[evlatests] 3-bit Science Tea 'answers'

[Ken Sowinski]

Though there may have been more questions than answers I can
record a few answers or comments.  My memory works better for
the more technical than astronomical questions.  Corrections
or further comments welcome.

On Mon, 25 Jun 2012, Barry Clark wrote:

> 1.  How do we set up for 3 bit observing?
>     We have a procedure.
>     a.  Is it optimal?  How much overhead does it entail?

The executor knows how to setup the T304s to optimally feed
the samplers by looking at the wideband correlator results
from the Widar station board.  This usually works, but is slow.
Therre is a bug which sometimes causes it to misstep andtake a
very long time to recover.  A script-callable funtion is being
developed.  When refined its use will be built into model2script.
It is not reasonable to think that this will take less than one
minute for each frequency setup.  Assume at least two minutes.

>     b.  Can we observe with it all day?

The consensus seemed to be: it depends.  I don't recall hearing
the suggestion, but it struck me that being able to servo either
T304 output or equalizer independently might be useful during
a long observation if power levels change.  The counter argument
is that we know that changing the attenuator does produce bandpass
changes of its own.

>         i.  Do we need to repeat it periordically?
>         ii.  Do we need to repeat it if system power changes?

Testing should be done to verify what theory says about 3-bit
sampler se4nsitivity.

> 2.  Can we flag data?
>     a.  Do we have ways of seeing bad and erratic samplers?

We need more experience.  To understand what the problems are
we have to distinguish sampler and DTS problems.  Sampler problems
lie in the realm of black art; DTS problems are understood by more
of us.  Some can be recognized in the deformatter and/or station
board.  Others seem to be well hidden and are seen only if there
is correlated signal against which to see the defect.  Careful 
examinatino of telcal output can find many of these problems, but
that is not (yet) automatic, and we don't know what to do if a
problem is recognized.

>     b.  What do we need in the way of subband based flagging?

I don't think this was discusssed.  There are DTS problems that 
might be manifest in a subband, but I think the entire baseband
is suspect in that case.  Interference?  There is an 'RFI detector'
for each subband which can, in principle, be used to generate
subband flags.

> 3.  How do we calibrate fluxes?
>     a.  An eight bit observation of the phase calibrator should work.
>         i.  May need corrections for the structure of the calbrator
>         ii.  May need corrections for elevation difference from source
>    b.  Or can the non-linearity of the dDiff system be corrected?

Tests are being planned to try to answer these questions.
I don't think any further questions were posed on this topic.

> 4.  How do we calibrate bandpasses?
>     a.  How near do we need to be in system power to our source?
>     b.  What are the symptoms of being too far away?
>     c.  Dare we stitch subbands?
>     d.  Does interference take out the whole subband in which it resides?

Again, these are questions we cannot yet answer.
While not strictly a three-bit system question, the tradeoff between
many short observations and one long observation was discussed.  There
was concern over how much additional time was necessary for bandpass
calibration.