[evlatests] Noise Diode Stability

[Rick Perley]

The recent interest in calibrating our visibility data by use of the switched power has me looking into the historical trends of the noise diodes.  For this, I have used the 'flux density' observations, dating back to 2014.  These data were taken using the 8-bit system.

There are nine databases that I used,  five when the array was in C or D configuration, the others when in A.  I extracted the 3C286 observations from each.  Switched power calibration was applied, and my best models were used for all calibration stages (delay, bandpass, gains).  This has been done for all 8 Cassegrain bands.

I've analyzed only the X-band data so far.  By remarkable chance, one antenna has had the same receiver on it for the entire duration — ea08, with receiver # 27.  This is important for judging stability, since antennas will have differing efficiencies, and noise diodes may change when receivers are brought back for maintenance.

The evaluation process is as follows:  The program CALIB is used to determine the antenna gains.  If the Tcal associated with the noise diode, and the efficiency associated with the antenna are both correct, then the output from CALIB will be 1.000.  If either is wrong (unless they are both wrong by the same factor), the resulting non-unity gain tells us that the Tcal (or efficiency) is incorrect.  The deviation from unity permits a correction to either the Tcal or efficiency.
I chose to calculate a new Tcal using the derived gains:  The relation is:

Tcal(new) = Tcal(old)*gain**2.  (It's squared since the CALIB gain is a voltage gain).

(If it were the efficiency we wanted to change, we would divide instead of multiply).

I'll note here that the derived value for Tcal is independent of the old value — the old value is used in the calculation of the visibilities, so an error in this is offset by the calculated gain.

The results for ea08 are quite remarkable — better than I expected, and even better than I had hoped for!  I used only two spectral windows for the test, one at 8464/8436 MHz, the other at 10936 MHz.   Listed below are the derived ('correct') Tcal values, presuming the antenna efficiency hasn't changed, and is equal to the value that Bob Hayward and I derived some 20 years ago...

8464 RCP   1.15   1.14   1.15   1.21   1.21   1.17   1.18   1.12   1.14
8464 LCP    1.05   1.04   1.03   1.08   1.12   1.05   1.06   1.04   1.04

10936 RCP  0.65   0.61   0.63   0.66   0.67   0.63   0.65   0.63   0.63
10936 LCP   0.54   0.53   0.54   0.56   0.58   0.53   0.54   0.53   0.53

Observations dates were (from left to right)

16Feb2014, 11Oct2014, 25Jan2016, 31Jan2019, 24Oct2019, 09Sep2021, 03Oct2023, 31Jan 2025, 09Mar2025.

The small changes seen in the data above could have four origins:

  1.
Change in opacity
  2.
Pointing errors
  3.
Change in noise diode power, mostly likely due to temperature change
  4.
Receiver malfunctions

The first three must all be small at this band, but could easily explain the few percent variations seen above.

Most of the other antennas show similar stability during period where the receivers did not move.  There are a few exceptions, presumably due to some receiver malfunctions.

Finally, I'll note that there is chance we can use these data to calculate antenna efficiency differences:  I noted that when a receiver moved from one antenna to another, the resulting 'new' Tcal was usually quite close to what it was on the preceding antenna.  The simplest explanation is that the 'new' antenna's efficiency is different from that assumed by the ratio of the old to new Tcals.

Rick




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20260828/3306b8cc/attachment.html>