[evlatests] More on Drop Behavior

[Rick Perley]

    I secured a half hour of time this afternoon (Wednesday) to check on 
X-band amplitude drop behavior.  At Walter's suggestion, I did this test 
at 9 different frequency pairs, spanning all of the (currently) 
available X-band -- from 8035 to 8885 MHz.  A result of this is a new 
failure phenomenon -- described below. 

    The data were taken in continuum, 0.417 sec. averaging.  

    The IFs were commanded to 9 different frequency pairs, 8n35 and 8n85 
MHz, with n = 0, 1, 2, ... 8.  (I tried n = 9, but got no data at 
all).    Each observation was two minutes in duration.  IFPair AC got 
the 8n35 assignment, IFPair BD got 8n85. 

    Drops were seen at all frequencies on antenna 13 on all IFs, and on 
antenna 18, IFs B and D only. 
No drops were seen on antenna 24.  All drops were 'on the 1s'. 

    Some more details:

    For antenna 13, every single 'on the 1' integration had a drop.  All 
drops, on all IFs, at all frequencies, were of the same magnitude -- to 
0.75 of correct amplitude.  (note:  it's hard to give an amplitude, as 
these drops don't close -- the drop is different on EVLA -VLA baselines 
than it is on EVLA-EVLA baselines). 

    For antenna 18 (IFpair BD only), there are two different drop 
amplitudes, with median drop amplitudes of about 0.7 and 0.2.  All drops 
were 'on the 1s'.  But for most frequencies, the deeper drop was 
accompanied by the shallower drop on the very next record.  The deeper 
drops were not evident at 8585 MHz, and at two frequencies, the presence 
of drops is masked by a new phenomenon (see below).  At 8885, antenna 18 
gave no fringes (but 18AC did give good fringes at 8835 Mhz). 

    New Phenomenon. 

    An extraordinary new phenomenon was seen on two frequencies,  -- 
8185 and 8785 MHz -- on the BD side only.  Characteristics are:

    - The amplitudes are 'scalloped', rising smoothly (like an inverted 
parabola) to a maximum, then
smoothly falling.  There is no smooth low point -- the amplitudes 
abruptly begin to rise.  For antenna
13, IF B, at 8185 MHz, the period is about 16 seconds for this scalloped 
pattern.  At 8735, for antenna 13,
IF B, the pattern is similar, but the period is much shorter -- about 8 
seconds.  The ratio of peak to trough of the pattern, for antenna 13, IF 
B, is about a factor of 5 at 8185, and a factor of 2 at 8785. 
    The phases show a repeating pattern also -- closely sinusoidal at 
8785 MHz, much less sinusoidal (but perfectly repetitive) at 8185 MHz. 
    The same effect is seen on IF 'D' for antenna 13, but with different 
amplitudes. 

    Other EVLA antennas show the effect in varying degrees:

    Antenna 14:  Clearly present, similar to antenna 13 in amplitude and 
phase at 8185.  Not clearly visible at 8785, but the amplitude stability 
is extraordinarily bad, so we can imaging the same effect is present at 
a much higher level. 

    Antenna 16:  Present, but much weak repeating pattern.  Amplitude 
stability is appallingly bad, so despite the absence of a regular 
pattern, something is bad...

    Antenna 18:  Very strong pattern seen at 8785 MHz.

    Antenna 24:  Has no BD side, so nothing can be said.

    It is natural to think of this as a beating between two signals.    
Ken has a conjecture on its origin, which I leave to him to describe.