A. Systematic
uncertainties which have an impact on the luminosity:
1.
Bunch to bunch current variation:
-
The bunches ( 43, 54, 108, 156, > 2000) will be spread in
intensity and emittance.
-
What is acceptable? Aim for < 10% in intensity and 20% in
emittance. Discussed with W. Herr, to be studied further.
-
Seems to match about what is feasible from injectors (G.
Arduini)
-
If not? Anything foreseen to equalize the bunches?; check
with scrapers.
-
Measurements: for optimizing collisions and total integrated
lumi it is sufficient to take the sum from individual
bunches.
-
For a full analysis and optimization of lifetime, background
and stability, measurements should be able to distinguish
between bunches, for quantities like: current, beam size
(emittance), tune and lumi.
2.
Non-Gaussian distribution (or truncated)
3.
Beams not
colliding head-on: When the
beams don't collide head on, there is a loss in luminosity. For
Gaussian beams the remaining luminosity fraction is given by the
expression:
£/£0 = exp[-(δx/2σx)2
- (δy/2σy)2],
where δx and δy are the horizontal
and vertical separation between the two beams and σx,
σy the r.m.s. beam sizes. Numerical values are listed
in Table 1.
δx (σ) |
δy (σ) |
£/£0
|
0 |
0 |
1.0000 |
0.1 |
0 |
0.9975 |
0.2
|
0 |
0.9901 |
0.3
|
0 |
0.9778 |
0.4
|
0 |
0.9608 |
0.5
|
0 |
0.9394 |
0.5
|
0.5 |
0.8825 |
1
|
0 |
0.7788 |
1
|
1 |
0.6065 |
2
|
0 |
0.3679 |
2
|
2 |
0.1353 |
Table 1: Remaining
luminosity fraction from 0 to 2 σ separation (Gaussian beams).
4. Beam-beam effect (in principle should be small)
5. Longitudinal
charge distribution: un-bunched
particles or extra bunches in one beam would be counted in the
intensity as measured with a DC-BCT but would not contribute to
the luminosity. This will be observable with several
instruments: the gap monitor, comparison of fast and DC BCT, RF
pick-ups, and to some extend using BLMs. The transverse damper
can be used to eliminate such unwanted beam components.
6.
Hour glass effect: For collisions of long bunches the
luminosity decreases because of the increase in beam sizes around the
IP. This is known as the hour glass effect. It is significant if
the beta* function and the bunch length are comparable.
Nevertheless, this effect is negligible in LHC.
B. Background rejection
C. Extended halo:
halo scraping
D. Solenoid compensation (small effect at 7 TeV)
E. Need to check corrector transfer functions/hysteresis
F. Going back to ADJUST
from STABLE BEAMS once "end of physics run" declared:
-
If there were more to do than simple things, HERA always dumped
both beams, introduced a special time slot for MD (~ 1 or 2
hours) and proceeded with a train of pilot bunches ( 1 or 10
bunches of limited intensity) to perform the study. The
reason was MPS driven; one single quench or beam loss that
triggers the MPS system was much more severe and time
consuming than a dedicated well prepared study. During
luminosity operation the machine was running at the beam-beam effects
limit, and therefore, emittance growth were expected. HERA
started with a typical beam emittance (2 sigma normalized) at
the beginning of the run of 15π mm mrad, and at the end of
the run the emittance grew easily to 25 - 30. This meant that at the end
there was not much room for beam gymnastics. The beam was filling
nearly the aperture defined by the collimators, and even
steering a bit the angle at the interaction regions, this led
very quickly to beam scraping, losses at the low beta quads,
etc.
-
For simple things like orbit correction in one beam,
steering to find better vertex position, etc, they dumped
one beam and go on with the other for a while. But this
was rarely done.
-
HERA could unsqueeze both beams
to guarantee more aperture for the studies (in the case of
the electrons they could even decelerate them).
-
At Tevatron they go to MD mode often after Physics is over.
The most common studies are fairly benign, but the
experiments turn off most of their sensitive equipment. They
often do things like crystal collimator studies, separation
scans, collimator alignment, etc.
-
They have unsqueeze beams and also decelerated protons, but
this is not very commonly used. Their biggest problem is to
have both protons and anti-protons in the same beam pipe.
|