Download Andrew Moss RF testing in the MICE hall

MICE RF distribution system
Andrew Moss
ASTeC
Collaboration meeting, DL,
19th – 23th October 2008
Collaboration meeting, DL, 19h – 23h October 2008
Contents
• RF distribution
– Advice on using the triode circuits
•
•
•
•
RAL
CERN
Brookhaven
Fermilab
– Coax power handling
– Waveguide system design
– discussion
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
Triodes
• Advice has been sought from RAL, CERN, Fermilab
and Brookhaven on how triodes of this type are
connected to accelerating structures
• In general, the triode is very tolerant of reflected
power
• However methods are used in certain cases to
increase the isolation between amplifiers and cavities
• Also methods are used to balance RF power to
multiple accelerating structures driven by one active
device
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
RAL
• On the ISIS linac, the TH116 triode circuits use 12 inch output
coax
• The amplifiers use an output tap on the grid circuit to take off
RF power, this is considered the best approach to maximise RF
power transfer
• The RF power is taken through a 12 inch coaxial motorised
trombone to set the phase for each linac tank
• No device to isolate the reflected power is used, although the
trombone phase shifter may well be shifting reflecting power
and reducing standing waves
• ISIS runs at 50Hz 1mSec ? Considerably higher PRF than
MICE will
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
CERN
• At CERN the TH170 triode is used
• This is a direct plug in replacement for the 116 but has
different characteristics in terms of gain and output power
• The CERN amplifiers use 2 output taps (6 inch coax) to couple
power out of the amplifier circuit
• A multitude of pulse regimes are used with these circuits
• The circuits are connected to linac tanks without reflected
power protection
• CERN are doubtful that > 1.8MW can be extracted from the
circuits
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
Brookhaven
• At Brookhaven triodes are used to power Linacs
• The RF distribution system includes high power
trombones to move reflected power away from the
amplifier and hybrids are used to split power to
differing sections of the Linac
• They would recommend that the MICE RF systems
follows this same format, it allows for more isolation
and adjustability of the coax distribution system
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
Fermilab
• At Fermilab the MTA is being operated now with the MICE
cavity
• System consists of an amplifier circuit some distance away,
with a 9 inch coax feed line
• In the cavity test area a motorised trombone is used to set
phase into the cavity, and also to ‘move resonance away from
the amplifier’
• Power is split close to the cavity using a 3dB splitter and the
coax size reduced first to 6 inch and then to 4 inch coax, which
is pressurised with SF6
• Fermilab recommend the same approach for MICE, however
they also recommend using a hybrid splitter rather than the
3dB type, this will isolated the cavity couplers to 30dB with
respect to each other
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
MICE 116 Amplifiers
• The amplifiers from Berkeley have a have single
9”coax output, this is a capacitive output coupling,
not sure what the power limit of this style of output is
• 2 CERN style large amplifiers have twin 6” coax
outputs, these are a direct grid tap , generally a better
way to couple power out , however CERN are
doubtful at achieving more than 1.6MW without
issues in the coax lines
– when are these going to arrive ?
– May need to adapt them to suit our systems/needs in terms
of controls and electrical interface
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
MICE cavity
1.2
9
0.8
0.6
7
.
8
1.0
6
Gradient / MV/m
Reflected power per coupler /
MW .
• The MICE cavity
will have a fill time
of <200uSec
• This means that at
the start of each
pulse, total
reflection of power
will occur
• Cavity filling time
is shown
• A triode can
tolerated this
reflection
Pr
5
U
4
0.4
3
2
0.2
1
0.0
0
50
100
150
200
0
250
Time / microseconds
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
Coax
• Coax system can be designed with minimum phase
error at cavity inputs
– Cavity input couplers used to match up coax feeds
– ‘Flexible’ sections used to take up misalignments in coax
– Copper or Aluminium section possible, aluminium easier
to work with – lighter !
– Cavity couplers need SF6 (used on MTA couplers) to stand
off RF voltages
– For the MICE coax system suggest using nitrogen
pressurised to 1 bar, coax system will provide a slow leak
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
RF power into Arcs
• RF power likes nothing better than to dump itself into
an arc !
• Once an arc has imitated, all of the power will
dissipate into that area potentially causing damage to
surfaces, and if the problem is not detected, burning
of the coax system will result and possibly ‘holes’
• Filling the coax system with pressured gas (nitrogen,
SF6) will increase breakdown resistance and provide
a useful interlock to stop running in the event of a
‘hole’
• Other interlocks to protect against these issues
include arc detectors and RF power level comparators
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
Coax power handling
Peak power level
3MW
From Mega Industries
Collaboration meeting, RAL, 7th – 10th October 2007
Average power
130kW
Andrew Moss
Spinner 6 1/8 coax parameters
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
Peak power handling
From Andrews
More conservative numbers*
1.5 MW
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
Waveguide feeder
• Cavity is fed from both sides, flexible sections to take up
alignment errors, couplers can be used to adjust incoming
phase (by how much?)
• Probe style phase shifters could be used to adjust further if
necessary, limited range but remote control possible –
relatively small
• Advice suggests that hybrid power splitters be used rather
than 3dB types, greater isolation between the output arms
and hence the cavity couplers, these items are not that
expensive
• Advice suggests that a trombone phase shifter be used on the
output of the tube to move any resonance out of band, these
are large devices
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
Feeder design
• Using this design is more costly, larger in terms of
space required but a safer design - easier to set up,
more adjustable, greater tolerance to problems
• This will more closely follow the MTA design, which
is proven, expert advice says it should be done this
way
• The hybrid will isolate the amplifier in a limited
manner, phase combined signals coming back at the
hybrid will be split into the load and back into the
input arm – advice suggests triode not concerned with
reflected power anyway- may simply help to make
system more robust
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
MICE hall GA for DL/LBNL Amplifiers
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
MICE hall GA for CERN Amplifiers
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
Block diagram of amplifier system
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
Waveguide distribution
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
Hybrids and loads
• Using hybrids requires a reject load, this will need to
be rated to 1MW for voltage stand off = large device!
• Possible to find another type, none find so far
50/50 Water/glycol coolant
12 inch input port
83 inch’s long
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
RF testing in the MICE hall
• RF testing will be possible with the MICE hall
searched and interlocked, RF personnel on limited
access
• For amplifier testing, coax switches will be used to
isolate cavity and direct power into the test load
• These switches will be interlocked by the PSS
system, only allowing RF test mode when all cavities
are removed from the system
• Possible that one load can be used with a combiner,
or a pair of loads to simulate cavity
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
LLRF
• Possibility of using Larry Doolittle’s LLRF4 board for
monitoring and controlling mice amplifier chain
• In house development possible, however DL RF group has a
lot to do already
• Outside help possible at a cost, need to check on more details
• Money is in the budget for LLRF at ~ 20K per cavity system
– This would provide all the measurement of power levels, digital
interlocks and controlling functions for the amplifier chain
– Power supplies are expected to look after them selves
– May leave a few interlocks for the main control system that would be
hardwired into the amplifier enable line
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss
Discussion
• Amplifier issues
– What is needed for the CERN contribution
• Hall issues, plan to finalise hall layout by July 09
• Power supply issues
– Simply a matter of work effort to finish them
• Waveguide issues
– The current design is back to the more complex, larger system with
trombones and hybrids/loads, will it fit ?
– On the positive side this will provide a more robust system that has
already been demonstrated at the MTA
– Cost, all of the components are in the original budget
• Planning issues
– Plan for 2009 is tight but we thin achievable
– Electrical plan for stage 6 available, includes RF amplifier milestones
Collaboration meeting, RAL, 7th – 10th October 2007
Andrew Moss