The example below has been supplied by Alberto, though some administrative details have been left out to make the example less cluttered.
First we define some convenient variables, so we can re-use them in the script. The rule of thumb should be any number, or certainly multiply occuring text, should be used in a (shell) variable. That rules is not quite followed in the current example:
set FILE="c0001.n604_coC.1.miriad" set SRC="NGC604" set CAL1="0205+322" set CAL2="0237+288" set PBCAL="3C454.3" set NOISE="NOISE" set FLUX="URANUS" set WIDE="channel,1,1,15,1" set LINE="velocity,63,-317.521,2.54,2.54" set CAL=$CAL2 set OCAL=$CAL1 set RESTFREQ="115.271202" set REFA=9
Some of variables may be quite obvious, others less. For example, the setting for LINE= less came from gleaning the output of uvlist:
% uvlist vis=$FILE options=spectra rest frequency : 115.27120 115.27120 115.27120 115.27120 115.27120 115.27120 starting channel : 1 16 79 142 157 220 number of channels : 15 63 63 15 63 63 starting frequency : 111.47148 111.08239 111.05307 114.93370 115.32280 115.35211 frequency interval : -0.03125 -0.00049 -0.00049 0.03125 0.00049 0.00049 starting velocity : 9854.752 10866.791 10943.046 849.513 -162.527 -238.782 ending velocity : 10992.691 10945.532 11021.788 -288.426 -241.268 -317.523 velocity interval : 81.274 1.270 1.270 -81.274 -1.270 -1.270
First we note that the data set from CARMA is a single miriad dataset that contains all the sources. It is often , except in the most simple cases, much more convenient to keep track of things if the data is copied to single-source (or even single-setting) datasets:
foreach i ($SRC $CAL1 $CAL2 $PBCAL $NOISE $FLUX)
uvcat vis=$FILE out=$i select="-auto,source("$i")"
end
It cannot be stressed enough to inspect the data data visually , in as many ways as you can imagine. Here are just a few examples:
# phase vs. time uvplt vis=$CAL device=/xs line=$WIDE axis=time,phase # amplitude vs. time uvplt vis=$CAL device=/xs line=$WIDE axis=time,amp # even for the source: it probably be random, unless there are false fringes # or it is a very strong source uvplt vis=$SRC device=/xs line=$WIDE axis=time,phase uvspec ...
As a result of this inspection perhaps we found some suspicious data, and it needs to be flagged. This could be in certain channels and/or time slots. Here is an example to flag a certain time-range for antenna 5:
uvflag vis=$CAL,$PBCAL,$SRC flagval=flag select="ant(5),time(21:30:00,22:15:00)"
First we proceed with (astronomical) passband calibration, to make sure the trends we saw in phase vs. time are not washed out by passband slopes. Notice we're compressing the whole time-ranges to get a single passband shape for all times:
mfcal vis=$PBCAL line="channel,282,1,1,1" interval=999 refant=$REFA
and inspect the result
uvspec vis=$PBCAL axis=chan,phase line="channel,282,1,1,1" device=/xs interval=999 yrange=-180,180
copy the passband from the PBCAL to the CAL
gpcopy vis=$PBCAL out=$CAL options=nocal
Create new dataset with calibration applied, otherwise linetype averaging does not work properly. Use all wideband channels.
uvcat vis=$CAL out=$CAL.pb select="window(1,4)"
Proceed with amplitude-phase calibration
gpcopy vis=$PBCAL out=$FLUX options=nocal uvcat vis=$FLUX out=$FLUX.pb select="window(1,4)" selfcal vis=$FLUX.pb options=apriori,amp,noscale interval=0.1 line="channel,1,1,30,1" refant=$REFA bootflux vis=$FLUX.pb,$CAL.pb primary=$FLUX line="channel,1,1,30,1" taver=999
Self calibrate the phase calibrator, with passband calibration applied, and imposing the flux found by bootflux solution
selfcal vis=$CAL.pb line="channel,1,1,30,1" options=amp,noscale,apriori flux=1.2 interval=20 refant=$REFA
Inspect again. Now each channel should have a zero phase
uvspec vis=$CAL.pb axis=chan,phase line="channel,272,1,10,1" device=/xs interval=999 yrange=-180,180
Show time series of selfcalibrated wideband channels:
uvplt vis=$CAL.pb axis=time,phase device=/xs line="channel,1,16,15,1" uvplt vis=$CAL.pb axis=time,amp device=/xs line="channel,1,16,15,1"
Or for a phase-only calibration, we self calibrate on the phase calibrator, with passband calibration applied. Most of the time the online amplitude calibration seems very good... Note that we are averaging over all wideband channels. Channel linetype averaging does not weigh by bandwidths and/or Tsys. This is why we split out only the continuum windows.
selfcal vis=$CAL.pb line="channel,1,1,30,1" interval=20 refant=$REFA
Inspect the results, again every channel should have zero phase:
uvspec vis=$CAL.pb axis=chan,phase line="channel,272,1,10,1" device=/xs interval=999 yrange=-180,180
Now show a time series of self calibrated wideband channels:
uvplt vis=$CAL.pb axis=time,phase device=$device line="channel,1,16,15,1"
Now that all calibration is done, it is a good idea to do some sanity checks. Looking
at the gain amplitudes 
indicate that the phase calibrator was weaker than
expected, perhaps due to pointing errors,
gpplt vis=$CAL.pb device=$device yaxis=amp
The phase gains should be smooth now:
gpplt vis=$CAL.pb device=$device yaxis=phase
Looking at the phase vs time after selfcal, they should be centered around zero:
uvplt vis=$CAL.pb device=$device axis=time,phase line=$WIDE
The phase vs baseline length plot should be inspected to assess atmospheric decorrelation, it should be flaring at the longer baselines but not overall decline:
uvplt vis=$CAL.pb device=$device line=$WIDE axis=uvdist,phase options=nobase
And finally amplitude vs. time: it should be about was it was set to in the selfcal if an amplitude selfcal was done:
uvplt vis=$CAL.pb device=$device line=$WIDE axis=time,amp
In the actual example script it now continues mapping the calibrator, and finally a number of the same set of observations for the source. It can be found in the examples directory as example-blabla.csh.