Beams-doc-2127-v1  February 1, 2006

Rob Kutschke

This is a first look at raw mode data taken by
Steve Foulkes earlier today.  The data was taken
during state 21 ( 6 batches to NUMI ).  The trigger
was set when there were 6 batches in the MI.

In this mode I do not get (I,Q) data.  The quantities
A and B are just 14 bit signed integers. 

The following notes refer to the pages of the
accompanying .ps (.pdf) files.

Page 1: 
The horizontal axis is ticks of the 10/7 RF digitizing 
clock. The vertical axes on the three plots are,
A, B, |A|+|B|, all in Echotek Units (EU).
In these figures, one turn is 840 ticks.
The abort gaps are clear.  For example there is one
centered about 500 ticks and another centered
at 1340 ticks.  The individual batches are sort of
visible but not well defined.
We clearly need more gain if we want to do this measurement
seriously.

It's not clear to me if the sum signal has any real 
meaning for this circumstance.

Page 2:  
Repeat of page 1 but with a different vertical scale.
The batches are now resolved with the eye.
On page 4 I will show the FFT of some data.  The 
red vertical lines on this plot delimit the data used
for the FFT, approximately one full batch.

Page 3: 
The plot is A vs B.  The data are scattered around the
main diagonal, which implies that the timing is reasonably
well matched and the filters on the transition board
are reasonably well matched.

Page 4: 
These are FFTs of the data between the red vertical lines
in Figure 2.  When I compute FFTs, I always subtract out
the mean value of the array before doing the FFT. 
 
The colored lines are drawn at 1/7 (Green),
3/7 (Red) and 4/7 (Yellow) of the RF frequency.  A and B
both have a strong line at 3/7 of the RF frequency.
This is the expected alias from undersampling
the beam signal with a 10/7 RF clock.   The green line
is drawn at the frequency with which batches go by.
We don't see anything here; this is a left over from an
earlier version of this analysis in which a mistake created
a feature at that frequency.  The sum signal has a strong
feature at 4/7 RF. Bob explained that |A|+|B| has
an implied square in it.  This modifies a sin(w) signal
to look like (constant+sin(2w)).  Sin(2w) has a frequency
of 14/7 RF, which will have an alias at 4/7 RF when
undersampled with a 10/7 RF clock.  The constant component
will have an alias at baseband; the main body of the 
this alias has been removed since I subtracted the mean
value of the data before doing the FFT. However its 
various side lobes could well explain the low 
frequency noise.