Run Number 19000 20000 21000 22000 23000 24000 DAQ. Livetime Ratio 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 Run 62 Run 63 Run 64 - 65 Run 69 24kW 27kW 27kW 39 - 41kW 41kW Run Number 19000 20000 21000 22000 23000 24000 Count rate/spill [kHz] 0 5 10 15 20 25 30 35 Level1 Requested Rate Level1 Accepted Rate Level2 Accepted Rate RCE DPM CI DTM IPMC RTM Goal: Search for K L π 0 νν - Rare decay: SM BR ~ 3 x 10 -11 - Small theoretical uncertainty: 1~2% Sensitive to the SM - 2nd order FCNC Sensitive to new physics beyond SM FB NCC MB CV CsI calorimeter CC03 OEV CC04 CC05 CC06 BHPV LCV BCV HINEMOS BH*& %30 1HZ %+&9 BP&V BBBB BBBB _ _ 2OG %+&9 _ _ _ Detect: 2γ on the CsI calorimeter + nothing else — Photon Veto — Neutron Counter — Charger Particle Veto — CsI Photon Detector Upgrades of the Data Acquisition System for the KOTO Experiment Stephanie Su, University of Michigan, U.S.A. for the KOTO Collaboration KOTO Experiment at J-PARC K L π 0—> γγ ADC ‣ Shape waveform signals using 10-pole Bessel Filter ‣ Digitize detector waveforms using 125 MHz and 500 MHz 14-bit ADC ‣ Compress data using lossless compression algorithm ‣ Store waveform information in the pipeline and wait for L1 trigger decision Data Acquisition System L1 Trigger System ‣ Make L1 trigger decision every 8 ns ‣ Use VME daisy-chain backplane to sum the energy and hit information ‣ L1 trigger requirement: CsI & !(Veto detectors) L2 Trigger System ‣ Make L2 trigger decision using Center of Energy (CoE) on the CsI calorimeter ‣ During a spill: - Store entire spill of data onto 2 Gbit onboard memory - Read out data from the other 2 Gbit onboard memory to L3 trigger system L3 Trigger System ‣ Build events using Infiniband ‣ Each Type I Node: - Receives event fragments from each L2 trigger module - Sends event fragments to Type II nodes via Infiniband (event ID & spill ID) ‣ Each Type II Node: - Builds a complete event - Decompresses and transposes the data to analyzable format - Recompresses events for storage ‣ Transfer files from each computer node to disk arrays, then permanent storage at KEK Conclusion 125 MHz ADC 500 MHz ADC L3 Trigger System Cluster-On-Board (COB) developed by SLAC 10-pole Bessel Filter Σ Et and Σ Hit Energy Detector 125 MHz / 500 MHz Waveform Digitizer ADC Data Compression 2 Gb DDR2 Memory L1 Trigger Module FPGA FPGA L2 Trigger Module Input FIFO Buffer FPGA L3 Trigger Type II Nodes L2 Trigger Write Read Disk Arrays L1 Trigger Master Sum by Daisy-Chain Sum all Tape Storage at KEK L3 Trigger Type I Nodes 20 Gbps Infiniband Detector ADC FPGA Σ Et and Σ Hit Energy Transpose Data 125 MHz / 500 MHz Waveform Digitizer 10-pole Bessel Filter Append MAC Address for Event Building L2 Trigger Module (COB) L1 Trigger L3 Trigger Type I Nodes L3 Trigger Type I Nodes … L3 Trigger Type II Nodes L3 Trigger Type II Nodes … 2 Gb DDR2 Memory L3 Nodes L3 Nodes L3 Nodes … Disk Arrays Tape Storage at KEK 2.5 Gbps Optical Link 1 Gbps Ethernet Link L1 Trigger Buffer Full Σ CoE Sum by Daisy-Chain Suspend L1 Trigger L2 Trigger Master Buffer Full 2.5 Gbps Optical Link 1 Gbps Ethernet Link 20 Gbps Infiniband Pipeline Pipeline FPGA L2 Trigger Cut (COE, Clustering) RCE Input FIFO Buffer 1 GB DDR3 Memory 10 Gbps links x 18 x 279 x 279 x 6 x 8 2015 - 2016 Runs Upgrades Trigger Rates for 2015-2016 Runs DAQ Livetime for 2015-2016 Runs ;Purpose of ‣ To accommodate higher beam intensity - Current bottleneck: 50 kW ‣ To enhance DAQ performance and lifetime Changes on the ADC ‣ Secure data quality - Bit checking using checksum Upgrades of the L2 Trigger System ‣ Event Building on L2 trigger system - Full backplane connectivity ‣ High performance Cluster-on-Board (COB) - Reconfigurable Clustering Element (RCE) • FPGA (Zynq 7030, Zynq 7045) with ARM processors • 1 GByte DDR3 memory • On-board Linux and RTEMS Operating Systems ‣ High-speed 10 Gbps links between components ‣ New L2 trigger cut (ex. CoE, Cluster counting) Upgrades of the L3 Trigger System ‣ Event categorization using trigger tag information from MACTRIS ‣ Developing analysis technique for L3 trigger cut Purpose of the Upgrades The current KOTO DAQ system successfully collected compressed and uncompressed ADC data since 2015, at beam powers of 24 kW to 42 kW. Upgrades of the hardware is in progress in order to improve DAQ performance with anticipation of increasing beam power. — with compression — no compression