[evlatests] Subreflector positioning (focus) errors at X-band

[Ken Sowinski]

On Tue, 13 Feb 2018, Rick Perley wrote:

>     I have a 6.5 hour block of data, taken on February 04, with X, Ku, K, Ka, 
> and Q bands.
>
>     An extraordinary number of system flags were set *at X-band only*, for 
> some antennas.  Examination of the data showed that the visibility amplitudes 
> were fine.
>
>     Closer examination revealed the following curiosities:
>
>     1) Antennas 3, 13 and 26 were flagged for most of the run with 'focus 
> errors'.  Examination of the amplitudes showed nothing amiss.  But 
> examination of the phases showed that indeed, the subreflectors were in the 
> wrong position.  I can tell this because, for some of the scans, the 
> subreflector moved to the correct position partway through the scan.  For all 
> three antennas, the transition from 'off' to 'on' took 5 seconds, and the 
> phase changed by 120 degrees.  This corresponds to a 
> ~ 6mm error  -- the same for all three antennas.  Not enough to  affect 
> the amplitudes, but definitely enough to dramatically change the phases.  The 
> key point here is that for all three antennas, the move from the wrong to 
> right position took place well into the scan (a couple minutes).  It rather 
> looks like a command got 'waylaid'.

There are two lines of evidence that suggest this is happening in
the MIB or the FRM controller module.  First, since the data is
flagged, it means that the MIB got the command, processed it at
the right instant, but the focus position did not change.  If
the command had not been received there would be no flagging.
Second, I ran a copy of Rick's file modified to go through the
sequence of bands Rick used, but do nothing else.  while it was
running It was clear that ea03, for example, sulked and made an
alert when asked to go from the Q band to the X band focus.
At the same time ea04 behaved normally.  I did not look at
ea13 or ea23.  It is interesting that all the earlier focus changes
are small and in one direction; the change from Q to X is larger
and in the other direction.  This may be a clue or a coincidence.

I remember that in early days we had a problem with losing commands
between the MIB and FRM mechanism which we learned was caused by a
bug in the FRM microprocessor which caused it to lose one of the
commands if two commands were sent on the same WG cycle.  We avoided
this by telling the executor to use different activation times for
the rotation and focus commands.  Perhaps what is happening now is
realted to that.

If needed I can provide a simple script which evokes this behavior
and does little else.

Ken


>     2) Antenna 5 is different.  Here the noted reason for flagging was 
> 'subreflector error'.  But there was no noticeable change in amplitude or 
> phase when the flags were set, or not set.
>
>     In the script, the observation prior to the X-band observations was 
> always at Q-band.
>
>     I emphasize that *no other band* showed this focus offset behavior.
>
>     An earlier run, taken in mid January, showed exactly the same behavior.
>
>
>     rick
>
>
>
> _______________________________________________
> evlatests mailing list
> evlatests at listmgr.nrao.edu
> https://listmgr.nrao.edu/mailman/listinfo/evlatests
>