CREAM: the new NA62 calorimeter readout board Stefano Venditti on behalf of the NA62 LKr group PH-ESE seminar - 03/12/2013
Download ReportTranscript CREAM: the new NA62 calorimeter readout board Stefano Venditti on behalf of the NA62 LKr group PH-ESE seminar - 03/12/2013
CREAM: the new NA62 calorimeter readout board Stefano Venditti on behalf of the NA62 LKr group PH-ESE seminar - 03/12/2013 OUTLOOK • The K→πνν decay • The NA62 experiment • The NA62 liquid Krypton calorimeter • CREAM: The NA62 Calorimeter REAdout Module • Board layout • Firmware properties • Analog signal characterization @ CERN • Outcome of in situ tests • Conclusions Stefano Venditti - CERN PH-ESE 2 THE K→πνν DECAY CHANNEL No FCNC → contribution to K→πνν decay from penguin diagrams only • Top quark contribution dominance • CKM matrix suppression ( Vts*Vtd) → high sensitivity to new physics • Hadronic matrix element can be related to K→eπ0ν decay BR • CKM triangle measurement SM PREDICTIONS: [Brod et al. - Phys. Rev D 83, (2011) 034030] BR(K ) (7.81 0.75 0.29) 1011 BR(KL 0 ) (2.43 0.39 0.06) 1011 PRESENT EXPERIMENTAL STATUS: 15 BR(K ) (1.7311..05 ) 1010 • 7 candidate events from 2 experiments (E787, E949 @ Brookhaven) • Probability of all events being background ~ 10-3 BR( K L 0 ) 2.6 108 (E391a @ KEK) HADRON MATRIX ELEMENT Stefano Venditti - CERN PH-ESE 3 THE NA62 EXPERIMENT NA62 is the last from a long tradition of fixed-target Kaon experiments in the North Area (NA) @ CERN Fixed target experiment: 1.1x1012 Hz 400 GeV SPS protons on a Be target Outgoing charged particles (750 MHz, ~6% K+) selected in momentum (75±1% GeV/c) and direction (X,Y spread<100 μrad) through achromats 10 MHz decay rate on detectors (mainly from K+) GOAL: collect O(100) K+→π+νν events with S/B~10 REQUIREMENTS: 1013 K decays collected, >1012 BG reduction, 10% precision on BG Stefano Venditti - CERN PH-ESE 4 K+→π+νν SELECTION SCHEME • • • • • • • • • Signal in KTAG compatible with a Kaon One reconstructed track in the Gigatracker, no activity in the CHANTI One reconstructed track in the Straws matched in time with a Kaon in GTK Signal in RICH compatible with a charged pion No signals in the LAVs, IRC, SAC compatible with a γ 1 deposit compatible with a pion in the LKr, no additional γs No signal compatible with in MUV3, pion shower in MUV1+2, signal in CHOD Pion vertex within the first 60 meters of the decay volume 15 GeV/c < Pπ < 35 GeV/c Stefano Venditti - CERN PH-ESE 5 K+→π+νν SELECTION SCHEME • • • • • • • • • Signal in KTAG compatible with a Kaon One reconstructed track in the Gigatracker, no activity in the CHANTI One reconstructed track in the Straws matched in time with a Kaon in GTK Signal in RICH compatible with a charged pion No signals in the LAVs, IRC, SAC compatible with a γ 1 deposit compatible with a pion in the LKr, no additional γs No signal compatible with in MUV3, pion shower in MUV1+2, signal in CHOD Pion vertex within the first 60 meters of the decay volume 15 GeV/c < Pπ < 35 GeV/c Stefano Venditti - CERN PH-ESE 5 K+→π+νν selection scheme • • • • • • • • • Signal in KTAG compatible with a Kaon One reconstructed track in the Gigatracker, no activity in the CHANTI One reconstructed track in the Straws matched in time with a Kaon in GTK Signal in RICH compatible with a charged pion No signals in the LAVs, IRC, SAC compatible with a γ 1 deposit compatible with a pion in the LKr, no additional γs No signal compatible with in MUV3, pion shower in MUV1+2, signal in CHOD Pion vertex within the first 60 meters of the decay volume 15 GeV/c < Pπ < 35 GeV/c Stefano Venditti - CERN PH-ESE 5 K+→π+νν SELECTION SCHEME • • • • • • • • • Signal in KTAG compatible with a Kaon One reconstructed track in the Gigatracker, no activity in the CHANTI One reconstructed track in the Straws matched in time with a Kaon in GTK Signal in RICH compatible with a charged pion No signals in the LAVs, IRC, SAC compatible with a γ 1 deposit compatible with a pion in the LKr, no additional γs No signal compatible with in MUV3, pion shower in MUV1+2, signal in CHOD Pion vertex within the first 60 meters of the decay volume 15 GeV/c < Pπ < 35 GeV/c Stefano Venditti - CERN PH-ESE 5 K+→π+νν SELECTION SCHEME • • • • • • • • • Signal in KTAG compatible with a Kaon One reconstructed track in the Gigatracker, no activity in the CHANTI One reconstructed track in the Straws matched in time with a Kaon in GTK Signal in RICH compatible with a charged pion No signals in the LAVs, IRC, SAC compatible with a γ 1 deposit compatible with a pion in the LKr, no additional γs No signal compatible with in MUV3, pion shower in MUV1+2, signal in CHOD Pion vertex within the first 60 meters of the decay volume 15 GeV/c < Pπ < 35 GeV/c Stefano Venditti - CERN PH-ESE 5 K+→π+νν SELECTION SCHEME • • • • • • • • • Signal in KTAG compatible with a Kaon One reconstructed track in the Gigatracker, no activity in the CHANTI One reconstructed track in the Straws matched in time with a Kaon in GTK Signal in RICH compatible with a charged pion No signals in the LAVs, IRC, SAC compatible with a γ 1 deposit compatible with a pion in the LKr, no additional γs No signal compatible with in MUV3, pion shower in MUV1+2, signal in CHOD Pion vertex within the first 60 meters of the decay volume 15 GeV/c < Pπ < 35 GeV/c Stefano Venditti - CERN PH-ESE 5 K+→π+νν SELECTION SCHEME • • • • • • • • • Signal in KTAG compatible with a Kaon One reconstructed track in the Gigatracker, no activity in the CHANTI One reconstructed track in the Straws matched in time with a Kaon in GTK Signal in RICH compatible with a charged pion No signals in the LAVs, IRC, SAC compatible with a γ 1 deposit compatible with a pion in the LKr, no additional γs No signal compatible with in MUV3, pion shower in MUV1+2, signal in CHOD Pion vertex within the first 60 meters of the decay volume 15 GeV/c < Pπ < 35 GeV/c Stefano Venditti - CERN PH-ESE 5 K+→π+νν SELECTION SCHEME • • • • • • • • • Signal in KTAG compatible with a Kaon One reconstructed track in the Gigatracker, no activity in the CHANTI One reconstructed track in the Straws matched in time with a Kaon in GTK Signal in RICH compatible with a charged pion No signals in the LAVs, IRC, SAC compatible with a γ 1 deposit compatible with a pion in the LKr, no additional γs No signal compatible with in MUV3, pion shower in MUV1+2, signal in CHOD Pion vertex within the first 60 meters of the decay volume 15 GeV/c < Pπ < 35 GeV/c Stefano Venditti - CERN PH-ESE 5 THE NA62 TDAQ SYSTEM Na62 will apply 3 trigger levels to events before writing data on disk • L0T: FPGA based, fixed latency ~ 1 ms 10 MHz: Physics events • L1T: PC based, all data available except LKr’s, max latency: ~1 s 1 MHz • L2T: final decision taken using data from all detectors, max latency ~ burst length 100 KHz 10 KHz: On-disk events The TALK BOARD processes TRIGGER PRIMITIVES from fast detectors and takes the final L0T decision. The TALK cannot be used at regime because of some intrinsic limitations. 2 SOLUTIONS are currently under test Stefano Venditti - CERN PH-ESE 6 TEL62 & TDCB TEL62 • Major upgrade of TELL1 board (LHCB) • 4 PP-FPGAs, 1 SL-FPGA (Stratix III) • 4 200 pin connectors for mezzanines • 4 DDR2 (2 Gbytes each) • Quad-GBE output mezzanine Data is collected from TDCBs and stored in the DDR waiting for the L0 trigger; the proper data is then sent to the PC farm through the GBE. The SAME DATA is used to produce L0T primitives TDC BOARD (TDCB) • Developed by NA62 Pisa group • 4 CERN HPTDCs onboard • 1 Altera Cyclone III • 2x 1MB SRAM TDC data is buffered,timestamped and sent to TEL62. TDC emulation in the FPGA available Stefano Venditti - CERN PH-ESE 7 THE NA62 LKr CALORIMETER The NA62 experiment will reuse the NA48 liquid Krypton calorimeter (LKr hereafter) in order to: • Veto particles in the forward direction • Provide a high-precision measurement of the electromagnetic energy deposits DETECTOR INFORMATION • 10 m3 liquid Krypton calorimeter, 1.25 m deep (27 X0) • 13284 2x2 cm2 cells, projecting geometry • Signal is formed by the collection of electrons ionized by the passage of a particle in the Krypton (~ 2.5 μA/GeV) • Preamplifiers inside the LKr tank • Calibration system mounted on the LKr tank THE LKR UNDER CONSTRUCTION CELL’S ZIG-ZAG SHAPE LKR SECTION (1/4) Stefano Venditti - CERN PH-ESE 8 THE CREAM BOARD The Calorimeter REAdout Module (CREAM) is a 6U VME board developped by CAEN upon CERN specifications.It will replace the old LKr readout used in the NA48 experiment FRONT PANEL Ethernet port: L1 data output Trigger sum links: Data to LKr L0T processor 8 Gbytes DDR3 module FPGA: Altera Stratix IV P0 custom bus 2 x 16 channels input connectors 14 bit ADCs Stefano Venditti - CERN PH-ESE 9 THE CREAM BOARD THE MOTHERBOARD digital logic, trigger, data buffer (8GB DDR3) THE DAUGHTERBOARD analog signal shaping, digitisation Stefano Venditti - CERN PH-ESE 10 THE CREAM BOARD FUNCTIONS • INPUT SIGNAL SHAPING: the 2.7 μs long triangular signal from LKr channels is shaped into a 70 ns FWHM pseudo-Gaussian signal • DIGITISATION: shaped signals are digitised @ 40 MHz by octal 14 bit ADCs and copied in a circular buffer • FIRST TRIGGER LEVEL (L0T): upon reception of the L0T signal through the custom P0 VME backplane, data is moved from the circular buffer to the L0 buffer • SECOND TRIGGER LEVEL (L1T): when a L1T signal is received through a Multiple request UDP packet (MRP) data is sent to the PC farm • TRIGGER SUM LINKS: the sums of the digitised samples from two groups of 16 channels each are serialized inside the FPGA and sent to the LKr L0 processor Stefano Venditti - CERN PH-ESE 11 THE CREAM BOARD FUNCTIONS • INPUT SIGNAL SHAPING: the 2.7 μs long triangular signal from LKr channels is shaped into a 70 ns FWHM pseudo-Gaussian signal • DIGITISATION: shaped signals are digitised @ 40 MHz by octal 14 bit ADCs and copied in a circular buffer • FIRST TRIGGER LEVEL (L0T): upon reception of the L0T signal through the custom P0 VME backplane, data is moved from the circular buffer to the L0 buffer • SECOND TRIGGER LEVEL (L1T): when a L1T signal is received through a Multiple request UDP packet (MRP) data is sent to the PC farm • TRIGGER SUM LINKS: the sums of the digitised samples from two groups of 16 channels each are serialized inside the FPGA and sent to the LKr L0 processor Stefano Venditti - CERN PH-ESE 11 THE CREAM BOARD FUNCTIONS • INPUT SIGNAL SHAPING: the 2.7 μs long triangular signal from LKr channels is shaped into a 70 ns FWHM pseudo-Gaussian signal • DIGITISATION: shaped signals are digitised @ 40 MHz by octal 14 bit ADCs and copied in a circular buffer • FIRST TRIGGER LEVEL (L0T): upon reception of the L0T signal through the custom P0 VME backplane, data is moved from the circular buffer to the L0 buffer • SECOND TRIGGER LEVEL (L1T): when a L1T signal is received through a Multiple request UDP packet (MRP) data is sent to the PC farm • TRIGGER SUM LINKS: the sums of the digitised samples from two groups of 16 channels each are serialized inside the FPGA and sent to the LKr L0 processor Stefano Venditti - CERN PH-ESE 11 THE CREAM BOARD FUNCTIONS • INPUT SIGNAL SHAPING: the 2.7 μs long triangular signal from LKr channels is shaped into a 70 ns FWHM pseudo-Gaussian signal • DIGITISATION: shaped signals are digitised @ 40 MHz by octal 14 bit ADCs and copied in a circular buffer • FIRST TRIGGER LEVEL (L0T): upon reception of the L0T signal through the custom P0 VME backplane, data is moved from the circular buffer to the L0 buffer • SECOND TRIGGER LEVEL (L1T): when a L1T signal is received through a Multiple request UDP packet (MRP) data is sent to the PC farm • TRIGGER SUM LINKS: the sums of the digitised samples from two groups of 16 channels each are serialized inside the FPGA and sent to the LKr L0 processor Stefano Venditti - CERN PH-ESE 11 THE CREAM BOARD FUNCTIONS • INPUT SIGNAL SHAPING: the 2.7 μs long triangular signal from LKr channels is shaped into a 70 ns FWHM pseudo-Gaussian signal • DIGITISATION: shaped signals are digitised @ 40 MHz by octal 14 bit ADCs and copied in a circular buffer • FIRST TRIGGER LEVEL (L0T): upon reception of the L0T signal through the custom P0 VME backplane, data is moved from the circular buffer to the L0 buffer • SECOND TRIGGER LEVEL (L1T): when a L1T signal is received through a Multiple request UDP packet (MRP) data is sent to the PC farm • TRIGGER SUM LINKS: the sums of the digitised samples from two groups of 16 channels each are serialized inside the FPGA and sent to the LKr L0 processor Stefano Venditti - CERN PH-ESE 11 THE CREAM BOARD FUNCTIONS • INPUT SIGNAL SHAPING: the 2.7 μs long triangular signal from LKr channels is shaped into a 70 ns FWHM pseudo-Gaussian signal • DIGITISATION: shaped signals are digitised @ 40 MHz by octal 14 bit ADCs and copied in a circular buffer • FIRST TRIGGER LEVEL (L0T): upon reception of the L0T signal through the custom P0 VME backplane, data is moved from the circular buffer to the L0 buffer • SECOND TRIGGER LEVEL (L1T): when a L1T signal is received through a Multiple request UDP packet (MRP) data is sent to the PC farm • TRIGGER SUM LINKS: the sums of the digitised samples from two groups of 16 channels each are serialized inside the FPGA and sent to the LKr L0 processor Stefano Venditti - CERN PH-ESE 11 CREAM CRATE ORGANIZATION • 16 CREAMs will be housed in a VME crate • 28 CREAM crates, organised in 8 racks, will readout the whole calorimeter • The TTC-LKr board is placed in the 11th slot of each crate SLOW CONTROL: up to 8 bridges (CAEN VX2718) daisy chained, four links controlled by a single A3818 PCIe card Stefano Venditti - CERN PH-ESE 12 THE TTC-LKr BOARD • 6U VME64x board, Xilinx Spartan-6 FPGA onboard • TTC-FMC mezzanine (CERN) using the ADN2814 clock-data recovery IC FUNCTIONS • Distribution of 40 MHz clock, L0T signal, start/end of burst and broadcast commands to the CREAMs in a crate • Crate ID number • Reception of CHOKE and ERROR signals from CREAMs and their transmission to the L0T processor TTC SOURCES • optical 160 Mbps BPM encoded bit-stream • electrical front-panel inputs • internal rate-programmable TTC signal generator • generated by a VME access The selected source is available for the CREAMs on the P0 backplane. All TTC signals are synchronised with the selected clock reference Stefano Venditti - CERN PH-ESE 13 INPUT SIGNAL The signals from the LKr channels go into the CREAM through two connectors delivering 16 differential pairs each to the CREAM input 16 CHANNELS CONNECTORS PLUGGED IN A CREAM The signal is shaped into a 70 ns FWHM Gaussian by shapers (20 ns differentiation +Bessel filter) mounted on the CREAM daughterboard THE SHAPING CIRCUIT AND THE ADC CHIP Stefano Venditti - CERN PH-ESE 14 DIGITISATION The shaped signals are digitised by four commercial octal 14-bit ADCs (AD9257 from Analog Devices) mounted on the daughterboard. Radiation hardness not required MAIN CHIP FEATURES • 14 bits dynamic range • 2V p-p input voltage range • SNR 75.5 dB • DNL ± 0.6, INL ± 1.1 (typical) • 55 mW per channel • Activable internal patterns Stefano Venditti - CERN PH-ESE 15 FPGA & DDR3 FPGA The firmware is housed in an ALTERA Stratix IV FPGA mod. EP4SGX180 • Equivalent LEs: 175750 • Registers: 140600 • Embedded memory: 11430 Kb TOTAL RESOURCE USAGE: ~32% (max 40% requested in specifications) 8 GB DDR3 SODIMM MODULE • Both circular and the L0 buffer are implemented in the DDR3 module directly connected to the FPGA • Circular buffer is 1M x 256-bit wide, allowing for 12.5 ms (>> L0T latency) data storage • L0 buffer is 255M x 256 bit wide, allowing for 16 s (> spill length) data storage at the nominal 100 KHz L1T rate and 8 samples/trigger Stefano Venditti - CERN PH-ESE 16 THE TSL The CREAM system data output is too large to be fully exploited for the L0T decision Each CREAM computes Trigger Sums (TS) and sends them to the LKr L0T processor • Samples from 2 groups of 4x4 channels summed every 25 ns • Baseline subtracted and gain correction applied to each channel of the sum • The two sums (16 bits each, the 2 LSBs are dropped) serialised on two pairs of a standard RJ45 connector • Serialization inside FPGA following Texas Instruments DS92LV18 IC coding • The TELDES board, plugged on a TEL62 and hosting 16 DS92LV18, deserializes the sums and sends the data to the TEL62 PPs. Stefano Venditti - CERN PH-ESE 17 CREAM L0 TRIGGER PROCESSING Sampled data are housed in a 12.5ms wide circular buffer built inside the DDR3. Data is continuously copied in the buffer, waiting for a L0 trigger 32x2 byte 32x2 byte 32x2 byte 32x2 byte samples samples samples samples EXAMPLE • samples collected: 4 • latency: 10 clock cycles 32x2 byte 32x2 byte samples samples CIRCULAR BUFFER ……….. SAMPLES MOVED TO L0 BUFFER L0T LATENCY (FIXED) L0T reception time When a CREAM receives a L0 trigger, a configurable number of samples is extracted from the circular buffer at fixed latency and moved to the L0 buffer EVENT NUMBERS 0 1 2 3 4 5 6 7 8 9 10 11 12 data data data data data data data data data data data data data L0 BUFFER Events are univocally identified by their event number and timestamp,both reset at SOB s Stefano Venditti - CERN PH-ESE 18 CREAM L1 TRIGGER PROCESSING Once the PC farm has taken a L1 decision, a L1 request is issued to all CREAMs through Multi-Request Packets (MRP). Each CREAM sends the data corresponding to the specific event number requested, through Sub-Detector Event (SDE) packets EVENT NUMBERS 0 1 2 3 4 5 6 7 8 9 10 11 12 data data data data data data data data data data data data data L0 BUFFER MRPs from PC farm are received and decoded in the FPGA Data is extracted from L0 buffer,given a header and sent to the PC farm as UDP packet Data is available for L1 requests during the whole burst duration FPGA SDE MRP Ev. number12 0 <ETHERNET Ev. number 8 4 Ev. number 8 0 Ev. Ev.number number12 4 Stefano Venditti - CERN PH-ESE 19 CREAM L1 TRIGGER PROCESSING • The 16 CREAMs in a crate are connected to a 10 Gbit switch, that allows to receive/send packets from/to the PC farm • The CREAMs cain join multicast groups: the PC farm sends to this group only one data request, which is forwarded to all CREAMs belonging to the group at switch level. 10 Gbit switch (one per crate) Stefano Venditti - CERN PH-ESE 20 L1 PACKETS FORMAT PACKET FORMATS REQUEST TO CREAM: MRP (Multi-Request Packet) UDP packet Number of events Total MRP length header sender’s IP address Reserved Timestamp Fine time Reserved (LSB: ZS on) L1T word Reserved L1 request L0T word Event number ………….. DATA FROM CREAM: SDE (Sub-Detector Event) UDP packet Source ID Event number Reserved Event Length header Timestamp Reserved L0T word Crate and slot ID Data 31 0 Stefano Venditti - CERN PH-ESE SDE header is standard for different Source IDs (NA62 detectors), data content is custom 21 L1 PACKETS FORMAT DATA PACKET (ZS and NZS mode) MSB1: L0 readout MSB2: ZS readout #SAMPLES Data Length data header Active channels (0xffffffff in NZS mode) Sample 0 (1° active ch.) Sample 1 (1° active ch.) …. …. Sample N-1(1° active ch.) Sample N (1° active ch.) Sample 0 (2° active ch.) Sample 1 (2° active ch.) …. Sample N-1(last active ch.) LKr data …. Sample N (last active ch.) Checksum 31 0 • Empty packet sent if no channel complies with the zero-suppression threshold • Possibility to readout events at L0 (no need for MRPs, lower request rate) • Up to 256 samples/L1 request can be sent • Use of jumbo frames implemented Stefano Venditti - CERN PH-ESE 22 L1 PACKETS FORMAT L1 MECHANISM AT WORK (WIRESHARK OUTPUT) MRP size = 10 MRP SDEs MRP SDEs MRP SDEs MRP MRP from PC IP to Multicast IP Address (all CREAMs) SDE from CREAM IP to PC IP (taken from MRP) Stefano Venditti - CERN PH-ESE 23 ZERO SUPPRESSION The CREAM bandwidth can be reduced by performing zero-suppressed(ZS) L1 requests. In this mode the triggered samples of a channel are sent if at least one of its samples is above a programmable threshold. Several zero-suppression scenarios will be tested: ZS scheme DRAWBACK ZS requests only Thorough calibration needed L0-based mode Highly dependent on the L0 algorithm Double request mode Total achievable L1 rate reduced Double request mode: a first ZS request is issued to all CREAMs in multicast mode; a second NZS request is issued only to modules wich are part of specific regions of interest CREAMs sending non-empty zerosuppressed data upon first request CREAMs receiving second request and sending non zerosuppressed data 1° REQUEST (ZS) 2° REQUEST (NZS) Stefano Venditti - CERN PH-ESE 24 SPECIAL TRIGGERS The TTC broadcast signal is used to deliver L0 triggers to the acquisition system CODE TRIGGER TYPE CREAM ACTION 0b0xxxxx Physics Readout data 0b100001 Synchronization Send special frame 0b100010 Start of burst (SOB) Send special frame 0b100011 End of burst (SOB) Send EOB report 0b100100 Choke on Send special frame 0b100101 Choke off Send special frame 0b100110 Error on Send special frame 0b100111 Error off Send special frame 0b101000 Monitoring Send monitoring data 0b10110x Random Readout data 0b11xxxx Calibration Readout data CHOKE/ERROR MECHANISM: A CREAM is in a critical or in error condition (e.g. a L0 fifo almost full) CHOKE line raised on P0 LKR-TTC CHOKE received by CREAMs, packet sent L0T Processor: Send CHOKE and stop L0s Stefano Venditti - CERN PH-ESE LKR-TTC If critical state solved, lower CHOKE (L0s restarted), if not keep it up 25 TESTS @ CERN First 4 CREAMs @ CERN in March 2013 FIRST GOAL: verify that the specifications on the quality of the analog signal are met MOST IMPORTANT FEATURES TESTED: • ENOB (>10 required) • Cross-talk (<-70 dB required) • Non-linearities (Differential < 2 LSBs, Integral < 5 LSBs required) • Non-coherent noise (< 2 LSBs) and coherent noise(< 10% non-coherent noise) • Signal shape (after shaping) is 70 ns±10% FWHM pseudo-Gaussian with ±1 ns uniformity Continuous acquisition mode used in the tests Stefano Venditti - CERN PH-ESE 26 TESTS @ CERN: ENOB • 5 MHz input sine wave synchronous with the 40 MHz clock to CREAMs • 5 MHz narrow-band pass filters before CREAM input • FFT and data analysis with ROOT • ENOB of all channels well within specification 5 MHz ZOOM 5 MHz SINE CLOCK TO CREAM FFT ON CREAM DATA (65K SAMPLES) 10 MHz ZOOM 15 MHz ZOOM Stefano Venditti - CERN PH-ESE NARROW-BAND PASS FILTERS 27 TESTS @ CERN: ENOB ENOB DISTRIBUTION FOR 1 CHANNEL (1K EVENTS) ENOB - 32 CHANNELS OF A CREAM Stefano Venditti - CERN PH-ESE 28 TESTS @ CERN: CROSS-TALK • 5 MHz input sine wave, FFT of non-pulsed channels computed • Low-band pass filters used • Only cross-talk of neighbor channels is slightly above specifications (<-70 dB) • The cross-talk is on the connector, not on the PCB LOW-PASS FILTERS CROSS TALK SHAPE: points are averages of samples CROSS-TALK (NEIGHBOR CHANNEL) Stefano Venditti - CERN PH-ESE PULSED CHANNEL (~FULL SCALE) 5 COUNTS 0.5 COUNTS CROSS TALK (NEXT-TO NEIGHBOR CHANNEL) 29 TESTS @ CERN: CROSS-TALK FFT SPECTRA PULSED CHANNEL NEIGHBOR CHANNELS NEXT-TO NEIGHBOR CHANNELS Stefano Venditti - CERN PH-ESE 30 TESTS AT CERN: NON-LINEARITIES • Non linearities measured using a sine wave histogram test (Ting, Liu, IEEE Tr. On Instr. & Meas., Vol. 57, N. 2) • Uncoherent sampling, sine amplitude > dynamic range to populate all ADC values • Given the number of events and the population of the first and last bins, the theoretical distribution in case of perfect linearity is computed and compared with data i • INL and DNL well within the specifications H (i) DNL(i) a exp H theo (i) 1 INL(i) DNL(i) k 1 b c a,b) INL and DNL values (1 channel), full ADC range c) Theoretical (RED) and experimental (BLUE) distributions of sample values INL and DNL max values – 32 CREAM channels Stefano Venditti - CERN PH-ESE 31 TESTS @ CERN: NOISE • The pedestal value of each channel is measured, pedestal mean values are removed • Two groups of 16 channels (even and odd) are formed • The ditribution of the sum (difference) of the two groups → total noise σSUM (total non- coherent noise σDIFF) → coherent(σCOH) and non-coherent(σNCOH) noise/channel • σNCOH and σCOH within the specifications (σNCOH < 2 LSB, COH σCOH < 10% σNCOH) 2 2 SUM DIFF 32 NCOH DIFF 32 PEDESTALS – 16 CHANNELS CHANNEL SUM AND DIFFERENCE DISTRIBUTION Stefano Venditti - CERN PH-ESE 32 SIGNAL SHAPE AND UNIFORMITY • LKr-like signal replicated using Tektronix AFG9252 pulse generators • Trigger mode used: 8 samples extracted using sync signal from the generator, delayed by the CREAM latency • Both the signal width and its uniformity are within specifications WIDTH DISTRIBUTION (1 CHANNEL) TRIGGERED SAMPLES 0 1 2 3 4 5 6 7 8 SAMPLES Stefano Venditti - CERN PH-ESE 33 TECHNICAL AND “DRY” RUNS December 2012: a daughterboard tested on an Altera test-bench (no motherboard available at the time), 8 channels readout. Real LKr signals from particles collected. TESTED FEATURES • Shaping: LKr-signals turned into Gaussian 70 ns wide ones •Sampling: digised samples collected on disk and analysed 7 3 CELLS POSITION ON LKR SURFACE 6 2 5 1 4 0 ADC COUNTS SAMPLE INDEX Stefano Venditti - CERN PH-ESE 34 TECHNICAL AND “DRY” RUNS Summer 2013: two CREAMs were used in the NA62 “dry” run (i.e. a run without particles, mainly meant to test the readout system) TESTED FEATURES • Network: MRP reception by CREAMs, CREAM data received by the PC farm and written on disk • TTC-LKr: tested for the first time • Switch: tested for the first time (not at the full required bandwidth) Stefano Venditti - CERN PH-ESE 35 TECHNICAL AND “DRY” RUNS November/December 2013: 16 CREAMs (a whole crate) were tested at the NA62 site. TESTED FEATURES • Trigger sum links: the TELDES board correctly received and deserialized data from CREAMs • Network: multicast requests sent to all CREAMs, switch tested at the required bandwidth • L1 rate: 100 KHz rate L1 triggers sent to CREAMs, no packet loss detected TESTS STILL ONGOING! Stefano Venditti - CERN PH-ESE 36 CONCLUSIONS • Exhaustive tests of the prototypes and preproduction modules have shown full compliance with specifications •“Green light” for the production will be probably given this week NEXT STEPS •1st production batch (220 modules) delivery is foreseen in February 2014; • 2nd production batch (220 modules) delivery is foreseen in May 2014; • Production tests foreseen in March-June 2014; • LKr readout commissioning is foreseen during summer 2014; In addition, we are very pleased to mention that our measurements show that overall performance of such complex module developed by CAEN is beyond our specifications and expectations, and now we are very confident that, in spite of some delay, the project is in very good shape. In particular we would like to thank: Luca Colombini Annalisa Mati Stefano Petrucci Andrea Romboli Carlo Tintori Stefano Venditti - CERN PH-ESE 37