Transcript workshop 2016x

Commercial high-power X-band
experience: klystron/modulators,
and waveguide vacuum flanges
CLIC Workshop January 2016
G. McMonagle
(on behalf of Xbox team)
•
•
•
•
Klystrons
Modulators
Klystron Drivers
Flanges
KLYSTRON
SLAC Collaboration
XL4 to XL5
2008-2010
KLYSTRON
XL4 USA frequency 11.4 GHz only high power X band tube in operation
No commercial option
Long experience at SLAC in building and using these tubes
Collaboration between CERN and SLAC to build 12 GHz design (XL5)
PSI and Trieste joined collaboration
5 tubes ordered
“Best efforts” collaboration, no guarantees
1st tube installed at CERN in 2010
KLYSTRON
Parameters
Specifications
units
RF Frequency
11.9942
GHz
Peak RF power
50
MW
RF pulse length
1.5
µs
Pulse repetition rate
50 (100)
Hz
Klystron voltage
410-470
kV
Micro perveance
1.15E-6
Good collaboration
But some problems
needed extra payments
due to best efforts
NO GUARANTEE
KLYSTRON
Call for Tender for commercial solution
Limited replies
Contract to CPI
2010-2014
SLAC Collaboration
XL4 to XL5
2008-2010
KLYSTRON
CPI VKX-8311A
Same specification as XL5
Ordered 2 new tubes plus option of buying a 3rd
1st tube successfully tested at SLAC
Delivered to CERN and installed in XBox1
Tube has been performing fine since installation
2nd tube also successfully tested at SLAC
Delivered to CERN and installed in Xbox 2
During SAT tests problems arose with tube performance
KLYSTRON
First diode tests (no RF) of CPI#2 tube showed that 0.7 GHz gun
oscillations starts at about 240 kV and even generate RF power
from the input cavity:
Fortunately, going higher in voltage, the instability zone moves
towards the rise/fall time, so the plat top can be used now.
No margin to reduce
Klystron operation voltage
Run at nominal even when
low power required to avoid
oscillations on RF
KLYSTRON
Why were oscillations not visible in SLAC tests ?
Consultation between CERN and CPI concluded it must be due to geometry
in the high voltage tank, CERN took reservation on warranty of tube
performance.
CPI manufactured pieces for 3 different solutions to try and resolve the
problem
1. Special plates with ferrites to be mounted on walls of modulator tank :Improvement but did not get rid of all oscillations but could run at a
lower voltage with good flat top
2. Plates to change geometry (reduce volume and mimic shape) to match
Xbox1 modulator :- No improvement
3. Solution 2 with ferrites :- Slightly better than 2 but 1 was best.
Lost 3 weeks of operations with these manipulations, CPI wanted to
investigate more but we decided option 1 is sufficient to allow us to
run
KLYSTRON
Are there any oscillations on 1st tube?
Yes but at within a small voltage at around 200kV and on falling edge
Above and below these voltages tube is fine.
3rd tube has been ordered and will be tested in a new Scandinova
modulator in CPI factory (not at SLAC)
FAT next week
First indications from CPI are that there are no oscillations like 2nd
tube but may have similar issue to 1st, will find out more next week !
KLYSTRON
Call for Tender for commercial solution
Limited replies
Contract to CPI
2010-2014
2013-2015
Xbox 3
Combination of lower
peak power klystron with
high average power,
turnkey contract awarded
to Toshiba for 4 klystron
modulators
SLAC Collaboration
XL4 to XL5
2008-2010
KLYSTRON
Toshiba E37113
Peak power: 6 MW
Beam Voltage: 150 kV
Beam current: 90 A
Average power: 12.4 kW
Efficiency: 47.5%
RF pulse width 5µs
Rep rate 400Hz
KLYSTRON
All 4 tubes have been successfully installed in the modulators at CERN
in October 2015
All have pulsed to nominal parameters in diode mode
First tube has also been operating with RF
Waiting for delivery of RF waveguide components, LLRF and klystron
drivers
RF commissioning hopefully starting March 2016
MODULATORS
Xbox 1
2010
MODULATORS
Xbox 1 modulator
CEA Saclay collaboration financed modulator
Area chosen for modulator installation had limited footprint
Therefore size of modulator was critical and CEA also decided to go for solid state technology
Contract awarded to Scandinova
ScandiNova solid-state modulator (Ser. #1)
for powering the XL5 klystron
- Designed for 400kV, 300A, 3.25us HV
pulse width FWHM, 1.5us RF pulse
width at 50Hz repetition rate
- Max. 450kV, 350A, 3.5us, 50Hz
- Specified 0.25% HV ripple and 0.1% HV
repeatability
- Compact design (3x2m2) including HV oil
tank and klystron
- Primary voltage 1400V max.
- Integrated control and interlock system
- Integrated vacuum and solenoid power
supplies for klystron
MODULATORS
Factory test done into dummy load (no klystron available)
Normally modulator delivered to client in fully assembled
unit (as in factory) but due to space restrictions had to be
disassembled the remounted at CERN
SAT tests with dummy load no klystron available
Did not achieve nominal parameters until 2015 with CPI
klystron installed (arcing in SLAC klystron gun)
MODULATORS
Issues with 1st modulator
Calibration measurements.
In factory no problem into resistive load (so we thought)
In SAT, modulator was commissioned with dummy load again as klystron was not delivered
yet.
When klystron arrived was difficult to correlate measurements from SLAC to
measurements at CERN, indications were that the perveance of tube was not correct.
Eventually found that SLAC measurements were erroneous by 10% (confirmed by Trieste
during their acceptance of tube #3)
MODULATORS
MODULATOR AND RF SYSTEM, OVERVIEW
External Water
Cooling System
Need a CT on high side
For future modulators
Water Cooling Distribution
L
L
1
L
2
GN
3
D
L
230VAC N
1
50Hz GN
D
Initially only SN built
HT measurement
Later commercial
calibrated solution
Tri
I
g
installed
puls
Local RS232Ue pul
Remote se
RS232
P
o
w
e
r
D
i
s
tr
i
b
u
ti
o
n
U
n
it
S
W
#1
3phase
480VA
C
HVPS
#1 - N
3x18
5A
Solenoid
>10
L/min
Modulator
>45 L/min
400VAC
50Hz
Water Flow
Interlocks
Collector
>120
L/min
Body
>10
L/min
12GH
z
Pulse
Transform
er Oil Tank
S
W
#2
Current Interlock
RFAmplifie
r
RF 3GHz
RFSource
Core
Bias
DCPS
C
T
1-phase
120VAC
Control System Unit
K2 Solid
State
Modulator
Vacuum
Interlock
CVD
S
W
#3
S
W
#n
Ion
Pump
PS 5X
Solenoid
PS #1-3
Filament
DCPS 030A
Waveguide
pressure Int., Arc
Int, VSWR Int.
Safety
Interface
Current measurements on low side of pulse transformer secondary
Not ideal due to stray capacitance giving overshoot on measurement
Difficult especially when starting up in short pulse to say where real peak is situated
MODULATORS
GUI interface was not very user friendly
No real time and date stamp on interlock occurrences
Some switch failure and reliability problems
May have been due to many tube arcs in the XL5 tube
Flat top on voltage difficult to achieve
many iterations on changing tuning elements on each switch
Flat top only really perfect at nominal voltage
Had some issues with switch units when at nominal voltage
Beware if you have to increase charging voltage to keep performance
Some switch units may not be triggering and overloading the others (no indication of non
triggering switches)
When in operation the modulator performance is very good
MODULATORS
Xbox 1
Xbox 2
2012
2010
MODULATORS
Xbox 2 modulator competitive tender from CERN
Scandinova awarded contract
Measurement requests implemented into new unit (Calibrated CVD, CT on high side), but
not real time and date stamp for intelocks
New design of switch unit (unfortunately not backward compatible with 1st unit)
Different spares needed.
3 charging power supplies integrated (2 in first)
Unfortunately we decided to commission this modulator with the XL5 as there was
pressure to put CPI tube #1 in Xbox1.
Resulted in getting modulator to work correctly but not at full pulse width due to tube
limitations.
Waited for CPI #2 to arrive before full commissioning and tuning of pulse shape
Apart from a few teething problems and of course tube oscillations system running very
well
MODULATORS
Xbox in building 150
More space, overhead crane available
Manipulation and operation a lot easier than Xbox1
MODULATORS
Xbox 1
2010
Xbox 2
2012
Xbox 3
2014
MODULATORS
Lessons learnt from previous two projects
Price enquiry to klystron manufacturers for turnkey solution
Klystron contract awarded to Toshiba who subcontracted modulators to Scandinova
Tube and modulators arrived at CERN after full SAT test with klystron
Very quick and successful integration at CERN
MODULATORS
Modified K1 ScandiNova modulator
The illustration of new 2 cabinet
concept that will be adopted for 6 MW
Toshiba klystrons.
 Doubled width oil tank. To facilitate installation
of the Toshiba klystron which has rather wide (
0.7 m) solenoid.
 Additional cabinet (comes for free). It can be
used for Klystron RF driver amplifier, Solenoid PS,
Ion Pump PS etc.
 New Control System that will simplify integration
of external parts and offer a lot of new features.
 Flexible design (klystrons positioning) to
Modulator parameters (max.)
minimize the length of RF waveguide circuit:
Peak RF power: 8.0 MW
Pulsed voltage: 175 kV
Pulse current: 115 A
Average power: 50 kW
Pulse length (flat): 5sec
Rep. rate: 400 HZ
1.6 m
1.3m
for 6 MW Toshiba klystron
MODULATORS
Xbox 3 in building 150
Please sign up for visits later in the workshop
MODULATORS
New improved performance in modulators due to new control system and GUI
FPGA programming of switch triggers and charging power supplies
allows easier tuning of flat top by delaying triggers
Also allows flat top to be tuned at lower voltages of operation not just at nominal
New GUI an excellent improvement
Real time and date stamp for interlocks and events
Interlock levels and settings a lot more user friendly
MODULATORS
Final words on modulator performance, measurements made on new S band K2 at SN
Same performance on Xbox modulators
K2-SYSTEM PULSE FLATNESS <0.4%
DC-LEVEL STABILITY K2-SYSTEM
USING PRECISION MULTIMETER
<0.002%
K2-SYSTEM PULSE-PULSE STABILITY <0.04%
USING OSCILLOSCOPE
Klystron Drivers
For Xbox 1 and 2
Using TWT 3kW pulsed driver to provide power to Xbox 1 and 2
~ 700w needed but allows tube to placed considerable distance away from klystron
Non member state
Expensive
Protected technology , end
user statements needed
Designed for high duty factor (high pulse rep)
Using at 50 Hz peak power drops off after few thousand hours operation
Recycling at high prf necessary to recondition tube
Recently seeing glitches in pulses causing problems in pulse compression
Finite lifetime of tubes, whole unit needs to be sent back for change (long and
expensive)
Klystron Drivers
Xbox 3
Development of solid state driver with a member state company
Prototype built
achieved 310W output power (initial goal >400W)
Tested on Xbox3 A
RMS phase and amplitude stability is very good with 0.17 degrees and 0.14%
respectively
Peak power limited due insertion loss by coaxial combination and isolator
inside unit.
4 new units ordered with waveguide combination and output
>1.1dB extra gain expected to achieve >400W output (delivery February 2016)
If successful two units can be combine to power klystrons in Xbox 1 and 2
Still cheaper than USA TWT
FLANGES
Xbox 1 and Xbox 2
Using SLAC type male/female type flange
Expensive to manufacture flanges and gaskets
Attention to detail on integration of “directional” components needed when
choosing flanges.
Experience in Xbox 1 shows copper plating on flanges is necessary to reduce RF
losses and RF heating
FLANGES
New sexless flange manufactured for Xbox3
(design agreed between laboratories that use x band technology)
FLANGE
GASKET
Initial testing of flange at SLAC
Initial problem with copper plating, now resolved with new technique at CERN
(for more info see Serge Lebet)
When mounting it was found difficult to align flanges and keep the flange gasket
in place.
New mounting jig developed prototype ready this week
Summary
•
New technology, fast learning curve for CERN
–
•
•
And manufacturers
Good collaboration between laboratories and commercial suppliers allowing fast
development in the field
Other competitors coming on market especially in solid state modulators e.g.
Ampegon (SwissFEL) and Jema (University of Valencia)
2010
2016
Thank you for your attention
Xbox operations at CERN
XFEL’S?
Medical apps?
University test stands?
Xbox3 technology in industry?