Towards wire scanner measurements with
Large Dynamic Range (> 106)
Pavel Evtushenko,
Jefferson Lab
Halo, SLAC, 2014
Outline

Motivation: Why large dynamic range diagnostics?

Experience with existing high average current FEL driver

Large dynamic range transverse beam profile measurements

Wire scanner measurements
 experience so far (CEBAF) counting
 PMT in analog mode

Signal generation
Halo, SLAC, 2014
Motivation: Why Large Dynamic Range?

there are several applications of electron LINACs under
consideration / design that require average beam power
of several MW

these applications also require very high peak beam brightness,
comparable to the one at pulse NC LINACs

similar to low average current (NC, pulses) LINACs, with high
average current LINAC a diagnostic beam mode must be used

the significant difference is the ratio of beam currents in the
diagnostic mode and full current mode

for a high average current LINAC this ration can easily be tens of
thousands

One example of an electron LINAC, which have operated with high
average current 9 mA, while driving FEL (also high average power)
JLab IR/UV Upgrade FEL.
Halo, SLAC, 2014
JLab IR/UV Upgrade: 1.2 MW beam power
Ebeam 135 MeV
average current 9 mA
(135 pC at 74.85 MHz)
Average beam power ~ 1.2 MW !
If lost beam average <P>=1 W
 possible problem for vacuum
 concern for the FEL
undulator livetime
25 μJ/pulse in 250–700 nm UV
120 μJ/pulse in 1-10 μm IR
Halo, SLAC, 2014
Lessons from high current FEL operation

when setting this machine up for high current operation, at fist
diagnostic beam mode is used, this gives “best” RMS setup, i.e.,
the setup which optimizes FEL performance and does not show
any measurable beam lose (at that current level)

then as average beam current is increased we always found that
there is a need to alter transverse match to further reduce beam
loss to allow higher current operation

important point is that, such adjustments of the transverse match
must be small

there are very small fractions of the beam, which could prevent high
current operation, but are not measured when diagnostic beam
mode is used

it also appears that, such small fractions of the beam have different
Twiss parameters than the core of the beam, i.e., transverse phase
space is not described well by a single set of Twiss parameters
Halo, SLAC, 2014
Beam dynamics driven halo generation
Measured: JLab FEL injector, intensity
difference of the peak and “halo” is about
300.
(YAG:Ce, standard CCD - 57 dB SNR
10-bit frame grabber)
Simulations: PARMELA, 3×105 particles; X and Y
beam profile and its projection show the halo around
the core of about 3×10-3.
Even in idealized system non-linear beam
dynamics can lead to formation of halo.
Halo, SLAC, 2014
LINAC’s non equilibrium (non Gaussian) beam
Propagating in drift space …
 This are not beam distributions from a
nominal setup, but an experiment that shows
complexity of the phase space distribution –
no single set of Twiss parameters describes
the beam
FODO matching section
 This is also not a halo. Dynamic range of this
measurements is ~ 500, all of this beam
later is matched to the FEL’s optical cavity
and participates in the FEL interaction
Halo, SLAC, 2014
Beam viewer wire-scanner combination
 Must have impedance shield, due to high average I
 Two diagnostics at one location
 Can use YAG:Ce or OTR viewer with easy switch
 Shielded, 3 position viewer design for FEL
Halo, SLAC, 2014
Wire scanner measurements: counting
A. Freyberger, in DIPAC05 proceedings,
Measurements made at CEBAF
 CEBAF uses wires scanners for
transverse beam profile measurements
 499 MHz repetition – very good for
counting
 One of a very few LDR beam profile
measurements examples
 Due to very low current (5 nA) made
with CW beam
 Max. counting frequency ~ 10 MHz (not
a dedicated hardware)
 Coincidence effective to reduce
background, but at the expanse of even
longer measurements time
 With CW beam measurements time of
about 15 min.
 for non-Gaussian beams 
Halo, SLAC, 2014
PMT current range

Counting can provide LDR, but is really practical only with high (~ 100 MHz) bunch frequency

For smaller bunch frequencies alternative is analog mode - PMT current measurements

Typically average PMT current must be ≤ 100 µA

With low duty cycle beam (100 µs @ 60 Hz) PMT current within the 100 µs can be much higher

PMTs with dark current of a few nA are available (low Q.E. cathode at long wavelength)

For low duty cycle systems like diagnostic mode beam, gated integrator (GI) is a for small signal
recovery

For a single GI dynamic range of 107 is very challenging and probably impossible (sub µV noise for 10 V
signals)
Halo, SLAC, 2014
Gated Integrator (GI)
Digitize
Ready
Integrate
Discharge
 PMTs with HV at the cathode and anode at the ground potential are used – this results in
negative current, which needs to be inverted
 A current mirror is used to 1. invert the current and 2. to make multiple “copies” of the
PMT current
 Two outputs of the current mirror:
#1 ~ 100 % of PMT current, #2 ~ 1 % of PMT current
Halo, SLAC, 2014
Ready
GI calibration with precision source
 Output of each GI is digitized with
16-bit ADC at 4 MS/s
 Output of a GI is available for
digitalization during charge
integration as well – better than the
gate width time resolution
 Results of GI calibration with a
precision DC current source
(Keithley 6221) in the range from 100
pA through 10 mA are shown
 RMS noise level ~ 250 µV
 Non linearity of the 1 % channel (red)
is du to nonlinear operation of the
current mirror,
too little current for bipolar transistor
 Preparing version two with FET transistor based current mirror
 The non linearity by itself is not a really a problem if the behavior is reproducible
 Calibration is to be used as a look up table
Halo, SLAC, 2014
Calibration cross-check
Halo, SLAC, 2014
GI stability
 the 1 % variation is attributed to the source stability
 GI stability is ~ 10 times better (0.1 %)
Halo, SLAC, 2014
GI + PMT test
 PMT driven by a pulsed LED, 100 us “macro pulse”
 LED is driven by pulse generator at fixed micro pulse rep. rate of 100 MHz
 Width of the LED pulse adjusted from 620 ps down to 380 ps to generate the plot
Halo, SLAC, 2014
PMT dark current measurements
 ~ 2 nA dark current is at the level of PMT specification (3 nA typical 20 nA max)
 measurements with two gates allows to subtract the dark current
 then limiting factor is the GI intrinsic noise level – equivalent to ~ 100 pA RMS
Halo, SLAC, 2014
Wire Scanner: analog mode
 an alternative to GIs are
commercially available Logarithmic
Converters
 Originally designed for photo diode
measurements
(fiber optics communications)
 Dynamic range of 160 dB and 200
dB
 Bandwidth of several MHz but
varies dependent of signal level
 Shows more complex than GI
noise behavior, which needs to be
studied further
 Calibration of AD8304 log-amp is
shows
 The calibration was made using
the same setup – DC current
source and ADC as used for GI
evaluation and testing
 4 calibration without and 4 with a
CM are shown
Halo, SLAC, 2014
Wire Scanner / Cherenkov converter
 one way to convert E-M shower
e- and e+ to visible photons
 “prompt” – much faster than a fast PMT
with few ns pulse length
 direction sensitive – to reduce
background, i.e., insensitive to particles
coming from “wrong” direction
 all reflective optics – to use wavelength
as short as possible (3 reflectors)
 output matched to a quartz fiber to
transport light to a PMT outside of the
accelerator tunnel (background
reduction)
 thicker converter generated more
photons, but limited by multiple
Coulomb scattering – beam energy
dependent
Cylindrical
reflector #1
Cone
reflector #2
Cherenkov radiator
 H20 n=1.333 > sqrt(2); Cherenkov
radiation is not trapped in the radiator
Optical fiber input
Halo, SLAC, 2014
90˚ off axis
parabolic
reflector #2
W-S signal via Cherenkov converter
How many photons would Cherenkov radiator make?
•
- 3 mm stainless steel wall;
- 50 µm W radiator;
- 200 nm – 650 nm wavelength range;
- 200 pC;
- 50 mm diameter, 125 µm thick radiator at 0.1 rad relative to the beam direction
~ 1.1×105 photons
Halo, SLAC, 2014
That is all folks. Thank you.
Halo, SLAC, 2014
back up
Halo, SLAC, 2014
FEL Injector as an example of #1 (1/6)
downstream of
the gun
Halo, SLAC, 2014
FEL Injector as an example of #1 (2/6)
upstream of the
buncher cavity
Halo, SLAC, 2014
FEL Injector as an example of #1 (3/6)
downstream of the
buncher cavity
Halo, SLAC, 2014
FEL Injector as an example of #1 (4/6)
upstream of the
SRF cavity 1
Halo, SLAC, 2014
FEL Injector as an example of #1 (5/6)
downstream of the
SRF cavity 1
Halo, SLAC, 2014
FEL Injector as an example of #1 (6/6)
downstream of the
SRF cavity 2
Halo, SLAC, 2014