Rob Kutschke
August 9, 2004

There are two accompanying postscript files,  ha34.ps, ha15.ps.
The files are also available as .pdf.

The first file shows 7 pages of plots for data taken today and
the second file shows the same 7 pages of plots using date
taken on June 2. 

I know that the data on June 2 were taken with long gate mode,
which I believe has a resolution bandwidth of about 1kHz.
I don't know the configuration of the board for today's data.

The bottom line is that there are some problems.  There is
a clue to the problem in the phase behaviour of the signals.

All data in this note are for protons only.

Page 1.

  The top plot shows the proton sum signal, |A| + |B|, 
for about 10 minutes around 1 PM.  The horizontal axis is
time of the day, in hours.  The vertical axis is Echotek
units.  On both the  ha34 and ha15 plots, the data is drawn
to appear at a mean of 75% of full scale.   The new
data clearly has poorer resolution on |A|+|B|.

   The bottom plot shows the proton position signal,  The horizontal
axis is time of day in hours and the vertical axis is position in 
mm.  The ha34 and ha15 files are on the same vertical scale.

  The position is defined as: 26*(|B|-|A|)/(|B|+|A|)

  There is clearly something wrong with the ha34 position signal.


Page 2.

  This shows a plot of the phase of the A signal vs the proton
sum signal.  Both axes are in echotek units.
Here phase is just the argument of the complex number (I,Q).

  The ha15 data shows the expected behaviour: there are 5 stable
values of the phase.  The proton sum signal looks much the same
for all 5 phases.
 
  The ha34 data has phases all over the place.  This suggests
a timing problem.  The amplitude clearly depends on the phase.

  The two sets of plots have the same horizontal scale, 95% to 105%
of the mean value ( I only eyeballed the mean value).


Page 3.

  The phase of the A signal vs the proton position, in mm.

The HA15 data shows the expected behaviour, there are 5 stable
phases and the position does not depend on phase.

  The HA34 data shows that the position depends strongly on the phase.

  The two plots have the same full scale on the horizontal axis, 2.5 mm.

Page 4.

  The phase of B vs the phase of A.  In both cases the data line
on the line arg(B)=arg(A).    There are only 100 points on this plot.

  In the HA14 data the 5 stable phases are clear.  In the HA15
plot there appears to be 7 groups of quasi stable phases.

Page 5.

  Repeat of page 4 but with 500 points.  For the HA34 data, the
full line has been filled in.  The data still lies on arg(B)=arg(A)
but there is no clustering.  The HA14 data are still strongly
clustered but a spread out a bit more.

Page 6.

  Phase A vs magnitude of A and Phase of B vs Magnitude of B.
This is the cleanest plot yet which shows the correlation between
phase and amplitude of A and B in the new data.  I don't yet understand
the pattern - in particular I don't understand where the 4-fold
symmetry comes from.

  The  old data show the expected pattern of 5 stable phases with a
magnitude which is independent of phase.

 In all of the plots the magnitude axis covers 95% to 105% of the mean
value.

Page 7.

  The magnitude of B vs the magnitude of A.  Both axes are
approximately  95% to 105% of the mean value.


Conclusions:

  I think that page 6 is the key.  It tells us that there is
a timing problem.  The 4 fold symmetry should be a good clue
about the source of the problem - maybe the measurement can
start at 4 distinct times relative to the arrival of the
first bunch?