Spectral Line Regression Test Analysis - July / August 2017

My scripts, results, et. in /users/rprestag/vegas/regression

TGBT15A_915_120 (evening of Wed 30th ET)

Bottom line: The spectra look reasonable, and switched-power baselines look good. But in a pattern I cannot discern, the system temperatures are intermittently way off.

• Here is a summary of system temperature as a function of software running, receiver and mode: TGBT15A_915_120.txt
• It seems the correct system temperatures are ~30K for X-band, and ~18k for L-band - this makes good astronomical sense.
• all "old, old" scans look correct.
• Some "new manager" scans have an incorrect value of Tsys. But not always. This is the case for all banks / all polarizations (for every one I have checked), so for the "hybrid" scans, both old and new HPC software give the wrong answer.
• The "new manager" scans are not always incorrect.
• Examples
  • Scan 3, X-band mode 10, bad (corresponding old, old would be scan 31, but note integration times are different)
  • Scan 20, X-band mode 10, good
  • Scan 18, X-band mode 4, bad (corresponding old, old would be scan 29)
  • Scan 8, L-band mode 4, bad (corresponding old, old would be scan 40)
  • Scan 10, L-band mode 10, good

-- RichardPrestage - 2017-08-31