Difference between revisions of "20181024-Analyzer-Mtg"

From PREX Wiki
Jump to navigationJump to search
 
(One intermediate revision by the same user not shown)
Line 13: Line 13:
  
 
== Agenda ==
 
== Agenda ==
 +
# Cavity BPMs
 +
#* CG: John Musson, Ye, and Cirprian looked at Cavity BPM signals on a scope last Friday.
 +
#* CG: It looks like the receiver box's output DAC is not configured correctly, and so a test of this firmware is being done by Trent and Musson.
 +
#* CG: As a backup, it may be possible to bypass the DAC conversion by using a fiber optic cable to take the original signal instead.
 +
#** KP: We should ask them to give us a good analog signal.
 +
#** Bob: One goal is for SEE BPM cross-checking, but another is for low-current EPICs variable data collection (already implemented), in which case an already-digital readout is desirable, particularly for commissioning.
 +
#** Ye Tian will work with Ciprian and Musson to follow through on this.
 +
 
# et_bridge
 
# et_bridge
 +
#* Carl Timmer helped Bob recompile and get it working ([https://userweb.jlab.org/~rom/TwoET.html Bob's notes]).
 +
#** Bob: Memory mapped file deletion was a problem, and has been fixed, so the 2-ET system is working fine now.
 
#* Cameron used it and tested it for deadtime [[:media:2018-10-24_Cameron-ET-deadtime.pdf|here]].
 
#* Cameron used it and tested it for deadtime [[:media:2018-10-24_Cameron-ET-deadtime.pdf|here]].
 +
#** Bob: Why is there deadtime at all? 1kB size events should not be a problem. Buffering of events in series may help.
 +
#** KP: Why measure deadtime like this? PK: Implement a counter in the VXWorks ROC software, and see if there are any skips there. Bob: Implement a scaler in the crate that will count the number of helicity flips - also it would be good to recreate the helicity information to see that no gaps appeared.
 +
#** KP: Each data buffer has an index associated with it that can be tracked to see how many buffers are lost.
 +
#** PK: Techniques exist for looking at ET's efficiency of moving data around, and ways exist for interrogating the ROCs buffers to find the bottleneck. Bob: Putting events back will allow us to use the ET more dynamically and make a copy of all events to file (to compare stored vs. analyzed).
 +
#** Bob: Check that this wasn't accidentally run in blocking station mode (which would make it slower).
 +
#** PK: Set the ET buffer to a larger number and see if that smooths out network traffic problems.
 +
 +
# WAC: online analysis scripts
 +
#* Ciprian will look at online analysis scripts, and possibly ask Cameron to help as well
 +
 
# inclusion of HAPPEX ADCs
 
# inclusion of HAPPEX ADCs
−
#*  
+
#* Bob: Getting mnemonic device names to work is proving non-trivial, but progress is being made.
 +
#** CG: [https://hallaweb.jlab.org/parity/prex/japan/Doxygen/html/ Doxygen] is hosted on Hall A Web and can be used to track class inheritance, etc.
 +
 
 
# naming conventions:
 
# naming conventions:
−
#* See Cameron's commit and comment on issue [https://github.com/JeffersonLab/japan/issues/22 22].
+
#* See Cameron's commit and comment on [https://github.com/JeffersonLab/japan/issues/22 issue 22].
 +
#* PK: Don't use a placeholder "ble" class name (the 4th word in the .map files is the japan class name).
 +
#** PK: You can just skip unknown channels and not include them in the map.
 +
#* PK: The BCM class is possibly not working, but should be used as the general default for getting a signal decoded.
 +
#* KP: What 4th word class name should be used? PK: Will check later in the code - "bcm" is generic, but "ble" should be replaced with either an empty data type or the general type of bcm class
 +
#** PK: The 4th word is just used by the .map file to indicate the class to be used, the 5th word should be "bpm", "cav", "bcm" + _ "channel name indicator"
 +
#** KP: 5th word variable names - underscores? Changing cm_us and cm_ds to bcm1 and bcm2 should be sufficient.
 +
#** PK: For example: the SEE BPMs 4 x, y, p, m channels are parsed (the trailing XP, etc. are removed) and new parallel branches are calculated and stored - leaves are sub-blocks and hardware sums. Similarly for Cavity BPMs (cavity class hasn't been written yet, but should be temporarily replaced with 3 "bcm" instances).
 +
 
 
# cavity BPM and configuration files (japan and pan): Ciprian, Bob, Cameron
 
# cavity BPM and configuration files (japan and pan): Ciprian, Bob, Cameron
−
#* BPM pedestal calibration?
+
#* Bob fixed the control.db in
 +
# BPM pedestal calibration
  
 
== Present ==
 
== Present ==
 +
Cameron, Bob, Paul King, Tao Ye, Ciprian
  
 
===Excused===
 
===Excused===
−
 
+
Ciprian
  
 
[[Category:Meetings]]
 
[[Category:Meetings]]
 
[[Category:Analyzer_Meeting]]
 
[[Category:Analyzer_Meeting]]

Latest revision as of 10:14, 24 October 2018

Back to Main Page >> Analyzer_Meeting

previous meeting << >> following meeting

Logistic information

 BlueJeans calling instructions:
 Toll-Free Number (U.S.&  Canada):  888-240-2560
 International toll number:         408-740-7256
 Bluejeans CODE:                  684 884 269
 Bluejeans link: https://bluejeans.com/684884269

github repository: JAPAN

Agenda

  1. Cavity BPMs
    • CG: John Musson, Ye, and Cirprian looked at Cavity BPM signals on a scope last Friday.
    • CG: It looks like the receiver box's output DAC is not configured correctly, and so a test of this firmware is being done by Trent and Musson.
    • CG: As a backup, it may be possible to bypass the DAC conversion by using a fiber optic cable to take the original signal instead.
      • KP: We should ask them to give us a good analog signal.
      • Bob: One goal is for SEE BPM cross-checking, but another is for low-current EPICs variable data collection (already implemented), in which case an already-digital readout is desirable, particularly for commissioning.
      • Ye Tian will work with Ciprian and Musson to follow through on this.
  1. et_bridge
    • Carl Timmer helped Bob recompile and get it working (Bob's notes).
      • Bob: Memory mapped file deletion was a problem, and has been fixed, so the 2-ET system is working fine now.
    • Cameron used it and tested it for deadtime here.
      • Bob: Why is there deadtime at all? 1kB size events should not be a problem. Buffering of events in series may help.
      • KP: Why measure deadtime like this? PK: Implement a counter in the VXWorks ROC software, and see if there are any skips there. Bob: Implement a scaler in the crate that will count the number of helicity flips - also it would be good to recreate the helicity information to see that no gaps appeared.
      • KP: Each data buffer has an index associated with it that can be tracked to see how many buffers are lost.
      • PK: Techniques exist for looking at ET's efficiency of moving data around, and ways exist for interrogating the ROCs buffers to find the bottleneck. Bob: Putting events back will allow us to use the ET more dynamically and make a copy of all events to file (to compare stored vs. analyzed).
      • Bob: Check that this wasn't accidentally run in blocking station mode (which would make it slower).
      • PK: Set the ET buffer to a larger number and see if that smooths out network traffic problems.
  1. WAC: online analysis scripts
    • Ciprian will look at online analysis scripts, and possibly ask Cameron to help as well
  1. inclusion of HAPPEX ADCs
    • Bob: Getting mnemonic device names to work is proving non-trivial, but progress is being made.
      • CG: Doxygen is hosted on Hall A Web and can be used to track class inheritance, etc.
  1. naming conventions:
    • See Cameron's commit and comment on issue 22.
    • PK: Don't use a placeholder "ble" class name (the 4th word in the .map files is the japan class name).
      • PK: You can just skip unknown channels and not include them in the map.
    • PK: The BCM class is possibly not working, but should be used as the general default for getting a signal decoded.
    • KP: What 4th word class name should be used? PK: Will check later in the code - "bcm" is generic, but "ble" should be replaced with either an empty data type or the general type of bcm class
      • PK: The 4th word is just used by the .map file to indicate the class to be used, the 5th word should be "bpm", "cav", "bcm" + _ "channel name indicator"
      • KP: 5th word variable names - underscores? Changing cm_us and cm_ds to bcm1 and bcm2 should be sufficient.
      • PK: For example: the SEE BPMs 4 x, y, p, m channels are parsed (the trailing XP, etc. are removed) and new parallel branches are calculated and stored - leaves are sub-blocks and hardware sums. Similarly for Cavity BPMs (cavity class hasn't been written yet, but should be temporarily replaced with 3 "bcm" instances).
  1. cavity BPM and configuration files (japan and pan): Ciprian, Bob, Cameron
    • Bob fixed the control.db in
  2. BPM pedestal calibration

Present

Cameron, Bob, Paul King, Tao Ye, Ciprian

Excused

Ciprian

Retrieved from "https://prex.jlab.org/wiki/index.php?title=20181024-Analyzer-Mtg&oldid=3680"