SISPI

1. SurveyImageSystemProcessIntegration-Realtime Software on the mountain SISPI’s job:• Coordinate camera operation (filters, shutter, readout, etc.) • Ensure smooth data flow and logging• Perform quality analysis • Provide a user interface (for monitoring and control) SISPI

2. Collect the data and deliver it, without loss, to Data Management° Focal plane (CCD) data° Instrument metadata° Status and history logging to database Perform Quality Analysis (QA), to monitor data integrity° Useful presentation of information via GUI° Software alarms (hardware alarms belong to subsystems) Perform manual or automated sequences of operations° GUI interface° Instrument calibration and configuration Maintainable by CTIO° DES only uses 30% of the telescope time. Image Builder (IB) Performance Metrics

3. Databases ObservationControl Alarms Input Status Log Image Stabilization Image Stabilization Image Stabilization Instrument Control Legend: Control paths(response required) Data paths Processes Data Subsystems Image Building & Analysis GUI SISPI definesthe interfaces Logical Connections Guide TCS Focus User commands Image Acquisition and Focus Here behumans Sky Cameras DM

4. Tasks, Milestones, & People All subsystems need an ICD. OCS draft exists. Management:JJT supervises . FRD/ICD & architecture definition (milestone) 12/06 . Estimate resource requirements 12/06 . Assign responsibilities & hire personnel 2/07 . Test complete system at FNAL (milestone) 3/09 . Install at CTIO 9/09 Infrastructure: OSU - Honscheid + postdoc . Choose message & data protocols. 12/06 . Complete focal plane emulation 6/07 . Integrate w/o CCDs (using emulators) 12/07 . Select database technology 12/06 Fermilab - Buckley-Geer . Demonstrate database functionality 1/08 Alarms (needs thought) Observation Control: OCS (Observation control system) UIUC - Qian & gs . Enumerate command sequences (use cases) 12/06 . Demonstrate OCS communications (with FCS) 7/07 . Prototype complete OCS 4/08 . OCS ready for test (milestone) 10/08 ObsTac (Observation tactics) FNAL - Annis . First version 7/07 . Ready for test 9/08 GUI UIUC - Qian & gs . List displays and controls 4/07 . Prototype display/control (of FCS) 10/07 . First complete version 5/08 . Ready for test 2/09

5. Tasks, Milestones, & People Image Acquisition & Focus FCS (aka Monsoon Supervisor Layer – MSL) UIUC - 2nd postdoc . First prototype 12/06 . Ready for test 6/07 FocusChicago - Gladders . Algorithm 9/07 . Feedback loop (milestone) 1/08 Image Stabilization GCS (aka Monsoon Supervisor Layer – MSL) UIUC - 2nd postdoc . First prototype (adapt from FCS) 3/07 . Ready for test 8/07 Guide Barcelona - Castander . Algorithm 6/07 . Demonstrate guiding with emulators 10/07 Data collection and analysis IB (Image builder) UIUC - 2nd postdoc . Specify content (metadata) 4/07 . Ready for data flow test 10/07 . Complete 12/08 QA (Quality analysis) UIUC - with DM people . Specify tests 7/07 . Software complete 7/08 ICS (Instrumentation control system) FNAL - Diehl Filter control OSU - Honscheid + gs . 1st prototype SW 6/07 . HW & SW complete (ready for test) 3/08 Focus, cryogenics, & filter control Cloud Camera FNAL - Tucker . Commission HW & SW 11/09

6. End

7. Cost Hardware Prototyping: Computers $10k Emulator 7k Installation: Network 25k Computing 36k Control room (displays, etc.) 25k Software Licenses (IDL, LabVIEW, etc.) 20k $123k Labor 3 faculty (part time) MG, KH, JJT 7 staff scientists (part time) JA, LB-G, FC, TD, IK, WM, DT 3 postdocs for 3 years Qian + two more 3 graduate students for 3 years tbd

8. 2 3 3 3 2 2 FCS & GCS DAQ electronics Fiber Ethernet PCs DHE PAN 5.3 PC farm 34 MBps 5.8 Sci. DHE PAN IB/QA/disk 5.3 + < 5.3 5.3 34 5.8 DM (not SISPI): Data staging on mountain Sci. 21 DHE PAN 5.3 + < 5.3 4 IB/QA/disk 5.3 34 5.8 Sci. DHE PAN 5.3 + < 5.3 IB/QA/disk 5.3 34 1.9 24-port switch Sci/Focus DHE PAN 5.3 + < 5.3 IB/QA/disk (the same) 24-port switch 34 < 21? 1.9 Sci/Focus DHE PAN GUI 8 0.8 Focus 1.3 Guide small Instrumentation Guide 0.01 0.8 small Control signals are not shown Database 0.01 Proposed SISPI and Front End Network Showing Bulk Data Rates (MBps)

9. Other Design Choices We do not want to reinvent the wheel. •Control protocols: Adapt the SOAR messaging system? Supported by CTIO. TCS & ICS will use LabVIEW. •Data flow: TCP/IP sockets. Client/server or publish/subscribe. •QA and GUI: Being studied. What diagnostics? What data to humans? JJT has visited CTIO & Keck control rooms. • Development: We will be able to operate a partial system. Before CCDs exist, we wll use emulator FPGAs. •Mountain storage: 3 days of data:~ 2 TB HD local to IB processors. Database (calibration, log, SkyCam, etc.) CTIO uses MySQL.

10. Other Design Choices (2) •Subsystems: Interface hardware is the responsibility of the subsystems (FCS, GCS, IC, & TCS). •Alarms: Safety is a hardware responsibility. SISPI handles other fault conditions.

11. UIUC Physics: Senior Physicists: Inga Karliner, Mats Selen, JJT Postdocs:Tao Qian, (+ another later in 2006) Students: Youcai Wang, Curtis Leung (summer) Engineers:Mike Haney, Todd Moore UIUC works on DAQ electronics and SISPI. I will only discuss SISPI (the people in green). Also: Ohio State University Klaus HonscheidFermilab Liz Buckley-Geer and Wyatt MerrittCTIO Tim Abbott UIUC and OSU will be the primary sites in the near term.Effort will ramp up over the next six months.Timely assembly of the team is our biggest risk. People

12. Glossary DCDisk Cache Local image storage DMData management Offline analysis FCSFocal plane control system Science CCD controller FDFocal plane data Science CCD data CGSGuider control system Guide CCD controller GDGuider data Guide CCD data GUIGraphical user interface Humans sit here. IBImage builder Assembles an image ICSInstrument control system Moves filters, etc. IDInstrument data Telemetry ObsTacObservation tactics Exposure sequencer OCSObservation control system Coordinates everything QAQuality analysis Real-time feedback SkyCamSky cameras Atmospheric monitoring TCSTelescope control system Tracking on the sky

13. Some Pieces Some pieces of this are described in other talks: Image Acquisition T. Shaw & V. Simaitis Focus M. Gladders & S. Kent (optics session) Guide F. CastanderTCS T. Abbott

14. Sequence commands Databases OCS Input Status Log Every process sends/receives data to/from databases. To avoid clutter, arrows are not drawn. Alarms ObsSeq Filter Focus Cryo Shutter Monsoon A GCS FCS ICS Monsoon B Legend: Control paths Data paths Processes Data Subsystems IB QA GUI More Detailed Logical Diagram Focus Guide User commands FD GD ID DC Sky Camera TCS DM

15. Control from Focus Control from Guide Control from GUI OCS Sequencedata ObsTac Control to: Image acquisition Image stabilization Instrument control Telescope control (and Image builder?) Observation Control

16. Controlfrom OCS Controlfrom OCS Monsoon B Monsoon A Guide Focus GD FD Focus controlto ICS via OCS Guide controlto TCS via OCS GCD FCD Data to IB and GUI Data to IB and GUI Image Acquisition & Focus; Stabilization The same interfaces?

17. Data from Focal plane & Instrument Control Controlfrom OCS Filter Focus Cryo Shutter DC QA ID IB QAD Data to IB and GUI Data to DM and GUI Instrument Control; Image Building

18. Rate calculation No data compression Electronics: Each ACQ card has 12 input channels. Produces 2 or 4 bytes/pixel. Each DHE crate has 2 ACQ & 1 CBB card or 3 ACQ & 2 CBB cards. Each CBB can control 9 CCDs. Science CCDs: 62 2k4k CCDs, read out @ 250 kHz, 2 video outputs. 32 MB/CCD Digitization time = 4 Mpixel * 4 µs = 17 sec Focus/WF CCDs: 8 2k2k CCDs, read out @ 250 kHz, 2 video output. 16 MB/CCD Digitization time = 2 Mpixel * 4 µs = 8.4 sec Guide CCDs: 4 2k2k CCDs, using 100 x 100 pixel ROI readout @ 1 MHz, 2 video outputs. 20 kB/CCD Digitization time = 104 pixel * 1 µs = 10 msec (+ ROI overhead) Science/Focus data rate to PAN: (on dedicated fibers) From 3-ACQ DHE: 34.3 MBps (18 CCDs) 3 PANs control 54 science CCDs. From 2-ACQ DHE: 34.3 MBps (8 CCDs) 2 PANs control 8 science and 8 focus CCDs. Science/Focus data rate out of PAN (to meet 100 sec cadence): 3-ACQ PAN: 5.76 MBps 2-ACQ PAN: 1.92 MBps Science/Focus data to IB: 3*5.76 + 2*1.92 MBps = 21.1 MBps. Perhaps /4, depending on SW protocol (push vs pull). Focus data to Focus process: 1.28 MBps Guide data rate to PAN: 8 MBps (on dedicated fibers) Guide data to Guide process: 0.8 MBps (to send in 0.1 sec) Guide data to IB: 0.08 MBps (if we keep all 100 guide images per 100 sec science image) Other data sources/sinks: Instrumentation: 100 kB per image to IB 1 MB to GUI (on observer demand) Database: 100 kB per image to IB 1 MB to GUI (on observer demand) GUI: Receives complete images (2 GB, on observer demand. Instantaneous data rate depends on the desired response time. Other information requires a small data rate. The main GUI issue is the complexity of the human/computer interface. NOTE: I assume that QA is done by the same computer as IB, and that 3-day data storage is local to each computer. This minimizes network traffic.

19. Science Exposure Prepare for exposure Exposure Telescope tracking, Filter ready, Focus adjusted, FCS ready, GCS ready, Guide ready Take a science exposure Image ready Status Data Shutter is open Not guiding Shutter is closed Guiding Commands and messagesreceived by OCS time Commands and messagessent by OCS Send status Slew telescope, Change filter, Adjust focus Prepare FCS, Prepare GCS, Prepare Guide Open shutter Start guiding Build image Exposure duration is controlled locally at the shutter. OCS Command Sequence We must specify similar sequences for every situation.

20. Division of Labor This assignment of responsibilities is not yet final.Names are of some who are helping with the design. AlarmsFermilab, OSU Merritt, HonscheidDBFermilab, OSU Merritt, Honscheid, Buckley-Geer FCS UIUC Karliner, ThalerFocus Chicago Gladders, AbbottGCS UIUC Karliner, ThalerGUIFermilab, OSU Merritt, Honscheid, Thaler GuideChicago, Barcelona CastanderIBFermilab, UIUC Mohr, StoughtonICSFermilab, UIUC Alspach, HaneyObsTACFermilab AnnisOCS NOAO AbbotQA Fermilab, NOAO, UIUC Lin, Rest, MohrSkyCamFermilab Tucker

21. Focal Plane Simulator (FPS) •Simulate the response of the front end electronics. . Exercise the HW & SW before CCD’s are available.. Evaluate proposed upgrade or other focal plane modifications. •Implement with field programmable gate arrays (FPGA’s).. Each FPS element will behave like a CCD.. Accept clock signals.. Generate “fake analog” Vout with programmable patterns. •Cost to simulate the whole focal plane ~ $7k. .A prototype will exist this summer. It is important to be able to test the partial system.