How to measure the stability of the receivers
CK 21-Jan-2012 (1st version: 15-Sep-2010)
The idea is to "observe" the hot load taking data over at least an hour and at a rate of about 0.5seconds. This allows to monitor the stability of the entire receiver/IF-distribution/backend chain. Such observations should be done on any of the receivers and with the continuum backends connected to measure the total power stability, but also with the spectrometers (e.g. the FTS) connected to measure the spectroscopic stability. The instrumental drifts are characterized by an Allan-variance analysis, by binning the data in bins of increasing integration time and finally plotting the variance versus integration time. For pure white noise, the noise will drop with T**-0.5. However, at some point drifts will start to dominate and the curve will deviate from the -0.5 slope. (References: e.g. Roelfsema et al. 2012, A&A, 537, A17; Ossenkopf et al. 2008; Schieder & Kramer 2001.)
Observing the absorber
The procedures lie on: t00-10@mrt-lx3.iram.es:~/pako/HemtStability
- For a first measurement, I suggest to connect E0 and the FTS. The tuned frequency does not matter. Once, we have obtained some data, we will device a systematic test plan.
- First, create a dummy XML file:
pako> cal /ambient yes /cold no /sky no /tcalibrate 60 ! longer than 60 is not allowed !
pako> set doSubmit no
pako> start
- edit the created XML file '~/pako/XML/calibrate.xml'
- modify all instances of tcalibrate (total observing time in seconds, e.g. 3600)
modify tphase time to e.g. 0.5sec (TimePerPhase, TimePerRecord, dump time / time bin length, e.g. 0.5 sec)
- on mrt-lx1: 'submitObservingBlock calibrate.xml' to start taking data
- Check that the FITS raw data file has been created on observationData/imbfits. Identify full scan number, e.g. 20100822s227.
Data reduction
THIS PART IS STILL INCOMPLETE AND MAY NOT WORK.
Routines were originally written by HW.
Create Allan-Stability plot of continuum scan: greg @ plotAllanCK no 20100822s227 CONTI this routines runs a fortran program to calculate the allan variances against time, outputs the data to allan.dat and then plots them. Each line in the input file allan.dat corresponds one variance point. Each column corresponds to a frequency. For continuum observations this can be the 4 4GHz IF channels. Create Allan-Stability plot of WILMA scan: greg @ plotAllanCK no 20100823s72 WILMA 1 11 21 31 41 1 corresponds to the backend part. 11 21 31 41 are the channel numbers to be used for the calculations. These should not be consecutive channels, to avoid any correlation between channels.