Private:Spring 2011 Beam Test Manual

The following note should prepare a person on shift operating the Spring 2011 Beam Test in the Hall B of Jefferson Lab for all the technical tasks required.

= General Notes =

General Info


A single password has been made for all temporary computer accounts (described for each setup below). It is hidden in the adjacent image. This goes for the beam test accounts on the two Windows machines serving the active collimator and the microscope prototype and VNC passwords.

Access to JLab computers


Computers in JLab cannot be accessed directly off-site. One must always:
 * 1) log in via SSH to login.jlab.org
 * 2) if you don't have your own, use bpratt with password shown on the adjacent image.
 * 3) from there, log in to the desired computer or...
 * 4) use the existing SSH session for "tunneling" i.e. patching through Remote Desktop of VNC connections. Quick port reference:
 * 5) * Remote Desktop port: 3389
 * 6) * VNC port: 5900 (Note that the Windows firewall may not be configured for this, but "loop-back" is allowed in VNC service so 5900 can be tunneled out)

Likewise data cannot be copied directly from the computers in JLab. One must ssh into the desired machine and push data to UConn.

= Active Collimator =
 * Logbook: Google Doc:Beamtest-Spring-2011_ActiveCollimator view, edit

DAQ Computer
Below is the information on the data acquisition (DAQ) computer set up to configure and collect data from the active collimator:
 * Name (local domain): GUMMO
 * Network name: clonwin4.jlab.org
 * User name: beamtest (to specify domain explicitly: GUMMO\beamtest)

Technical Issues
Due to some peculiarities of Windows 7 (or lack of understanding of its "features") here are some specific issues encountered and their workarounds:
 * Cygwin's sshd could not be installed as a service - we did not seem to have sufficient privileges. sshd is run instead on startup using registry entry in HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\Run Unfortunately it doesn't even do this (as it would in previous Windows versions) until the first user has logged in. Note a CMD prompt that appears briefly upon first login.
 * The installed UltraVNC service seems to respect Windows' edict of "one computer - one user" and does not bring up the current screen when accessing with a client but offers a login a kicks off the current user: same behavior as Remote Desktop, defeating the point of this alternative. But, launching WinVNC daemon allows the usual behavior of several users sharing one screen. WinVNC should run at startup in the same way as sshd described above, but this may need to be checked. A shortcut has been added to the desktop to (re)start WinVNC.

Organization

 * directory on DAQ computer: /c/work/GlueX/Collimator
 * Note that C: drive had only 55GB free at the time of installation and 30GB as of the start of beam test due to baseline noise data sets. Use "data" directory under the above path which leads to drive F:
 * space allocated on the UConn cluster: /scratch/BeamTest-3-2011/ActiveCollimator

Keeping track of data

 * Waveform traces consume ~4.2MB/s!! (counting acquired time window only) DAQ can quickly overrun a disk if left unattended.
 * Please keep track of space consumption using: df -h

Operation


The following settings on the DAQ "virtual instrument" should be checked during a run:
 * 1) board number = 0 This is not the default. "Invalid board number" error will occur otherwise (see #11)
 * 2) Number of channels (Nchan) = 4
 * 3) Channels read must be "1 - 4"
 * 4) Range should be as low as possible (for maximal dynamic range) without clipping the waveforms Apparent clipping on the plots below do not necessarily indicate clipping. Clipping can be verified by clicking "Reset range" to the right of the plots so that their voltage range corresponds with range set on the DAQ. Make sure there is no clipping! Set a wider range as necessary, but gain setting should not be altered without consultation.
 * 5) During regular run the acquisition should be free-triggered. Other trigger settings are irrelevant if "free" is set
 * 6) Set "interval" (time between acquisitions in "Run" mode) carefully: it should be about twice the duration of the "time window" or more to allow ample time for the system to save traces. The general rule is that the acquisition LED (under "interval") should turn off between acquisitions.
 * 7) Note: "Acquire" orders a one shot acquisition, "Run" begins batch mode acquisition, repeating every "interval" set on the left.
 * 8) "Close" does not actually close the window but is recommended before shutting down the program for safe termination of analog input.
 * 9) To avoid the overhead involved in FFT calculation, turn this feature off during data acquisition unless online tracking of some harmonics is desired
 * 10) When setting up data recording, ensure the following
 * 11) * The correct path and history file name. Please make sure the file name is descriptive of the run number/name.
 * 12) * Prefix of the files in which traces are saved matches in some way the history file name to understand how the files are associated (e.g. as shown: run0.dat with trace files trc0_#.dat to indicate that run=0)
 * 13) * Both check boxes are checked (without the latter, only the history with DC values will be saved and not the traces)
 * 14) Ensure that the error panel is clear of any errors and that the "Actual rate" is reasonable: it should be about 333kHz/Nchan

= Tagger Microscope Prototype =
 * Logbook: Google Doc:Beamtest-Spring-2011_proto</tt> view, edit

Control Computer
Here is the information on the Windows PC set up for controlling the prototype's alignment motors and configuring/monitoring the electronics.
 * Name (local domain): clonwin3
 * Network name: clonwin3.jlab.org
 * User name: proto

Startup checklist

 * 1) Install the equipment (with SiPM bias still off)
 * 2) Find the plane of scattering
 * 3) Put detector into alignment (plumb bob etc.)
 * 4) Find signals in the trigger scintillators, equalize gains
 * 5) Find the accelerator R.F. in TDC, verify <100ps &sigma;
 * 6) Bias SiPMs, check noise on outputs
 * 7) Take spectra, note any correlations between signals
 * 8) Optimize elevation and orientation