CollimatorMtgs-7-31-2013
From GlueXWiki
Hall D Collimator Meeting Minutes (Wednesday, July 31, 2013)
Minutes courtesy of Todd Satogata
Contents
Attendance
- University of Connecticut: Richard Jones, Alex Barnes, James McIntyre
- Jefferson Lab: Alex Somov, Todd Satogata, and Trent Alison
Topics
- Richard reviewed data in the presentations linked below.
- In the first, the prototype active collimator was installed in the Hall B alcove and tested in late March 2011. Run conditions are on page 7, and photos of the installation are on page 8. Only horizontal sections were instrumented (right photo on p. 8), and the collimator was installed on a horizontal motion stage. The data shown on page 9 -- note that the photon collimator has a 5mm diameter aperture for an expected ~5mm rms photon spot size, so the aperture is about 1/2 of a horizontal (1cm) tick centered around x=14.2cm where the signals are equal and very nearly linear over a +/-5mm range. Note that there is a baseline signal, and enough current between inner and outer wedges that a simple difference over sum calculation does not adequately linearly correlate to position. There was some discussion of signal contamination by front-facing signal connections and cables; the connections cannot be moved to the back of the collimator, and the L joints help keep the cables away from the central collimator area where position readings will be important.
- The second presentation shows collimator requirements (page 5) and design (page 6). Page 9 shows raw data, after baseline subtraction and what appears to be a cut of negative signals. Page 10 shows difference over sum vs stepper motor position (without raw offsets removed), showing structure through the aperture that is unacceptably nonlinear. We also discussed the relatively low noise of the system (pages 11-12) and suspect that this noise floor is acceptable for fast feedback, but this will need to be checked in light of the feedback gain and resolution and final position calculation algorithm.
- Trent summarized conclusions from a meeting on July 12 with Todd and Alex. We decided to use CEBAF standard VME setup with existing 4-channel VME3115 16-bit ADC cards, one for each plane, for operability and to eliminate new board layout and spare inventory costs. There is some FPGA work to be done to pass through 16 bits of digitized data, including data snapshots that could be accessed though EPICS. Trent also thinks that a simple position calculation may fit on the FPGA. Signals can be converted from TTL to fiber on the front of the board with some work, and will need to be established to communicate fiber positions to the fast feedback system. Overall system latency is also an open concern.
- Discussion of possible position algorithms suggested that difference of inner signals divided by the sum of all four signals may provide acceptable position correlation. We also concluded that a position calculation status bit would be useful; data from the second slide suggests that thresholds on all four signals (above threshold on inner wedges and below threshold on outer wedges) may provide a simple acceptable status.
Action Items
- Richard will share raw (ASCII) four-channel data with Trent and Todd for evaluation of different position algorithms.
- Trent will evaluate some algorithms to determine whether they can be performed on the VME3115 (old) FPGA. If not, they will need to be calculated on the EPICS processor from raw data piped on the VME backplane.
- Another meeting will be called if needed after Trent and Todd have a look at UConn raw active collimator data.
