Two Devices for HINS Robyn Madrak Accelerator Physics Center (APC)

HINS - Purpose
Robyn Madrak - FNAL APT Seminar - 12/16/2008
60 MeV Linac under construction at Fermilab’s meson building
R&D Linac which will demonstrate novel technologies used for the first time
Technical feasibility proof of (front end) for
8 GeV Linac, Project X, etc.
High intensity proton source for neutrino physics/
muon storage ring experiments

Unique Aspects/Challenges
Solenoidal focusing ⇒ cleaner, axisymmetric beam
Use of SC spoke resonators
Fast ferrite phase shifters
will allow multiple cavities (and RFQ) to be driven by a single 2.5 MW, 325 MHz klystron => cost savings
Fast Beam Chopper
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Robyn Madrak - FNAL APT Seminar - 12/16/2008
Parameters
* full un-chopped 3 msec pulse at klystron-limited 20 mA

FNAL HINS
Robyn Madrak - FNAL APT Seminar - 12/16/2008

HINS Chopper – Part I
Should the HINS be extended to an 8 GeV Linac, output beam would be transferred to Fermilab’s Main Injector, with 53 MHz RF frequency
HINS Linac Bunches are spaced by 325 MHz (3.1ns)
In MI, RF frequency is ~53 MHz (~19ns)
Don’t want bunches in the 53 MHz separatrix
⇒ Chop out ~1 of every 6 bunches
Additional complication: 325 ≠ n G 53
⇒ Sometimes chop 1, sometimes 2
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Traveling Wave Chopper Structure
beam is deflected by traveling pulse (electric field) β(beam)=0.073 => must slow down pulse
Use traveling wave “meander” structure:
50 cm long
16 mm between chopper plates
2.4 kV per plate
deflection of 6mm at end of plates
6 mm
.015 ″
100Ω
20 mm
⅛″ thick substrate (ε = 9.6)
β=0.073
β=0.073
d = 16mm
V = +2.4 kV
V = -2.4 kV
~6ns
β=0.073
chopper plates
(meanders)
deflection θ = 24mRad
Pwid < 6 ns
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Chopper in MEBT
length of chopper plates: 50 cm
drift space downstream: 30 cm
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Deflection
length of chopper plates: 50 cm
drift space downstream: 30 cm
plate separation: 16 mm
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Pulser Development
Robyn Madrak - FNAL APT Seminar - 12/16/2008
We need
Two pulsers to drive the ~50 Ω meanders: +/- 2.4 kV
Max ~5.5 ns pulse width (including rise and fall time)
53 MHz rep rate
burst of 3ms @2.5Hz, or 1ms@10Hz
Programmable pulse width
(may sometimes chop 1 bunch, sometimes two)
→ Specs do not lead to an “obvious” solution

Similar Choppers
CERN-SPL
LANL-SNS

Combining lower voltage pulses?
scope:
sees ¼ of ~120V signal
(25Ω/100Ω)
Fet A
(~60V)
Fet B
(~60V)
75Ω
50Ω
100Ω
ferrite
50Ω
t_rise =1.4 ns
t_fall = 2.2ns
width = 4.1ns
2.5ns/div
repeat for 10ms
Basic Concept:
Two 60V→50Ω pulses
Combined to
One 120 V→100Ω pulse
scope
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Kentech 500V Pulser
pulse
control cards
PSU
pulse
cards
trigger and power dist cards
output
before fully assembled:
one side of combiner:
five 25Ω semirigid cable
in parallel, with ferrite
25Ω semirigid cable
Robyn Madrak - FNAL APT Seminar - 12/16/2008

~520 V pulse
5.5 ns
1 ms of pulses
@ 53 MHz
3 ms of pulses
@ 53 MHz
500V Pulser Output
(repeats @ 10Hz)
(repeats @ 2.5Hz)
June ‘06
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Pulser output
200V/div
5ns/div
3ms burst
200V/div
400μs/div
1.2 kV Pulser
Nov ‘07
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Kentech Pulsers
500 V Pulser was a success
Subsequent 1.2 kV pulser was a success
Plan: two (1.2kV→50Ω) into one (2.4kV→100Ω) output
This requires a combiner and a meander with 100Ω impedance
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Microstrips in General
phase velocity and impedance are
determined by effective dielectric constant:
Delay time (β) and Z0 may be adjusted by
Adjusting d, W, and also meander pathlength
Using only one trace or two in parallel
Adding an air gap beneath the dielectric (changes εe); can be used to tune β
view from end
view from top

FNAL Fabricated Meanders
We have pursued the following:
Use double meander design with air gap between meander and ground plane (50Ω w/no gap, 100Ω/w gap)
Using single meander
Material: Rogers TMM10i, Cu clad; ε =9.8, 18’’ long (46 cm)
Meander is formed by
routing out traces
double meander
single meander
single meander
20mm
40mm
78 mm
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Chopper Meanders
Important Aspects
Material: Outgassing ?
Impedance (avoid reflections)
Delay time (match beam β)
Pulse Behavior along length (dispersion)
Coverage Factor
20mm
40mm
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Meander Substrate
Meander traces are generated by routing out traces on Rogers TMM10i, Cu clad
Cheaper, faster than paste/firing/electrochemical deposition
Test indicates acceptable vacuum properties
Terry Anderson
@55C 8E-08 torr
0.63 L/s
(with⅓ final surface area)
bakeout
begins
nitrogen
backfill
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Double Meander, Impedance
Impedance measurements from 1 – 500 MHz
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Double Meander, Delay
Robyn Madrak - FNAL APT Seminar - 12/16/2008
Want β = 0.073 to match beam speed
Measure pulse delay in meander:
Varying distance between meander and ground plane shows sensitivity
input pulse
output; delayed

Single Meanders: Impedance and Delay
Meander 1: 95 Ω @~100 MHz
Meander 2: 100.5 Ω @~325 MHz
input pulse
nominal
0.010″shims
t delay=20.8ns
t delay=20.24ns
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Dispersion
input pulse
beginning
half way
end
single meander 2
single meander 1
“low dispersion”
input pulse
beginning
half way
end
double meander
Meanders are 18″ long
Look at pulse behavior along length using high f scope probe
2 ns/div
2 ns/div
2 ns/div
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Coverage Factor
The electric field between the chopper plates is less than that for a structure in which the entire surface is conducting
This must be accounted for in the chopper design when determining the voltage needed for the desired kick
conductor
dielectric
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Coverage Factor Measurements
High frequency probe
Tip is at top ground plane
xy stage for position dependent measurements
meander
Ground;
top ground plane
at beam height above
meander
network analyzer
port 1
network analyzer
port 2
Robyn Madrak - FNAL APT Seminar - 12/16/2008

input pulse
@ end
@end
Coverage Factor Normalization
Normalize to stripline with wide trace
Use geometry for 50Ω – convenient
For striplines
Z = 120π 2/8(ln2 + πw/4h)*
Use w = 25mm, 2h = 16mm
* R. Collin, Foundations of Microwave Engineering
Probe pickup signal (S21), 50 – 150 MHz
Robyn Madrak - FNAL APT Seminar - 12/16/2008

All Measured Coverage Factors
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Position Dependence
beam RMS size
100%
x
Robyn Madrak - FNAL APT Seminar - 12/16/2008
20mm = 0.79″

3 ms pulse
combined output
input
50Ω cable
V1→50Ω
V1→50Ω
50Ω cable
MN60 ferrite: three 11’’ OD, 4.5’’ ID, 1’’ thick cores
58 turns of ¼″ “superflex” cable
ferrite
1 ms pulse
Expect behavior to be better than this:
currently we have extra unneeded cable length
matching resistors (100Ω to scope 50Ω) add extra inductance
Combiner
2V1→100Ω
Test combiner by splitting and recombining (using our 500V pulser):
Vout = 95% Vin
scope

input
ferrite
46 turns of ⅜″ superflex around five 1″ MN60 cores
1700 V →100 Ω
1200 V pulse → 50 Ω
500 V pulse → 50 Ω
combiner
100 Ω
meander
structure
Kanthal 100 Ω
High f probe, ⅜″ away
Combiner Optimized
200 μs/div
1ms pulse
5ns/div, 1600 V
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Heating in Meander
Current in meander will be 2.4 kV/100 Ω = 24 A
Need to test heat/current handling capacity
Use 1ms/3ms pulses: (24A)2 x ⅓ x 5.3 = I2test ⇒ Itest = 32 A
actual
pulsed
current
chopping
DF
Skin
depth
factor
Fuses @ 180A, 3ms pulse
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Chopper: Summary
We have built prototypes for the necessary components for the chopper: the pulser, meander structures, and combiner
The prototype pulsers from Kentech performed to specs;
For a complete chopper system we need 3 more
We have built a combiner suited for combining these fast pulses
We have explored different layouts for the chopper plates (meander structures). The higher coverage factor single meander is the best candidate.
For more details, see proceedings of Linac’08 :
R. Madrak et al., “A Fast Chopper for the Fermilab High Intensity Neutrino Source (HINS)”
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Aside: Application of Chopper R&D
to the current accelerator
Initially explored the option of using a few fast, 1 kV FETs from DEI for
Chopper pulser
Realized these could be used for notching in the 750 keV line:
create a notch for booster kicker rise time
(minimize losses)
This effort was begun initially in collaboration with Doug Moehs
(first attempt was chopping in the source)
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Combining three DEI FETs
from DEI/IXYS RF
Use the same scheme as HINS pulser,
combining three ~1kV → 16.7 Ω signals
(x 30 = 50 Ω)
~40ns pulses 2.2μs spacing
Burst of 15 pulses, repeat at 15 Hz
Two pulsers: ±1.9kV
1.9 kV,
~40 ns wide pulse
(on test bench w/60 dB atten)
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Plates: 0.9″
spacing
W1
Notching Plates in 750 keV line (H-)

In Linac after tank 2
ΔV plates = 2.9 kV
~100 ns wide notch
50 ns/div
20 ns/div
In Linac after tank 2
ΔV plates = 3.8 kV
~100 ns wide notch
In booster
~40 ns wide notch
In booster
Notching Study
~40 ns wide notch
B. Pellico, R. Tomlin
B. Pellico, R. Tomlin
400 ns/div
40 ns/div

Robyn Madrak - FNAL APT Seminar - 12/16/2008
Part II – Vector Modulators

HINS 325 MHz RF
Pulse Transformer& Oil Tank
IGBT Switch & Bouncer
CAP
BANK
10 kV
110 kV
Charging
Supply
300kW
MODULATOR
325 MHz
2.5 MW
1ms@10Hz
or
4ms@2.5 Hz
WR2300 Distribution Waveguide
I
Q
M
I
Q
M
I
Q
M
I
Q
M
I
Q
M
Fast Ferrite
Vector
Modulators
RF
Couplers
I
Q
M
I
Q
M
I
Q
M
I
Q
M
I
Q
M
I
Q
M
500kW
I
Q
M
I
Q
M
I
Q
M
I
Q
M
I
Q
M
I
Q
M
10kV
50 kW
circulator
M
E
B
T
R F Q
S
S
R
S
S
R
S
S
R
S
S
R
H-
Medium Energy
Beam Transport
Copper Cavities
Radio Frequency
Quadrupole
Cryomodule #1
Cryomodule #2
TOSHIBA E3740A
independent phase and amplitude control in each cavity
Robyn Madrak - FNAL APT Seminar - 12/16/2008

How it Works
In a coaxial line filled with some dielectric (ε,μ)
v = c/√εμ
We vary μ and thus v and phase by varying
H applied to the ferrite.
H
I
ferrite
outer
conductor
inner
conductor
slot
in outer
conductor
supplied by
solenoid
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Robyn Madrak - FNAL APT Seminar - 12/16/2008
Operates in full reflection mode (end is shorted)
Use solenoid along with shifters: phase of reflected wave determined by μ of ferrite (μ depends upon applied H Field)
Ferrite is Al doped Yttrium Iron Garnet (YIG) – TCI Ceramics AL-400
Required rate: 1º/μs
Power Rating: ~50kW (Room Temp Cavities)
or ~500kW (RFQ)
Fermilab’s Ferrite Phase Shifter

Modulates phase and amplitude independently:
With ΔΦ = (Δφ2 - Δφ3)/2
Φ = (Δφ2 + Δφ3)/2
90 Degree
Quad Hybrid
Input
(split between ports 2 and 3)
Output ←
(Cavity)
Δφ2
Δφ3
circulator
Vector Modulator:
Output power ~ cos2(ΔΦ)
Phase shift ~ Φ + 3Π/2
Phase
Shifters
shorted
end
solenoid
flux return
slot
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Two Phase Shifter Types
For Cavities (~75 kW):
1.5″ OD X 0.65″ ID X 5″ long garnet
For RFQ (~500 kW):
3″ OD X 0.65″ ID X 5″ long garnet
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Shifter Design Details
Center conductor: shrink fit during assembly
Use quarter wave matching section
(for 50Ω)
Outer conductor has 0.020″ slot (length = 9″) to reduce eddy currents
(gives faster response)
solenoid
(12 awg wire around G10)
flux return
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Other VM Parts
hybrid for 1⅝ vm: Dielectric
circulator for 1⅝ vm:
Ferrit-Quasar
circulator load:
5kw CW
water cooled
Altronics
6 ″hybrid for RFQ vm: MCI
Filled with SF6 to prevent sparking
Robyn Madrak - FNAL APT Seminar - 12/16/2008

useful phase shift range ~120 deg. (loss < -0.2 dB)
Gyromagnetic resonance
(lossy)
@ 2.8 MHz/Oe
Phase vs. Applied Field
325 MHz
Low Power meas:
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Small Signal Frequency Response
Open loop bandwidth: 15 kHz
> 35 kHz w/feedback
Response
(mixer)
solenoid I
Program, 10 kHz
0.1 ms
30 deg.

Slew Rate
Phase Shifter Slew Rate:
(above resonance)
6 deg/μs
Current risetime limited by supply output, solenoid inductance
Fast 300A power supply thanks to
Brad Claypool, Steve Hays, Howie Pfeffer
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Robyn Madrak - FNAL APT Seminar - 12/16/2008
Beam Loading - Simulation
Cavity 6
Starting and stopping the compensation 4 usec prior to beam arrival time
Beam current 26 mA
phiS = -45 deg
Results courtesy
Julien Branlard

Meson Building Test Facilities
325 MHz
RF Test Cage
Please do not
feed the animals
2.5 MW klystron
First room temperature cavity
vector
modulator
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Meson Building Test Facilities
Testing the RFQ vector modulator
Testing the 1⅝″ vector modulator
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Power Capabilities
phase ~ (Δφ1 + Δφ2)/2
here, both shifters’ solenoids driven by one power supply
(Δφ1 = Δφ2)
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Power Capabilities
1⅝ VM for cavities:
good to >75 kW
Shifters alone could be used in a 200 kW VM if used with oil in ferrite part of coax line
(higher power quad hybrid & circulator would be needed)
RFQ VM
Shifters and Hybrid filled with SF6: Good to > 500 kW
Current hybrid is 6″
Stripline “500 kW” Circulator failed: Getting a new one from Ferrite
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Vector Modulators: Summary
The 1⅝″ vector modulators can operate well up to 75 kW (more than needed for the RT cavities)
The RFQ vector modulator elements:
phase shifters good to ~600 kW
hybrid good to ~600 kW
initial circulator failed; new Ferrite™ model on order
The speed of the response for the cavity shifters is 6X as fast as the original spec
bandwidth > 35 kHz for a first attempt at feedback
For more details, see proceedings of Linac’08 :
R. Madrak and D. Wildman, “High Power 325 MHz Vector Modulators for the Fermilab High Intensity Neutrino Source (HINS)”
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Robyn Madrak - FNAL APT Seminar - 12/16/2008
Conclusions
HINS is a key part of Fermilab’s Accelerator R&D program, and likely a key part of its future physics program
We have demonstrated the workability of two of its more challenging components

Backup Slides
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Kentech 500V Prototype Pulser Scheme
10 pulse cards: 50V→5Ω
5 FETS/card (in parallel)
each FET drives 25Ω
center conductor
outer conductor
5Ω
5Ω
5Ω
5Ω
5Ω
5Ω
5Ω
5Ω
5Ω
5Ω
25Ω
25Ω
pulse control cards
output: 500V→50Ω
25Ω semirigid cable
with ferrite
five 25Ω semirigid cable
in parallel, with ferrite
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Coverage Factor, Meanders
single meander signal to 500 MHz
double meander (100Ω)around 100 MHz
single meander (100Ω) around 100 MHz
Which gives*:
Robyn Madrak - FNAL APT Seminar - 12/16/2008
*Relative to normalization measurement;
After correcting for impedance difference and reflected power

Heating in Meander
Final Pulse: 2.4 kV→Z= 100Ω, 1ms@10Hz or 3ms@2.5Hz, Chop ≤ 30%
Meander Traces: 70μm thick, R(DC) = 2.7 Ω
- Or -
Measure power spectrum of pulses; normalize to (2.4kV)2/100 Ω x ⅓ x 0.01 = 192W
Convolute with S21 thru meander; this give Pdiss = 46 W
I2test x DFtest x 2.7 Ω = 46W ⇒ Itest = 41 A
actual
pulsed
power
chopping
DF
DF
1ms @ 10Hz
⇒ With safety factor, test at 50 A
low power, spectrum analyzer
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Notching Study
+1.9 kV
-1.9 kV
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Using one or two single power switches:
from DEI/IXYS RF
min width pulses @20MHz:
20ns/div
1ms/div
→ cannot get to a narrow enough pulse …
With two switches:
Pulses are narrower, but
Still not narrow enough
4ns/div
Robyn Madrak - FNAL APT Seminar - 12/16/2008

Using one or two single power switches:
The DEI FETS can be used to make a very narrow pulse by charging cable in the drain
But in this case we cannot attain the desired 53 MHz rep rate
20ns/div
4ns/div
With a slightly lower current version of the switch:
Robyn Madrak - FNAL APT Seminar - 12/16/2008