Location and regular date and time of meeting - agreed weekly on Mondays; GB-137 / Soc-280; 2-3pm ET
RICHMOND DECKER, JUSTIN
Marty, Ryan, Ray, John, Richard, Justin, Arindam, Dave MacMahon, Jack Hickish
Timing issues in 1500 MHz modes * Arindam sat down with Jack and Dave and tried manually placing blocks to achieve timing. New floor plan currently being tested. If it returns no timing errors Arindam will proceed with builds.
Ryan and Justin did extensive tests with Manager and grail set up last week
Data look good but we found some situations where scans failed.
Bank managers seemed to get into some weird states when running two banks. Parameters from one were apparently being sent to another. Multiple prepare statements and a manually turning off one bank seemed to fix this but it will complicate testing with two machines.
There were instances when the wrong bof/DD mode was being loaded. Unable to recreate.
When running multiple scans in a loop, data stopped being written after 4 scans with STR(?) errors
Ray recommends checking configurations and using a more robust method to detect scan stops and restarts. Also should keep an eye on guppi_daq to make sure the problem isn't there. * Roach crashes * Justin and Ray will make sure that the proper procedure is being followed whenever valon frequency is set. Namely, totally unload bof, make valon changes, reload bof * Dave asked if MMCM calibration is being done to make sure clock and data packets are aligned * Justin thinks this should be inherited by Backend.py base class, but will double check with Ray to make sure * Justin has written up some documentation * Ray and Joe will give a lunch talk next week on the new Matrix framework
Ray passed along repo information to Paul to patch guppi_daq
New wideband AP
Resource allocations for Q4
Should be announced soon
Richard thinks that the VEGAS "bump" artifact caused by the interleaved ADC's will be visible in pulsar bandpass
Size of the "bump" should be small unless the on/off continuum calibrator scans differ by a substantial fraction of Tsys
Should only be ~1 MHz wide
Ryan doesn't think this will have a significant impact, except for wide band flux density calibration