[Sane-analysis] Update to CVS for g_analyze_scaler_bank.f

[Mark Jones]

Hello all,
      The scalers measure the average beam current (cur) ,
the current with helicity plus (pcur) and helicity minus (mcur).
The beam current is monitored by a cavity which produces a voltage
which is then converted to a frequency (V to F converter).
An additional complication is that the there is always a positive offset 
voltage to the signal going into V to F converter, so even at zero
current the "charge" scalers are counting. This frequency is about
250kHz. When there is a 100nA beam the frequency rise only to
257kHz. The time is needed to convert the current to a charge.
The scalers measure the total time (time), the time with
helicity plus (ptime) and helicity minus (mtime) using a 1Mhz clock.
The actual beam current is calculated as

beam current = ( cur/time - 250kHz)*(conversion const)
charge = beam current * time

There is a 500usec blanking period when the beam helicity
is flipped to let the polarization of the beam settle.
The beam is flipped at 30Hz so one expects that the
ratio of (ptime + htime)/time = 0.985 . Also
same for the charge (pch + mch)/charge = 0.985 .


      Hoyoung noticed that for run 72377 that the
(pch + mch)/charge  was not 0.985. I actually found
it to be larger than one. Because of
problems with the 1Mhz signal down in the hall
during runs 72476 to 72588 we had be using scaler
174 which was also a 1mHz clock. I looked closer at
run 72377 and found that scaler 503 was actually getting
0.998 MHz so this signal was also going to the helicity
plus and minus clocks. So the total charge was using
the 1Mhz clock while the charge scalers were using a
0.998MHz under the assumption that it was 1.0MHz.
This change in the clock from 1MHz to 0.998MHz for run
72377 must be do to missing some of the pulses. Eventually
the signal failed downstairs for runs 72476 to 72588.
After run 72588 the scaler 503 is 1mhz and agrees with
scaler 174, so the problem doesn't exist.


To fix the problem before run 72476 I scaled

delta_time_help =delta_time*gscaler_change(516)/gscaler_change(503)
delta_time_helm =delta_time*gscaler_change(515)/gscaler_change(503)

where delta_time is the time using scaler 174.


For runs 72476 to 72588, I use the fact that over
a long time period the time for each helicity is 0.985/2.
and

delta_time_help = 0.4925*delta_time
delta_time_helm = 0.4925*delta_time

Of course for a 2 sec period this is wrong, since 1/30
is a significant part of 2.  But over the period
of a run then 1/30 sec is small.


I did notice the
  (ptime + htime)/time = 0.98527 which is
close to the expected 0.985 .
But for the charge
(pch + mch)/charge = 0.9839 .

The ratio is 1.0014 which is close the
false asymmetry that we are seeing. But
I can't assign the missing part to the
plus or minus helicity since the charge
asymmetry is at this level.

So I have put a new version of g_analyze_scaler_bank.f
in the CVS.


                        Mark



-- 
				Mark Jones
				757-269-7733
				jones at jlab.org