[evlatests] [Fwd: Failures to Tune, at Q-band]
Ken Sowinski
ksowinsk at nrao.edu
Mon Nov 28 15:00:25 EST 2011
On Mon, 28 Nov 2011, Rick Perley wrote:
> The Wednesday evening calibration test cycled through the 8 bands,
> at a rate of one per minute. Previous tests used the cycle:
>
> X -> C -> S -> Ku -> K -> Q -> Ka -> L --> X
>
> and resulted in a high fraction of tuning failures at Ka band on B/D
> side only. Note that this IF was tuned to 29.4 GHz, while the AC side
> was at 36.4 GHz. The preceding band was Q, and the frequencies were
> 47.9 and 41.9 GHz on the AC and BD IFs.
>
> For this new experiment (run Wednesday night), I perversely reversed
> the order of the tuning sequence, to see what this would do. So, the
> order was now:
>
> X -> L -> Ka -> Q -> K -> Ku -> S -> C -> X
>
> For this ordering, the failures to tune are profoundly different!
> The failures now occur at Q-band, on the B/D side only. The frequencies
> tuned were 47.9 (AC) and 41.9 (BD), while the preceding band (Ka) used
> the same pair as listed above: 36.4 GHz (AC) and 29.4 GHz (BD).
> There were 37 separate scans, and the total number of failures was 55 --
> a failure rate of (55/(26*37)) = 5.7%. Far too high! And none of these
> failures was flagged by the on-line system. The failures are uniformly
> distributed over the antennas -- only two antennas showed no failures (1
> and 28), while one antenna had five (ea04) and two antennas had four (26
> and 22).
>
> No other band showed an unusual number of tuning failures.
>
> It would be good to end this problem. And it would be useful if
> these failures could at least be detected and flagged by the on-line
> system.
When this happens it is often the case that the module has given
up and placed itself in 'standby' mode which means it will do
nothing until it gets a new command to tune to a different frquency.
I have no way of knowing whther this happens in every case, but I
use it as a diagnostic when troubleshooting non-working antennas.
There is a monitor point (obviously enough called 'standby')
to reflect this state. It could generate an alert which would cause
flagging for the appropriate IF pair. As a reminder:
L302-1 -> A0C0 or A1C1
L302-2 -> B0D0 or B1D1
L302-3 -> A2C2
L302-4 -> B2D2
In my experience the L301s are much less likely to fail in this way.
More information about the evlatests
mailing list