Properties of X-Ray Beam Position Monitor at

Properties of XX-ray
ray Beam
Position ii Monitors i at the h
S Swiss i Light Li ht S Source
Thomas Wehrli, Paul Scherrer Institut

Abstract
• XBPM types and their application application.
• XBPM performance.
p
• Use of single blades to judge the
quality li of f an XBPM reading.
di

Undulator and Bending XBPMs
Undulator XBPM:
4 blades arranged b1 b2
like an X,
TTungsten t blades bl d b3 b4
Bending XBPM:
4 parallel blades blades,
staggered by ±Δ
b1 b2
Copper blades } 2Δ
b3
b4

Undulator XBPMs
• Determination of horizontal and vertical
bbeam position. i i
• Beam position is estimated with
asymmetries: ti
– horizontal position
– vertical position
b1 b2
b3
b4
: calibration
factor
: calibration
factor

Use of undulator XBPMs at the SLS
• One XBPM is used per beamline assuming
only angular beam motion.
– Feed forward tables: Correction of the gap g p
dependence of the XBPM reading.
– Only used to determine relative position
changes, h which hi h are used d iin a XBPM ffeedback. db k
• Approved performance: μrad.
– According to the C. Schulze (PX beamline) 25
μm shift is accepted in 25 m distance from the
source point point, 100 μm is not not.

Bending XBPMs
• The monitors are constructed to detect vertical
beam positions only.
• The staggering of the blades results in selfcalibrated
readings under the assumption of:
P ≈ P = c⋅ a +Δ = c⋅a −Δ
asym
with P the real position,
13 24
a13 =
b −b b + b , a24
b −b
= b + b
a + a
⇒ P = 13 24 ⋅Δ
asym y
a −
a
13 24
1 3 2 4
1 3 2 4

Bending XBPMs at the SLS
• Two XBPMs are used to determine angle
and position of the beam.
� Higher demands on accuracy.
XBPM1 XBPM2
source point real beam error bars
error bars
error bbars direction for XBPM1
for XBPM2
source
point Source point errors are enhanced due to the
lever arm from XBPM1 to the source point.

Source
Point
Schematic beamline design
Photon Beam
Electron
Obit Orbit
Ring
Absorber Absorber XBPM1 XBPM2
Beam
Mask
Absorber
Block
0 m 1.3 m 3.2 m 3.4 m 4.1 m 6.1 m 6.5 m
2 m

Calibration factors and Sum signal
slope k
flection
def
1/k
sym.
Δ s

Symmetrical and asymmetrical
calibration lib i ffactors are diff different
bump angle [mrad] symmetrical deflection [mm]

Single blade signals
Blade1 Blade2
Blade3 Blade4
�No obvious shadowing. Almost linear dependence
�of the blade signals on orbit bumps.
Increasing
symmetrical
deflection
Decreasing
symmetrical
deflection

Single g blade signals g
Blade1 Blade2
Blade3 Blade4
�No obvious shadowing. Almost linear dependence
�of the blade signals on orbit bumps.
Increasing
bump angle
Decreasing
bump angle

sym
P as
Horizontal Bumps have an influence
Blade1 Blade2
Blade3 Blade4
� The blades react as if we had driven a vertical bump!

Results of the calibration tests
• Auto-calibration Auto calibration of bending XBPMs
doesn’t work.
• The absolute reading of an XBPM is
doubtable. At a given working point, small
position changes still can be determined
and used in a feedback.
• Si Single l bl blades d can bbe used d iindividually. di id ll

One XBPM = four blades = four
monitors
• Correct known systematic effects effects.
– In our case: normalize the blade signals bi with the
storage g ring g current I sr : bni= n,i b i / I sr where i = 1,...,4 , ,
• Identify a certain blade signal with an
according beam position position.
– The bigger the signal the closer the beam.
• CCalibrate lib t th the single i l bl blades: d
k b = P i n,i i

What can we do with four results?
• Ignore those blades that make problems problems.
Only use the „good blades“ to estimate the
beam position, p , for example: p
1
P ∑ P
13
i
2
= ∑
i=
1
ii=
3
Remark: If all blades are good we get the arithmetic mean:
4 1
Parith = ∑
Pi
4
i=
1

Wecandoevenmore
We can do even more
• Calculate standard deviations!
4
using i all ll 2
blades σ
σ
4 1
≈ ∑(
P −P
)
3
tot i arith
i=
1
≈ ∑ ( P − P P13
)
2
2
using i bl blade1 d 1 tot i
and blade3
i=
1
i=
3
2

Why we look at single blade signals
blade1 blade2
Decay not due to
beam motion
blade3 blade4
� The single blades behave differently.

Bending XBPM feedback run (I)
blade1 blade2
blade3 top-up
blade4
injection
� Beam was shifted towards blade3, blade4 by ~40 μm
� through the XBPM feedback. P asym was kept constant.
!

The use of standard deviations
time [h]
� Correlation between vacuum pressure and standard
� Correlation between vacuum pressure and standard
� deviation of the XBPM reading.
� XBPM design was not UHV compatible. (Problem fixed now.)

Advantages using single blades
• One gets four results instead of one one.
• Signal quality can be taken into account.
– “B “Bad” d” bl blades d can bbe excluded. l d d
• Standard deviations can be calculated.
– One can judge if the accuracy of the XBPM
readings is sufficient to improve beam
stability. t bilit
• No need for bias voltage:

Blade behavior on different bias
negative g
signals
!
� Changing of the blade signals on orbit bumps doesn’t
� Changing of the blade signals on orbit bumps doesn t
� depend on the bias voltage. (But the use of P asym (Δ/Σ)
� would be critical due to a potential division by zero.)

Bending XBPM feedback run (II)
P1-P3 [μm]
• Machine in thermal
equilibrium
– Run time before
feedback start > 40 h
• Total correction by
the XBPM feedback
~ 5 μm
�Possibly y mainy y
artificial correction
time [h]
time [h]
� In case of thermal equilibrium, residual drifts of the beam
position and remaining XBPM artifacts are ± equal at the SLS.

Bending XBPM feedback run (III)
Correcttion
[μraad]
Angle
correction of an electronic artifact of the RF-BPMs
time [h]
�Additional beam motion in the marked range expected
g p
�and corrected.

Conclusion
• Use of single blades as individual monitors
can reveal l diff different effects. ff
– Vacuum pressure dependence of the readings.
• P Problem bl is i fi fixed d now. S Stabilizing bili i plate l on the h b backside k id
of the XBPMs has avoided ventilation.
–Performance Performance e o a ce of o clean c ea bending be d g XBPMs Ms in the t e
order of some micron over 11-2
2 day(s).
• Undulator XBPMs are in use for a long g time
yet.
– Detailed investigation of technical properties of
undulator d l XBPM P i is more diffi difficult l d due to gap
dependence and two two-dimensional dimensional position
estimation.