This meeting qualifies as a requirements review and conceptual design.
Attendees:
diane, sheng, terri, hamid, antonio, chris, amedeo
Minutes:
Our current situation: the e-beam 120Hz capable multicast network and data will not be installed and available until ~January 2010. We will need both a short-term solution to providing data to PDCS, and a long-term plan.
We discussed the requirements for this interface between the e-beam and PDCS, both short-term and long-term.
REQUIREMENTS
Data Required short-term:
- Beam charge
- Beam energy
- Beam position and angle (in x and y) at the LTU *The data must be timestamped with the 17-bit pulseid (original source from an IOC w/EVR)
Data Requirements long-term:
- data from e-beam will be 120Hz data produced by the charge, energy and position/angle feedbacks *data will be timestamped *Only data requirements listed above so far, but we will need to be flexible as PDCS physicists will likely add more...
Other Requirements short-term:
- Data should arrive at PDCS within the pulse interval (ideally before the next pulse), highest expected rate 30Hz.
- The packet can arrive within about 3 pulses. PCDS can use a timeout to adjust.
- Can drop packet once every 3 days
- Data from e-beam at best will be rough estimates of above, using simple calculations on timestamped data.
- All e-beam data will be sent in one packet (& one group ID) using the PDCS format described in Chris O=Grady's ICD.
- Commissioned in mid July.
- Experiment runs in mid-August
Other Requirements long-term:
- The goal is to have data arrive at PDCS within the pulse interval, 8.3ms for 120Hz beam.
- packet format may change as data requirements change
We discussed the various proposed approaches to design and agreed upon the following implementation plan:
CONCEPTUAL DESIGN
A 'concentrator' IOC will be developed. This IOC will have an network interface to the e-beam Channel Access network, and a separate interface to the LCLSBLD subnet for nulticast to the PDCS multicast network.
- the concentrator will collect the required e-beam data from e-beam IOCs via channel access.
- the concentrator will calculate the charge, energy, position and angle and *the concentrator will create the PDCS data packet and send on PDCS network
- if the channel access approach does not meet latency requirements then the e-beam IOCs must send unicast UDP messages to the concentrator.
- PCDS has tested RTEMS code that can be used as example for UDP approach and data packet formation
- Long-term: the Channel Access interface will be replaced with an interface to the e-beam multicast network. The rough calculations will be removed, the e-beam feedback data will be placed in the PDCS data packet and sent.
ACTION ITEMS
- Diane/Hamid - identify and engineer to do this work.
- Diane - prepare list of rough estimate calculations and review with Emma, Iverson
- Diane - send list of calculations to group and John Bozek
- Terri - ask Ernest to identify an existing or new crate.
- Sheng - provide IOC
- Chris - send Tony's code (example) that PCDS is adding to their IOCs.
- Antonio/Terri - drawing of network this week
- Antonio: implement next week
- Engineer?: implement IOC with CA high priority client, do calculation, create 1 packet, and multicast from ebeam meas group.
- Engineer?: convert data collection to non-EPICS if needed. Use code/idea created by PCDS to UDP scalar and timestamp to concentrator IOC.
- PCDS and ebeam: test latency; PCDS will create a way to measure