[evlatests] [Fwd: Antenna-Based Wobbles -- it's all about 2 and 23!]

[Rick Perley]

    The (long) note  sent around summarizing the 'wobble' situation 
included a critical error.  The effective LO offsets are 128 or 256 
*MHz*, not kHz. 
    The version below has been corrected for this...

    Thanks to Keith for finding this. 



-------- Original Message --------
Subject: 	[evlatests] Antenna-Based Wobbles -- it's all about 2 and 23!
Date: 	Thu, 24 Jan 2013 09:58:07 -0700
From: 	Rick Perley <rperley at nrao.edu>
To: 	evlatests at aoc.nrao.edu



      Barry has proposed a mechanism to explain the slow sinusoidal 
wobbles in amplitude and phase which are seen on some baselines while we 
were in the A-configuration.  In short, the mechanism requires a 
secondary LO, offset by a couple hundred kHz from the correct value.  
The effect of this is to cause a beat in the visibilities between two 
'copies' of the source spectrum. 

      The proposed mechanism allows predictions:  the most important is 
that it predicts the beat frequency, which would be proportional to the 
E-W baseline length, measured in wavelengths of the LO offset.    
Comparison of the observed beats shows this prediction is dead on, 
provided the offset is 128 MHz at Ku-band, and 256 MHz at K band.  It is 
useful to point out here that a 128 MHz frequency beat on a 35 Km 
baseline has a fringe frequency of ~1 Hz -- so we expect significant 
attenuation of these beats on the long E-W baselines.

      But the mechanism fails in two other predictions:  (a) We should 
see the beats in both polarizations, but only see them (strongly) in 
LCP.  (b) The beats should be seen equally across the entire spectrum, 
but are seen (strongly) only in subbands 1 and 2. 

    I've done a little more poking around in the data taken 3 weeks ago 
which so clearly showed these beats.  Attached are two pdf files showing 
that the effect is *only* seen on baselines which include antennas 2 and 
23, and is (almost) completely absent on all others.  Both show the 
interferometer phase, in subband #2, in LCP, at 25 GHZ, versus the 
coordinate 'u', (measured in wavelengths at 25 GHz). 

         (a)  No2or23.pdf shows all baselines which do not include 
either 2 or 23.

         (b)  Only2and23.pdf shows only baselines connected to 2 and 23. 

Both plots are on the same vertical scale.  The difference is dramatic, 
and show that the wobbles are only found on baselines to ea02 and 
ea23.   The 'block' of high phases seen in the center are all from the 
single baseline 2 x 23.  The plots indicate that the mechanism is indeed 
present on all, or nearly all antennas, but that it is only really 
effective on the two antennas 2 and 23.  The effect is weak enough on 
other antennas that their cross product is negligible compared to the 
noise.  On 2 and 23, the wobble amplitude is 2 degrees  = 1/30 radian.  
We're looking for a 3% effect -- about -15 dB.  On the other antennas, 
the effect must be at least an order of magnitude weaker. 

    Yesterday, Keith and Eric went 'sniffing' in ea02 and ea23, plus one 
other (control) antenna, looking for spurs.  Unfortunately, nothing 
obvious was found.  Keith will send around an appropriate summary, after 
the data are reviewed.  We may be disappointed, but not surprised at the 
null result -- the characteristics of the observed wobbles cannot be met 
by the simple theory.   We need to think of a more obscure phenomenon 
which only affects LCP, and can only be seen in the lower two subbands 
...  However, the success of the simple theory in predicting the wobble 
frequency says (to me, at least) that the offset LO concept is in the 
right direction. 

    Unfortunately, we are now in the D configuration, and the two 
discrepant antennas are adjacent on the north arm.  The predicted 'beat' 
frequency will now be minutes, making diagnosis on real data rather 
difficult.  Had we arranged to put those antennas at the ends of the 
East and West arms, we would see beats of period ~ 35 seconds, which 
would be quite nice ... 


-------------- next part --------------
A non-text attachment was scrubbed...
Name: No2or23.pdf
Type: application/pdf
Size: 1584320 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20130124/9b9b4416/attachment.pdf>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Only2and23.pdf
Type: application/pdf
Size: 1451099 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20130124/9b9b4416/attachment-0001.pdf>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: Attached Message Part
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20130124/9b9b4416/attachment.ksh>