[b1_ana] Fwd: Uncertainties in measuring b1/Azz

[O. A. Rondon]

We need to emphasize that for a single pol+unpol period, nothing changes
  (intentionally) except the polarization, which is Pzz or 0. Each
polarization period is an independent measurement, all errors are
"local", they don't transfer to the next period. The only things that
matter are changes from 10 hours pol. to the next 10 hours unpol.

I specifically suggest that in the appendix about the dilf. we add an
equation showing that A1 = A and l1 = l, since the cup does not move,
the field stays on and the packing fraction won't change in 20 h of beam.

Also we need to emphasize that the 1 cm slow raster averages over the
area of the cup, so if the top gets a bit depleted, the material that
goes to the bottom compensates. But this does not happen in 20 h. And if
it does, we can check it with raster cuts. The top raster yield vs the
bottom raster yield have the statistics of total counts, which are in
the tens of thousands over 10 hours integrated over the acceptance.
But, again, the only thing that matters is the average over the raster
area, which is what we detect.

In summary:

- explain the procedure: multiple independent measurements from pol and
unpol periods.

- nothing changes during each cycle, except Pzz

- clarify in the appendix that A and l are identical during each ~ 20 h
cycle, to better than 0.1%

- slow raster averages over the cup. Local depletions would take much
longer than one data cycle. Comparison of yields with raster cuts would
allow to identify unexpected issues.

Settling has been seen at SLAC and during SANE. If we see it, we can cut
bad runs, etc. if needed.

Cheers,

Oscar


Dustin Keller wrote:
> I not sure why he lists luminosity as an additional issue.  We have a good 
> handle on acceptance differences so its really just beam current 
> calibration differences and drift in detector efficiency.
> 
> dustin
> 
>