaser.com [8] A. Balla, Technical User Manual Rev. 1.0 [Online]. Available: http:// www.lnf.infn.it/~balla/ Datasheet/AMB_manual_v1.pdf [9] R. Brun and F. Rademakers, “ROOT-An object oriented data analysis framework, Proc. AIHENP’96 workshop,” Nucl. Instrum. Methods Phys. Res. A, vol. 389, pp. 81–86, 1997. [10] H. Spieler, “Rate of noise hits in threshold discriminator systems,” in Spring Semester 1998. Berkeley, CA: Univ. of California Press, Lecture Notes, Physics 1998 . [11] S. O. Rice, “Mathematical analysis of random noise,” Bell Syst. Tech. J., vol. 24, p. 46, 1945.