thanks for making our meeting on Friday very focused and productive. I wi=
ll
post the
minutes as soon as a find a little time in my schedule. =
Right now I would like your input on 2 questions we discussed on Friday:
1) Beam energies. Since the Readiness Review Committee recommended we sli=
p a month,
we lost one of our low beam energies (1.833 GeV) pretty much completely. =
It
was suggested
that we rearrange the remaining runtime to get back to a similar split of=
low to
higher energy.
I ran a Monte Carlo optimization and found that we get the best error on =
the
"GDH" integral
at low to moderate Q2 if we run twice as long at 2 GeV than at 3 and 4 Ge=
V. Specifically,
according to that optimization we should increase the 30 days at 2 GeV to=
45
and decrease
3 GeV from 30 to 21 and 4 GeV from 30 to 24. There is a very easy way to =
do that:
- change the 7 days 10/22 - 10/28 from 4.01 GeV to 2.03 GeV. This does no=
t
interfere with
the other 2 halls
- change the 10 days 11/29 - 12/8 from 3.02 GeV to 2.03 GeV. Again, no
interference and
we even save one energy change (the rest of the run is at 2.03 GeV anyw=
ay).
Please send in your opinions on this proposed change. If it looks accepta=
ble
to all
respondents, I will make this our official request (probably by Monday
afternoon - sorry
for this extremely tight deadline).
2) Helicity scheme. I have posted a proposal for the helicity scheme at
http://www.cebaf.gov/~kuhn/Helicitymemo.html which was agreed upon by bot=
h
other halls.
There are 2 remaining questions:
a) Should we change helicity after 0.5 s or after 1.0 s? I would especial=
ly
like to hear
from DAQ people about this. Is it easier to read scalers only once a seco=
nd
than twice?
Are there other reasons to change faster?
b) Should each helicity period be determined by the ACTUAL AC phase (i.e.=
, a discriminator
on an AC outlet counting 30 or 60 cycles) or by the NOMINAL time (0.5 or =
1.0
s) for an
even number of phases? This seems to be the typical "theological" questio=
n
where there
are no obvious preferences (obviously, to zeroth and even first and secon=
d order
it just doesn't matter) but a lot of strong opinions. I include myself: m=
y
opinion is
we should use a fixed TIME (i.e., a clock), not a fixed PHASE. Here are m=
y reasons:
- I am more worried about normalization problems stemming from uneven dur=
ation of
opposite helicity periods than about electronic noise (we count real elec=
tron tracks,
not electronic currents like Parity).
- Using a fixed phase relationship may even by counterproductive: If ther=
e is
a strong
subharmonic noise (e.g., a blip every second), we might actually ENHANCIN=
G
that using the
AC phase. Using fixed time would give us a random phase walk which would
average out any
such noise.
- With a fixed time window, we could run for exactly 500 ms, then change =
helicity
(which takes several 100 =B5s of dead time) and then run again for exactl=
y 500
ms. This
would actually give us a better averaging over an exactly integer number =
of
phases than
a scheme were a helicity change always occurs at a fixed phase relative t=
o AC (because
there would be necessarily a blankout period of "bad helicity" which woul=
d
take up
always the SAME part of a full phase.)
O.k., blast away, let me have your counter-arguments. Send them to eg1_ru=
n so
we all
can participate in this discussion. However, sometime next week we should=
try
to come to
a consensus and then deliver the proposal to Larry Cardman and Charlie Si=
nclair.
Greetings -
-- =
- Sebastian