[esa-t474] ELOG and paper updates

[Mark Slater]

> 
> The hypothesis (scale changes coupled to the difference in offsets) is
> reasonable, but do the numbers make sense ? In other words, if you plug in the
> measured offsets, what kind of scale drifts would you need to explain the
> observed relative drifts ?

The plots that are simulated (that are in the ELOG entry) are with
sensible offsets (as measured) and with a 1% scale change (which we also 
measure from the mover scans).

> 
> I am a bit confused what plots you would like to leave in the paper. I think
> some 20 hour plots should still be there, with a proper hypothesis/explanation
> for the cause of the drifts.
> 

We were going to leave in the 1 and 20 hour plots, but just remove the 
correlation plots as these don't tell you very much. We can hopefully 
discuss this later though.

 > > http://pcfj.hep.phy.cam.ac.uk:8080/slater/38
> > 
> 
> Ah ! I think I did ask before which temperatures you used to plot the drifts
> against (see my previous message). I guess you were plotting against BPM
> temperatures ? BPM temperatures only affect the frequencies, which you already
> showed not to be a dominant cause of the drifts. The rack temperatures, on the
> other hand, would affect electronic gains (of course). The scale was measured
> by Maria and Erik last summer. See
> 

I was previously plotting against BPM temperature but the shapes are all 
the same (which is what I was worried about before). The electronics 
temperatures in ESA show the same time dependence but have a larger range 
(~1.5 deg C). At present however, I have no idea how the electronics gains 
are affected (other than from Maria's plots). Alex is looking into whether 
this temperature change could cause the measured 1% scale change.

> https://www.hep.ucl.ac.uk/elog//T-474/83
> 
> and in particular plots in
> 
> http://www.slac.stanford.edu/~chistiak/MccNoOutliers/Amplitude/
> 
> Looks like the scale drifts in processors 3,4,5 are O(0.1-0.15%/degree F), or
> 0.2-0.3%/degree C. Phase drifts seem negligible. Is this sufficient to
> generate observed differences, per your simulation ?
> 

Very nearly, but I've assumed a 1% scale change as that's what we measure. 
This gives a ~2um drift if all the gains change the same which is in 
reasonable agreement with the measured drift.

> 
> > We also got through English corrections for Sections 1, 2 and most of 3.
> > These are now included in CVS with a rough change to the results section
> > to represent what we found out.
> > 
> > If I remember correctly, the next meeting about this will be on Monday
> > after the ESA meeting. I'm hoping at this meeting to finalise the plots
> > that people want to show in the results section, then hopefully one more
> > meeting next week at UCL will complete the English corrections.
> > 
> 
> Yes, would be good to have a concrete proposal, e.g. a new paper draft, or at
> least a list of proposed plots. I will try to make the meeting, although other
> priorities (my daughter's birthday party with 6 tween girls in the house :-o )
> may interfere. My apologies in advance.
> 

Thanks for the comments and hopefully talk to you later!

Mark


> Yury
> 
>