[evlatests] Tests of the new ACU on ea21

[Rick Perley]

    Ken and I have done a number of performance tests of the new-style 
ACU on ea21.  The tests were taken with two different modes:

    1) Using the holography modes to step across the beam at L and X 
bands. 

    2) Using referenced pointing mode to step between the half-power 
positions, and the beam center. 

    These tests showed curiously different behavior -- here we summarize 
the results from the second test. 

    For referenced pointing, the executor moves the antennas around the 
following cycle:

    On -->  +El  -->  -El  -->  +Az  -->  -Az  --> On ...

    Each off-axis position is to the half-power position (or 2.6 
arcminutes at the band utilized here), and the duration of each step is 
20 seconds.  The data from this test were recorded at 10 Hz -- giving us 
excellent resolution to see any oscillatory behavior. 

    The pointing data were calibrated by using the on-source position, 
assuming a unity flux.  Then the antenna gains were solved for, using 
'CALIB', for each 0.1 second record.  The attached plots show *inverse* 
antenna voltage gain:  A value of near 1.0 means the antenna is on 
source.  A value greater than 1 indicates the antenna is off-source.  
The half-power will be at a level 1/sqrt(2) = 1.41.  A value of 10 
indicates a pointing at the -20dB level.  (10*log(1/10^2)). 

    Attached are five plots, showing the give transitions shows above, 
for ea21 (the upgraded antenna), and ea28 (representing the old ACU 
system).  All other antennas behave similarly to ea28, differing only in 
magnitude of the oscillations shown. 

    1) NegAz2On.eps 

    The antennas moved clockwise by half the beamwidth, to the 'on' 
position.  The plot shows that ea21 executed this move in 0.4 seconds, 
with virtually no overshoot.  Ea28 clearly overshot, with at least 5 
seconds needed to get stably on the beam axis.  A curiousity is that 
ea21 started its motion 0.3 seconds before the other antennas. 

    2) On2+El.eps

    The antennas moved upwards by 2.6 arcminutes, from the 'on' position 
to the half power.  ea21 takes 0.8 seconds to reach this point.  Both 
antennas oscillate at 2.0 Hz following the move, but with the old ACU 
antenna the oscillation very much larger.  For some VLA antennas, the 
amplitude of the oscillation is at least 2 arcminutes beyond the 
targeted position.  The oscillation amplitude for ea21 is barely 
perceptible.  Both old and new systems show a mean overshoot, but ea21's 
overshoot is much less, and the antenna appears stable after 4 seconds, 
less than half the time taken for the old system to stabilize.  Ea21 
again starts earlier than ea28 (by about .25 seconds). 

    3) +El2-El.eps

    The antennas moved downwards by 5.2 arcminutes, going from one 
half-power, through the main beam, to the opposite half-power.  The 
difference between the two systems is most clear in this plot:
    The new ACU moves smoothly through the beam (upper plot), showing a 
slight overshoot and very minor 2 Hz oscillation.   Data are stable 
after 5 seconds. 
    The old ACU (ea28) far overshoots the the target position, reaching 
to about the first null (about 5.5 arcminutes from the beam center), 
followed by large-amplitude 2Hz oscillations, which continue for well 
over 10 seconds. 
    ea21 again started 0.3 seconds earlier than the others. 

    4) NegEl2+Az.eps

    The antennas move diagonally, from the 'down' position to the 
'right' half power.  The plot for ea21 (upper) shows the motion was very 
nearly diagonal -- the lowest value is barely below that corresponding 
to the half-power positions.  There is a fairly notable 2 Hz oscillation 
in this motion, (more than in any of the other motions).  These die away 
after about 6 seconds. 
    For the 'old' ACU, a very different profile is seen:  The motion is 
nearly through the center of the beam (as the gain drops very closely to 
1.0) -- we think the elevation motion is much the greater, hence driving 
the antenna in an curving arc.  There is a spectacular azimuth 
overshoot, -- nearly to the first null -- before the antenna corrects 
with a slow, ~ 8 second, feedback loop.  Superposed is the usual 2 Hz 
mechanical oscillation.  About 10 seconds is needed for the system to 
become stable. 

    5) +Az2-Az.eps

    Finally, the antennas are driven horizontally by 5.2/cos(el) 
arcminutes, in a anti-clockwise motion.  The plot for ea21 shows a 
smooth crossing of the beam, with essentially no overshoot, other than a 
small 2 Hz oscillation.  About 4 seconds is needed for complete stability. 
    By contrast, the old-style ACU again shows a huge overshoot -- again 
essentialy to the first null, before a slow correction to the commanded 
position.  The 2 Hz oscillation is embedded in the overall overshoot 
curve. 

    Summary:
   
    The new ACU system nearly eliminates the overshoot problem, both in 
azimuth and elevation motions.  The smoother acceleration greatly 
reduced the 2 Hz oscillation, particularly in the elevation motion 
direction.  The only adjustment that I think would be useful is to 
reduce, if possible, the 2 Hz oscillation in the azimuth driving 
direction. 



   
-------------- next part --------------
A non-text attachment was scrubbed...
Name: NegAz2On.eps
Type: application/postscript
Size: 15317 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20140103/3906bfcf/attachment.eps>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: On2+El.eps
Type: application/postscript
Size: 24506 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20140103/3906bfcf/attachment-0001.eps>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: +El2-El.eps
Type: application/postscript
Size: 28674 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20140103/3906bfcf/attachment-0002.eps>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: NegEl2+Az.eps
Type: application/postscript
Size: 39904 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20140103/3906bfcf/attachment-0003.eps>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: +Az2-Az.eps
Type: application/postscript
Size: 23828 bytes
Desc: not available
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20140103/3906bfcf/attachment-0004.eps>