2017 November: Difference between revisions
(→Nov 07) |
(→Nov 07) |
||
Line 2: | Line 2: | ||
= Nov 07 = | = Nov 07 = | ||
'''12:28 UT''' There have been some notable issues in the past few days. The first is that, around Nov. 1 I rebooted the ACC, which loaded it from the disk copy of the LabVIEW code. After that time, we noticed some glitches that I recalled from a previous time, but I could not recall the source. Looking at some old emails from Aug. 31, we realized that the glitches were solved by loading the ACC code from memory, i.e. by downloading the source code from the Win computer. This is done by starting LabVIEW and launching ''Targets/ACC/Startup/ACC Master'' from the project page. After the target is runnning on the ACC, right click and select Disconnect from the menu. At present, we cannot explain why this should be any different from running it from the disk copy, but when run this way the glitches do disappear. The ACC was reloaded in this way on Nov. 6. We did an experiment by recompiling the deploying the Win1 copy back to the ACC, and that '''did not''' change the behavior. So there are a couple of scans on Nov. 6 that have the glitches. | '''12:28 UT''' There have been some notable issues in the past few days. The first is that, around Nov. 1 I rebooted the ACC, which loaded it from the disk copy of the LabVIEW code. After that time, we noticed some glitches that I recalled from a previous time, but I could not recall the source. Looking at some old emails from Aug. 31 (see the Aug. 31 observing log entry for more information), we realized that the glitches were solved by loading the ACC code from memory, i.e. by downloading the source code from the Win computer. This is done by starting LabVIEW and launching ''Targets/ACC/Startup/ACC Master'' from the project page. After the target is runnning on the ACC, right click and select Disconnect from the menu. At present, we cannot explain why this should be any different from running it from the disk copy, but when run this way the glitches do disappear. The ACC was reloaded in this way on Nov. 6. We did an experiment by recompiling the deploying the Win1 copy back to the ACC, and that '''did not''' change the behavior. So there are a couple of scans on Nov. 6 that have the glitches. | ||
A completely separate issue is that the dynamic spectrum seen in the daily spectrogram overview plots began to show an amplitude pattern starting around Oct 22 that became progressively more pronounced until the ACC was rebooted on Nov. 1. After that, the spectrogram went back to its usual appearance. I am not sure whether calibration of the data from Oct 22 - Nov 1 will be impacted, but this does seem to be an issue with synchronization of the DCM attenuation cycling. The lesson is that we should reboot the ACC from time to time, perhaps once/week. | A completely separate issue is that the dynamic spectrum seen in the daily spectrogram overview plots began to show an amplitude pattern starting around Oct 22 that became progressively more pronounced until the ACC was rebooted on Nov. 1. After that, the spectrogram went back to its usual appearance. I am not sure whether calibration of the data from Oct 22 - Nov 1 will be impacted, but this does seem to be an issue with synchronization of the DCM attenuation cycling. The lesson is that we should reboot the ACC from time to time, perhaps once/week. |
Revision as of 14:14, 7 November 2017
Nov 07
12:28 UT There have been some notable issues in the past few days. The first is that, around Nov. 1 I rebooted the ACC, which loaded it from the disk copy of the LabVIEW code. After that time, we noticed some glitches that I recalled from a previous time, but I could not recall the source. Looking at some old emails from Aug. 31 (see the Aug. 31 observing log entry for more information), we realized that the glitches were solved by loading the ACC code from memory, i.e. by downloading the source code from the Win computer. This is done by starting LabVIEW and launching Targets/ACC/Startup/ACC Master from the project page. After the target is runnning on the ACC, right click and select Disconnect from the menu. At present, we cannot explain why this should be any different from running it from the disk copy, but when run this way the glitches do disappear. The ACC was reloaded in this way on Nov. 6. We did an experiment by recompiling the deploying the Win1 copy back to the ACC, and that did not change the behavior. So there are a couple of scans on Nov. 6 that have the glitches.
A completely separate issue is that the dynamic spectrum seen in the daily spectrogram overview plots began to show an amplitude pattern starting around Oct 22 that became progressively more pronounced until the ACC was rebooted on Nov. 1. After that, the spectrogram went back to its usual appearance. I am not sure whether calibration of the data from Oct 22 - Nov 1 will be impacted, but this does seem to be an issue with synchronization of the DCM attenuation cycling. The lesson is that we should reboot the ACC from time to time, perhaps once/week.