310 likes | 460 Views
Virgo Data Acquisition D. Verkindt, LAPP. DAQ Purpose DAQ Architecture Data Acquisition examples. Connection to DAQ and monitoring tools Data Streams Online analysis tools. DAQ purpose. DAQ requirements: collection of distributed data (timing system, optical links)
E N D
Virgo Data AcquisitionD. Verkindt, LAPP • DAQ Purpose • DAQ Architecture • Data Acquisition examples • Connection to DAQ and monitoring tools • Data Streams • Online analysis tools GREX 2002, Pisa D. Verkindt, LAPP
DAQ purpose • DAQ requirements: • collection of distributed data (timing system, optical links) • flexibility in data flow (frame format) • reliability (at least 1 month without crash) • easyness of use and restart (DAQ graphical client) Laser GREX 2002, Pisa D. Verkindt, LAPP
North Building Control Building Central Building DAQ purpose Get data from various synchronized sources, sometimes 3 km away GREX 2002, Pisa D. Verkindt, LAPP
Environment Controls Detection Output MC Bench Detection Bench Env. monitoring Data Acquisition Suspension control Env. monitoring DAQ purpose • Collect distributed data from: • ITF environment • ITF controls • ITF output detection GREX 2002, Pisa D. Verkindt, LAPP
Input bench monitoring, Vacuum monitoring, Environment monitoring Suspensions data Locking and alignment data Photodiodes data det. Bench monitoring frames frames frames local data collector local data collector local data collector 2.7 MB/s 3.3 MB/s 3.0 MB/s Central data collection 9 MB/s (compressed = 4MB/s) DAQ architecture Environment Monitoring Controls Detection GREX 2002, Pisa D. Verkindt, LAPP
DAQ architecture Environment Monitoring Controls Detection Global Ctrl Susp. Ctrl Alignement Photodiodes Det. Bench Ctrl Laser + env + towers + tubes + calib.itf DOL DOL > 30 VME crates DOL 1 Gx 6 Gx FbS FbF FbS 4 Fbf FbF FbF 3 Gx FbS frames frames frames local Main Frame Builder local Main Frame Builder local Main Frame Builder 2.7 MB/s 3.3 MB/s 3.0 MB/s Central Main Frame Builder 9 MB/s (compressed = 4MB/s) GREX 2002, Pisa D. Verkindt, LAPP
DAQ architecture • More than 30 VME crates, but a reduced set of standard tools: • Digital Optical Links (DOL) for controls • Fast Ethernet and Gbit Ethernet for central data collection • VME crates for front-end data acquisition • Workstations for central data collection • Standard format for data collection : frames encapsulated in Ethernet messages GREX 2002, Pisa D. Verkindt, LAPP
Time GW signal channel 1 channel 2 channel 3 channel n frame 1 frame 2 ... Frame format Frame = elementary time slice of data • Contains: • GPS time stamp • ITF informations • raw data channels • processed data • events Common format of several gravitational waves detectors GREX 2002, Pisa D. Verkindt, LAPP
CPU Thanks to A. Masserot GPS Generator & Distributor Part Timing Information (Cm) GPS Timing Timing Timing Distributor Crate User’s Timing Crates Timing Distributor Crate User’s Timing Crates Timing Distributor Crate User’s Timing Crates TTL/OPT Fast Clock Timing Distributor Crate User’s Timing Crates OPT/TTL Fast Clock OPT/TTL Fast Clock OPT/TTL Fast Clock OPT/TTL Fast Clock OPT/TTL Run OPT/TTL Run OPT/TTL Run OPT/TTL Run GPS TTL/OPT Sampling OPT/TTL Sampling OPT/TTL Sampling OPT/TTL Sampling OPT/TTL Sampling Environment Controls Detection TTL/OPT Frame OPT/TTL Frame OPT/TTL Frame OPT/TTL Frame OPT/TTL Frame TTL/OPT Run Optical fibers Coax Cables Coax Cables Coax Cables Coax Cables OPT/TTL Sampling TTL/OPT Frame Sampling TTL/OPT Frame Sampling TTL/OPT Frame Sampling TTL/OPT Frame Sampling Laser OPT/TTL Frame Monitoring & Control Part Build. Return Timing • all VME crates synchronized by Master clock • Fast Clock (2.5 Mhz) • Sampling (20 kHz) • Frame (1 Hz) CPU CPU CPU CPU CPU CPU CPU CPU CPU CPU CPU CPU CPU CPU CPU CPU Timing Timing Timing Timing Timing Timing Timing Timing Timing Timing Timing Timing Timing Timing Timing Timing ADC,DAC Camera,DOL ADC,DAC Camera,DOL ADC,DAC Camera,DOL ADC,DAC Camera,DOL ADC,DAC Camera,DOL ADC,DAC Camera,DOL ADC,DAC Camera,DOL ADC,DAC Camera,DOL ADC,DAC Camera,DOL ADC,DAC Camera,DOL ADC,DAC Camera,DOL ADC,DAC Camera,DOL ADC,DAC Camera,DOL ADC,DAC Camera,DOL ADC,DAC Camera,DOL ADC,DAC Camera,DOL Return GPS Data Acquisition Timing system overview • Purpose • Synchronization (of controls) • Frame and sampling numbers • GPS time stamp for data exchange GREX 2002, Pisa D. Verkindt, LAPP
Sensor (temp. pressure…) Slow Monitoring Stations query SMS data Main Frame Builder frames Slow Frame Builder timing info GPS Timing timing info Data acquisition examples GREX 2002, Pisa D. Verkindt, LAPP
BNC cables Eth. 100 Mbps frames accelerometers, microphones, … Fast Frame Builder Main Frame Builder timing signals GPS Timing timing info Data acquisition examples GREX 2002, Pisa D. Verkindt, LAPP
Optical line Interferometer controls (DOL) Pre-ampli , demodulation& filtering Optical line (DOL) Eth. 100 Mbps Photodiode Readout Photodiode Fast Frame Builder frames Main Frame Builder timing info timing signals GPS Timing Data acquisition examples GREX 2002, Pisa D. Verkindt, LAPP
Dynamical connection • connect: send request with list of channels • disconnect: automatic • minimal perturbation on main stream. requested data Monitoring world request dataDisplay Data Storage Connection to DAQ • Main Frame Builder: • Use shared memory and 2 processes • Producer: merge input frames and put result in shared memory • Consumer: read frames in shared memory and send them on network Fast Frame Builders Slow Frame Builder Main Frame Builder Producer Shared Memory Consumer 2 Consumer 1 Main data stream DAQ world Central Main Frame Builder Online Processing GREX 2002, Pisa D. Verkindt, LAPP
Online Monitoring using dataDisplay GREX 2002, Pisa D. Verkindt, LAPP
Offline use of dataDisplay GREX 2002, Pisa D. Verkindt, LAPP
DAQ control and monitoring GREX 2002, Pisa D. Verkindt, LAPP
Web DAQ Monitoring GREX 2002, Pisa D. Verkindt, LAPP
nADC nBytes DAQ Performances • Run almost continuously since Sept. 2001 • DAQ efficiency during last engineering runs > 99.8% • Minimized latency --> DAQ can be used for online control GREX 2002, Pisa D. Verkindt, LAPP
Raw data frames: • most of channels sampled at 20 kHz or 10 kHz • frame = 1 sec of raw data = 4MB (day=345 GB year=120 TB) • 50Hz data frames: 3% of raw data storage • provide fast access to raw data in low frequency band • resampling at 50Hz (with filtering) all the fast data channels • frame = 10 sec of resampled data = 1.1 MB (day=9 GB year=3300 GB) • Trend data frames: 0.1% of raw data storage • provide fast access to long (weeks) stretch of data • trend data = min, max, mean, rms computed for each fast sampled channel, over one frame • frame = 30mn of trend data = 9.6 MB (day=460 MB year=170 GB) Current data streams GREX 2002, Pisa D. Verkindt, LAPP
Trend Frame Builder Web Trend Frames Disks Vega DB (Root) Trend data acquisition Env. Moni Frame Builder Controls Frame Builder Detection Frame Builder Main Frame Builder Full Frame Storage (disks) GREX 2002, Pisa D. Verkindt, LAPP
Example 1 : output of ITF over 8 hours, during engineering run E4 (min, max, mean) Online Monitoring using trend data Use of Vega tool and Web browser GREX 2002, Pisa D. Verkindt, LAPP
Offline use of trend data Example 2 : max of output of ITF, building temp. and seismic motion near north tower over 3 days GREX 2002, Pisa D. Verkindt, LAPP
50Hz processing 50Hz processing 50Hz processing Trend Frame Builder 50 Hz Frame Builder Web Trend Frames Disks 50Hz Frames Disks Vega DB (Root) 50Hz data acquisition Env. Moni Frame Builder Controls Frame Builder Detection Frame Builder Main Frame Builder Full Frame Storage (disks) GREX 2002, Pisa D. Verkindt, LAPP
Online monitoring using 50 Hz data Example 1 : monitoring of seismic activity over 8 hours, in 3 frequency bands GREX 2002, Pisa D. Verkindt, LAPP
Offline use of 50 Hz data Example 2 : spectral density of output of ITF over 3 hours of data (made in 30 sec) GREX 2002, Pisa D. Verkindt, LAPP
Online analysis tools • GAI (General Algorithm Interface): • A software tool to interface algorithms to online processing stream of data. • Used to run online algorithms during engineering runs • Used also offline to analyse engineering runs data • Improved thanks to requests and comments from users and algorithm developers • Some of the algorithms developed up to now with GAI for online and offline analysis: • Algorithm 1 : monitoring of spectral lines in ITF output channels • Algorithm 2 : search of glitches in ITF output channels • Algorithm 3 : monitoring of the stationarity and gaussianity of the ITF output. GREX 2002, Pisa D. Verkindt, LAPP
Online analysis tools Ethernet Disk Shared Memory frames GAI process Algorithm Gai library frames Shared Memory Disk Ethernet GREX 2002, Pisa D. Verkindt, LAPP
Online analysis tools Ethernet Algorithm1 Disk Algorithm4 Shared Mem Shared Mem Algorithm2 Algorithm3 Ethernet Algorithm5 Disk : data under frame format GREX 2002, Pisa D. Verkindt, LAPP
Algorithm 1 Algorithm 2 Algorithm 3 Algo3 data storage Algo1 data storage Algo2 data storage Online Analysis: current scheme Online Processing Frame Distributor Main Frame Builder frames raw data frames Full Frame Storage (disks) GREX 2002, Pisa D. Verkindt, LAPP
Algorithm 1 Algorithm 2 frames Algorithm 3 Processed data storage Online Analysis: futur scheme Main Frame Builder Trigger manager frames raw data frames Full Frame Storage (disks) GREX 2002, Pisa D. Verkindt, LAPP
Conclusion Virgo DAQ and online monitoring tools like dataDisplay or Vega+Web have been extensively used since year 2001. • DAQ has shown to be: • modular (lego pieces with standard connections between them) • reliable and quite easy to use (and to restart) • flexible and evolutive • latency minimized • Beyond DAQ: • Useful data streams (raw data, trend data, 50Hz data, processed data, …) are under definition • Online analysis has started GREX 2002, Pisa D. Verkindt, LAPP