The plots below show the gain ratio for all  calibration sources plotted versus frequency. There are 4 plots per page. Each plot holds 5 sources. Sources are distinguished by color and symbol. The large  variation for some sources (especially at higher frequencies) is caused by   collimation errors (and the gain fits inability to fit it).

• jan04 to current date:  gainMeasured/gainFit all datapoints (.ps) (.pdf):
This plots all of the measurements taken between jan04 .
• jan04 to current date:  Median by frequency of gainMeasured/GainFit (.ps) (.pdf):
• For each source at each frequency the median ratio was computed and plotted.

Some sources  that have been sampled across the entire frequency range show a drop in their ratio with frequency: B0221+276, B0932+089, B1622+238, B2018+295, B2251+158, B2337+220.  Not all of these sources are 'good' calibrators (B2251+158 3C454.3 is a strong variable).

How the fluxes were checked:

let y=log10(fluxJy)

let x=log10(freqMhz)

y= a0 + a1*x + a2*exp(-x)

    To check that the source fluxes used are ok we have computed:
 

We do this for each calibration scan taken for all receivers.
Some things to keep in mind are:

• When taking the ratio we always use the gain curve that was valid during that epoch . The gain curves have changed over time do to alignment procedures.
• A gain curve is made and then not changed until something  requires a new gain curve (new receiver, telescope alignment, etc.). We continue doing calibration scans to check things out. So there are many calibration scans that were not used in creating a gain curve.
• A point whose ratio is an outlier could be caused by:
• The source flux at that frequency is inaccurate
• The measurement may have had some problems (e.g.. we were using high frequencies and it was cloudy or raining).
• The gain curve fit is not doing a good job of fitting the az or za dependence.

 Systemp_performance page:

 home_~phil