High Rate Kicker Preliminary Study (Quick Update)
Tony Beukers/Tao Tang
4/8/14
2
Parameters for Spreader
Parameter Value Unit
Energy 4 GeV
Kick 1 mRad
Rate 1 MHz
Length 6 ? m
Aperture 1 cm
Pulse to pulse stability 100 ppm
Jitter When Off (compared to main pulse)
50 ppm
• “Jitter When Off” from “Post Laser Heater Diagnostic Beam-Line PRD”
• Works out to integrated field of 13.4mT-m required
• For all topologies considered, 5mT most reasonable 3ea. 1-meter sections
• Bi-polar pulse possible with two separate kickers, so not an area of extreme focus.
3
Magnet Cross Section
II
• 2-types of magnets shown to the right. C-core and Window frame. Best choice depends a bit on driver topology.
• Ferrite loaded magnet. Losses from drive field highly dependent on the core type. For 5mT, 100ns sine wave excitation at 1MHz 11W/m for 4M2, 450W/m CMD5005 (common kicker material).
• On the order of 2% change in effective µ over duration of pulse.
• Coated beam-pipe used to shield beam current from magnetic coupling to ferrite.
4
Beam Coating Losses
Figure from [2]
• Losses in conductive coating increase with beam current, high frequency components of pulse, and pulse rate.
• Eddy currents from the driver pulse go down with increased coating resistivity. . .but loss from the beam image current goes up.
• Beam image current losses (length =8.3µm, q=0.5nC):
∫ 𝑖2𝑑𝑡= 𝑞2
2√𝜋 𝜎𝑡
𝑃= 𝐼𝑅𝑀𝑆2 𝑅=𝑅× 𝑃𝑢𝑙𝑠𝑒𝑅𝑎𝑡𝑒×∫ 𝑖2𝑑𝑡
• Eddy current losses found through simulation.
• Total loss is 600W/m with 90Ω/m coating.
• Reduce loss (if necessary) with conductive strips,
5
In-Tunnel Driver
• Mount MOSFET drivers
directly on Ferrite loaded
magnets in tunnel.
• Multiple drivers reduces the
inductance of each section
so each driver can rise in
tens of ns.
• Easily redundant for longer
system lifetime.
• 5mT/m achievable goal.
Figures from LBNL NGLS paper. [2]
6
Driver Types
5 Segments/meterMOSFET Losses 520W/m
9 Segments/meterMOSFET Losses 1000W/mResistive Losses 3285W/m
3 Segments/meterMOSFET Losses 456W/m
4 Segments/meterMOSFET Losses 128W/mResistive Losses 944W/m
*Losses assume NON rad-hard MOSFET
7
Driver Types
5 Segments/meterMOSFET Losses 520W/m
9 Segments/meterMOSFET Losses 1000W/mResistive Losses 3285W/m
3 Segments/meterMOSFET Losses 456W/m
4 Segments/meterMOSFET Losses 128W/mResistive Losses 944W/m
*Losses assume NON rad-hard MOSFET
Too much power!
8
Tunnel Radiation
• Total ionizing dose causing non-recoverable failure in MOSFET is main problem.
Back-of-envelope yields ~15kRad/year.
• 1 rad-hard device rated at 100kRad (6 years). Expensive, not electrically great,
hard to get.
• Collimator reduces radiation by a factor of 10-100.
• Like to put NON rad-hard device in total dose test. Could it survive behind a
collimator?
• More input and/or modeling from RP may be useful.
Volume of electronics: d3
R
Length contributing Radiation: Lrc
𝑃=𝑊𝑎𝑡𝑡𝑠𝑚
×𝐿𝑟𝑐×𝑑2
2𝜋 𝑅×𝑋 0
9
Transmission Line Kicker
T-Line
Magnet LoadII
I+
I- Ferrite
• Loaded sections of ferrite and discrete capacitors simulate a transmission line.
• Used at SLAC in damping ring and at CERN.
• Typically used in high voltage. But for our low voltage, possible to tune magnet impedance with small chip capacitors.
10
Transmission Line Ringing
• Ringing damps to below 50ppm of the main pulse
within 1µs. Ringing reduced with more sections.
L 1
{L c }
1 2L 2
{L c }
1 2L 3
{L c }
1 2L 4
{L c }
1 2L 5
{L c }
1 2L 6
{L c }
1 2L 7
{L c }
1 2L 8
{L c }
1 2L 9
{L c }
1 2L 1 0
{L c }
1 2
C 1{C c }1
2
C 2{C c }1
2
C 3{C c }1
2
C 4{C c }1
2
C 5{C c }1
2
C 6{C c }1
2
C 7{C c }1
2
C 8{C c }1
2
C 9{C c }1
2
C 1 0{C c }1
2
V 1 TD = 3 0 n
TF = 3 0 nP W = 7 0 nP E R = 1
V 1 = 0
TR = 3 0 n
V 2 = 5 0 0
A C = 1
PARAMETERS:L c = {1 1 1 1 . 6 6 6 6 7 n / n u m }
C c = {7 1 1 4 . 6 6 6 6 7 p F / n u m }
R l = {S q rt (L c / C c )}
n u m = 2 0
V 1 V 3V 2 V 4 V 7 V 9V 6 V 8
R 1{R l}
0
V 1 0V 5V 0
L 1 1
{L c }
1 2L 1 2
{L c }
1 2L 1 3
{L c }
1 2L 1 4
{L c }
1 2L 1 5
{L c }
1 2L 1 6
{L c }
1 2L 1 7
{L c }
1 2L 1 8
{L c }
1 2L 1 9
{L c }
1 2L 2 0
{L c }
1 2
C 1 1{C c }1
2
C 1 2{C c }1
2
C 1 3{C c }1
2
C 1 4{C c }1
2
C 1 5{C c }1
2
C 1 6{C c }1
2
C 1 7{C c }1
2
C 1 8{C c }1
2
C 1 9{C c }1
2
C 2 0{C c }1
2V 1 1 V 1 2 V 1 3 V 1 4 V 1 7 V 1 9V 1 5 V 1 8V 1 6 V 2 0
VV
V
Sum of all magnet currents.
Traveling Waves
11
Conclusions
• “In Tunnel” and “Transmission Line” Kicker both still options.
• “In Tunnel” Kicker does not have a perfect driver solution, but
three topologies are possible. Need testing and more RP
input to fully evaluate radiation effects.
• “Transmission Line Kicker” looks promising according to
simulations. Some testing on the spare NDR magnet would
be useful.
• Bottom line: Both methods look feasible. Additional testing to
determine which is best.
12
References
[1] M.J. Barnes, L. Ducimetiere, T. Fowler, V. Senaj, L. Sermeus. “Injection and extraction
magnets: Kicker magnets” Mar 2011. 26 pp. Published in CERN-2010-004, pp. 141-166.
Presented at Conference: C09-06-16 Proceedings.
[2] M. Placidi, G.C. Pappas, J. Galvion, M. Orocz. “Update on Kicker Development for the
NGLS”, TUPPR095, Proceedings of IPAC2012.