June 11, 2010
We started EARL up at 1:19pm GMT this morning. We only got to run until 6:43pm GMT, and then it started raining. Weather radar suggested that the rain would last for a couple of hours, so we just headed home early for the day. The data we did get was useful for clear skies. From 1:19pm GMT until 2:36pm GMT, we set the "data sets per save" to 8 to see how it would affect clear skies data. From 2:36pm GMT until 3:47pm GMT, we set it to 16 data sets per save. Then we set it at 32 data sets per save until it got cloudy around 4:44pm GMT. Because of the high amount of saturation, we placed a cloud cover on the primary mirror, but continued with the 32 data sets per save to see how that worked with clouds. From 6:00pm GMT until 6:35pm GMT, we used 8 data sets per save because it was so cloudy and noisy; then from 6:35pm GMT until we shut the hatch, we set it to 16 data sets per save. This is highly skewed data, in other words. It would probably have been more useful to vary the data sets per save over several days rather than a few hours, but Haviland wanted to view them side-by-side during the same day.
Obviously, this data set did not go so well. Changing parameters that quickly was probably not a smart idea. Also, the laser power seems to have gone temporarily wonky, if "temporarily" can be translated as "for two and a half hours". Most of the data was thrown out. Thankfully, those were mostly cloudy times, and Haviland had experimented with different data sets per save previously on the 10th. The Background Subtracted graph shows the same error as that received on the 12th. Matlab could not read in the "NaN" that EARL came up with at about 4:20pm, 4:40pm, and 5:40pm GMT.
Questions to be answered: (A) What happened to the laser power? The laser was on for the entirety of the run, so it was not due to our tampering with the "Laser On" button. How is this graph even calculated and plotted? What does it mean? (B) What is the NaN difficulty? How can we fix it? Does it mean that the mirror is too saturated? (C) Can any accurate information be gathered about how data sets per save affect clear skies data and/or cloudy skies data?
Most of the cloudy skies data resulted in Time Skips on ProcEarl, so those changes were probably not observable in the graphs. However, the clear skies data seems to suggest that there is not difference between 8 and 16 data sets per save for clear skies, except that file folder for that day contains more files and is, therefore, much larger and much more difficult to send via zip files over email.