New RF Systems for the Super-ISR and Super-SPS

1
New RF Systems for the Super-ISR and Super-SPS

• Joachim Tuckmantel, CERN

Thanks to Trevor Linnecar and Eric Montesinos
2
Contents

• Hardware Considerations, an Overview
• cavities,power transmitter, transmission lines
• What exists around LHC where we stand
• f-swing, loop gain and RF power basics
• Study of Options
• Input from Elena Shaposhnikova
• Conclusion

3
CW1 Traveling Wave Cavities

• Wide f-range without tuner follow speed
  modulation at lower particle energy (SPS 10 GeV
  to 8)
• Fast, allow e.g. fixed f acceleration (ions)
• Good RF power efficiency to beam
• Long (large centralized power) not cheap
• Low gradient -gt large parasitic impedance per MV
• HOM damping not straight-forward
• Impossible (and useless) superconducting version

1 In contrast to pulsed TW as in 2-mile SLAC or
CLIC
4
Standing Wave Cavities

• High gradient less parasitic impedance per MV
• Superconducting OK (with different shape)
• Sharp resonance Wide range tuner following speed
  modulation at lower particle energy
• Slow (no tricks)

5
Multicell SW Cavities

• Less ancillaries (couplers, control-units,)
• Pass-band modes with closely grouped frequencies
• Field polarity between power coupler and
  reference antenna (compared to accelerating mode)
  opposite for at least one mode.
• With RF vector feedback (necessary) have to
  prevent auto-oscillation add sharp filter(s) -gt
  control-loop delay, gain limitation, stability
• Avoid if possible

6
Superconducting Cavities (1)

• Less wall plug energy per MV key importance LEP,
  TESLA, LHC no real importance (small system)
• Much Higher gradient much less parasitic
  impedance per MV
• Shape (fabrication reasons) lower R/Q also HOMs
• HOM couplers in any case build max. damping
• Due to low R/Q and high voltage less reactive
  beam loading detuning -gt the tilt for sc. cav.
  in LHC
• Larger QL -gt larger critical loop gain
• More sensitive to any dirt in the vacuum
  chamber
• Need cryogenic installation, He recovery, storage

7
Superconducting Cavities (2)Tuning Range

• Plunger tuner forbidden (multipacting)
• Tuning done by elastic homogeneous longitudinal
  shape deformation, today only distinction
  between labs how produce the deformation (gear,
  piezo, magnetostrictive)
• Limited tuning range due elastic limit LHC
  (stiff) 200 kHz/400 MHz, LEP (softer) 300
  kHz/352 MHz (50 kHz exploited by drive, largely
  sufficient in LEP)
• Material fatigue for (rapid) cycling machines. No
  free choice of material high RRR Nb (or Cu for
  Nb/Cu)

8
Superconducting Cavities (3)Design Frequency
Range

• Today major sc. cav. accelerators 352 MHz 3
  GHz solid experience in technology
• Cavity shape spherical size scales with
  inverse frequency (i.e. 200 MHz a big beast,
  1.5 m diam.)
• Prototype 200 MHz single cell Nb/Cu (µ
  acceleration) worked (not immediately) but below
  expectations, development suspended (done at
  CERN, sponsored by Cornell University)

9
Superconducting Cavities (4)Field Limitations

• Sc. cavities are limited by real-life defects and
  fundamental physics properties of
  superconductivity
• Todays record 45 MV/m (laboratory, th. lim.)
• LHC 400 MHz cavities 5.5 MV/m (2 MV/cavity) ???
• RF power proportional Voltagebeam-current !!!
• LHC cavity field (voluntarily) limited by
  Power Coupler Capability
• (matches well with single cell cavities)

10
Superconducting Cavities at CERN

• After development, installation, operation and
  dismantling for LEP2 (1979 - 2000)
• All first generation people are retired or work
  on (completely) different problems
• There is a minimum team for the LHC 400 MHz
  modules (there is more to it )
• Today till luminosity in LHC, no serious RD for
  (sc.) cavities and ancillaries possible at CERN
  except by hiring (and teaching ?) staff.

11
High Power Transmitters (1)(Wall plug power -gt
(RF) power)

• Historically used gridded tubes
• Stray capacitances more important for higher f -gt
  limit tube size -gt limit power
• Way out for larger power Combine output of many
  tubes, e.g. in SPS system of 32  tubes of 35 kW
  (nom.) 800 kW (true) expensive, very complex
  in set up and maintenance
• (usine a gaz, come to SPS-BA3 to have a look )

12
200 MHz Power system (Philips) at BA3/SPS Each
of the 4 cupboards delivers 2x35 kW 1
MV/cavity
13
200 MHz Power system (Philips) at BA3/SPS
Hybrids to join power flux from different
cupboards
14
200 MHz Power system (Siemens) at BA3/SPS The
man-high tube amplifier (right) contributes 135
kW . 1 MW/cavity
15
High Power Transmitters (2)

• Large size of high power grid tube installations
  House them on ground level, very long
  transmission lines into tunnel
• Gridded tubes have low gain at high frequency -gt
• Need chain of amplifiers with increasing output
  power (e.g. 5 levels)
• Significant increase of loop delay -gt reduction
  of loop gain if RF vector feedback (necessary for
  high beam current application)

16
High Power Transmitters (3)

• e-speed modulation tubes e.g. klystrons
• High gain small solid state driver OK
• Scale (about) with wavelength get very big for
  lower frequencies
• Need circulator/insulator against reflected wave
• No principal obstacle to extend ranges but today
• Down from 200 MHz gridded tubes
• Up from 400 MHz klystrons, ..
• Important Industry catalogues reflects this,
  each transgression will be very expensive

17
High Power Transmitters (4)

• Some recent developments
• 1) Prototype of diacrode Thales 1 MW _at_
  200 MHz, BUT price 32 classical tubes (1000 h
  test only, suspended).
• 2) Prototype combination of several hundred solid
  state amplifiers of 300 W for 30 kW _at_ 353 MHz
  (klystron too expensive) -gt 190 kW in total
  envisaged Soleil
• 3) IOT merge tetrode and klystron (tests low
  power only)

18
Transmission Lines

• Wave guides or coaxial lines ?
• Wave guides have lower losses, less expensive
  (for LEP precision ventilation air ducts)
• Wave guides scale with wavelength
• Coaxial lines are laborious to use (2 separate
  conductors, bends), expensive
• No holy grail but
• 200 MHz Coax, 400 MHz waveguide

19
Real Life High Power RF (1)

• Today no more high power TV emitter (main market)
  (cable TV or satellite)
• 1) High power RF equipment is getting more and
  more expensive (small series)
• 2) Spare parts get sparse (we stockpile ...)
  adapt hardware for different parts Work
• 3) Industry is less ready for RD (except )
• 4) Future engineers/technicians prefer to flip
  bits instead of learning high power RF

20
Real Life High Power RF (2)

• It is very difficult to find qualified staff
• -gt Before throwing away old hardware think
  twice (not only )
• New hardware needs manpower time
• 12.5 ns bunch distance in LHC -gt
• Death of any 200 MHz (incl. capt. RF in LHC !!)
  PS, booster, ...
• -gt use 10 ns or 15 ns (check PS, )

21
What exists (1) shows the ballpark

• In SPS
• 200 MHz TW cavities (2 x 2 cavities)
• total RF voltage 7 MV
• RF power factory 4 x 800 kW (1.1 MW nom.)
• ( 2 x 2 different designs 32 or 8 tubes in
  end-stage)
• on the surface (due to size of installation)
• (Long) coaxial lines form surface down and along
  the tunnel to cavities
• No fast RF vector feedback ramping of beam
  induced voltage at head of batch causes phase
  position slip -gt 1-turn-delay feedback, feed
  forward not perfect

22
200 MHz TW cavity (1 of 4 cav.)
23
What exists (2) shows the ballpark

• For LHC
• 400 MHz superconducting cavities 2 MV (5.5. MV/m
  ) per cavity 1 set of 8 per beam R/Q45W
• (grouped 4 cavities per cryo-module)
• HOM impedance a few kW (strong HOM couplers)
• Power coupler rated 300 kW (tested gt 500 kW
  reliability)
• RF Power one klystron 330 kW per cavity -gt
• allows RF vector feedback without RF summing
• all under ground (Aleph-cavern) short delay
  (600ns)
• (Qext 75,000 sc. cav. in coast)

24
ACS 400 MHz superconducting 4-cavity module
25
What exists (3) shows the ballpark

• 200 MHz capture system 3 (short time 4) MV
• 2 x 4 Cu cavities1 0.75 (1) MV R/Q192W (built)
• Staged (data estimated) to be designed/adapted/bui
  lt
• Main HOM impedance estimates (couplers !)
• Rsh23 (247) / 15 (487) / 5 (714) kW (MHz)
• Power coupler 300 kW
• gridded tube RF power chain (hybrids) 4x60kW gt
  240 kW CW (300 kW short time)
• Position under ground (LEP klystron. Gallery)
• (short delay for RF vector feedback) not more
  space
• 1 design constraint 2nd beam at 42 cm distance
  has to pass outside

26
ACN 200 MHz capture cavity
27
Frequency swing in SPS ring
For ions read p same momentum/charge (or
B-field) as proton with
Tuner range must also cover fabrication
tolerances, changes in cooling water temp., atm.
pressure, power level (T), . Classical tuner
slowly a little bit from time to time .. Now
full stroke for each acceleration cycle
(bellows, )
28
Loop gain and loop delay (1)
Critical gain (when it really starts to
auto-oscillate) depends on round-trip loop
delay T (as short as possible !!!)
Example for 200 MHz SW (ACN) capture
cavities for nominal LHC beam current Qsyst
6000. If Power factory on surface T300 m
(ampli.)210-6 s gcrit ltlt 7 g 3 usable
Rsh,0192W60001.2MW -gt 400kW (per
installed cavity of 0.75 MV) (1-turn-delay
feedback (filter!), feed forward SPS close to
limits) Higher beam current -gt set lower Qsyst-gt
lower gcrit
29
Loop gain and loop delay (2)
RF vector feedback to reduce fundamental mode
impedance RF power factory has to be under
ground close to cavities In SPS tunnel only
possibility (except new excavations) ex-UA-1
or ex-UA-2 caverns Infrastructure ? High
Power Transformer AC input feeds, cooling
water, ventilation if superconducting He
recovery (refrigerator can be partly on surface)
30
RF Power Basics (1)

• Major problem with strong RF vector feedback
  (as necessary in LHC / new injectors) and a beam
  with gaps
• Not a smooth beam but reactive beam loading to
  be compensated for two states - beam on
  (train) and off (gap / rest of turn) - while
  tuner too slow to adapt. Optimize a static tuner
  setting but remainder to be done by
• brute force RF power
• In LHC only weak acceleration (0.5 MV/8), but in
  fast ramping injector several MV for acceleration
• supply beam power RF power

31
RF Power Basics (2)
complex (model) generator current
Real world generator RF power
V accelerating voltage to be obtained (defines
real axis) (R/Q), Q0, cavity constants Qext
power coupling constant I b,DC instant. DC beam
current fb normalized bunch form factor ( fb
1) f synchr. phase angle (proton
convention zero for rising RF zero crossing) Dw
(static) cavity detuning -gt half-detuning in
coast with cos(f)-1 Half detuning is not optimum
under acceleration with f?0º,180º
32
RF Power Basics (3)
LHC nominal beam current on train I b,DC
instantaneous DC beam current on train nb
1.11011 p -gt qb 1.610-19 C nb 1.7610-8
C Tinterbunch 25 ns -gt I b,DC 0.7 A what
cavity sees on train (nominal) (ltIgt
1.7610-8 C Nbun frev -gt ltIgt 0.55 A
(average nominal in LHC) fb normalized bunch
form factor 1 for point
bunches Bunch-length changes but sufficiently
short compared to l (especially at 200 MHz) fb
1 good enough (we look for estimate, not
5th digit)
33
SPS 25 -gt 150 GeV (1)

• Elena 1 400 MHz l (bucket) too short for
  capture from PS -gt need 200 MHz system
• Existing 200 MHz TW Good for frequency swing,
  7 MV deliverable, but no tight RF vector
  feedback (long delay)
• Elenas scenario(s), most demanding situation
  V9.2 MV (to contain a bunch of 0.5 eVs) while
  accelerating in 1s with 160 GeV/c/s _at_ 200 MHz
• 1 previous talk

34
SPS 25 -gt 150 GeV (2)

• 1) Exploit the existing 200 MHz TW system and add
  5th cavity to reach 9.2 MV cavity has to be
  produced as copy of the existing 4 a 5th RF
  power factory has to be established (diacrode
  ?) with coaxial lines down into the tunnel (space
  for lines, factory in BA3 ?) -gt
• Impedance in the SPS 25 higher
• no fast RF vector feedback (for future higher
  currents)
• RF power per section is limited to present value
• Trevor L. for higher current, rearrange cavity
  sections -gt 5 shorter cavities, but still new
  power factory needed

35
SPS 25 -gt 150 GeV (3)

• 2) Add 3 SW cavities (30.752.2 MV) to existing
  TW system technically no improvement, still TW
  system without RF vector feedback is dominant
  rapid frequency swing SW cavities
• 3) Remove whole 200 MHz TW system, replace by 12
  SW cavities (120.759 MV) technical advantage
  full RF system under RF vector feedback, less
  apparent fund. mode impedance rapid frequency
  swing SW cavities
• ( The 800 MHz Landau system would also need
  upgrade for higher beam current, not treated here
  )

36
SPS 25 -gt 150 GeV (4)

• 3) -gt Accelerate in SPS in 1 s -gt 160 GeV/s 1,
  frev 43 kHz -gt
• 3.7 MV/turn or 310 kV/cav (12 _at_ 0.75) -gt
  f180-24.5º
• At 0.75 MV, Q031000 45 kW Cu heating to keep
  field up
• For nominal train current 0.7 A 220 kW power for
  acceleration
• Qext 6000 (as designed for the ACN cavities)

(High V/(R/Q) -gt low Dw -gt favors superconducting
cavities) Half-detuning -gt 75 detuning 270 kW
/ cav (12) (Qext5000) About ACN design setting
errors, . need more in reality
1 DEcDp here for protons
37
SPS 25 -gt 150 GeV (5)

• Accelerate in 2 s Elena -gt 6 MV to keep bucket
  size for 0.5 eVs bunch (existing TW for nominal
  Ib no feedback)
• -gt 8 cavities of 0.75 MV can do the voltage-job
• SPS 80 GeV/s, frev 43 kHz -gt
• 1.86 MV/turn or 230 kV/cav (8 _at_ 0.75) -gt
  f180-18.1º
• Qext 5000 (Qext 6100 foreseen for the ACN
  cavities)
• 230 kW / cav (8)
• 2x nominal current (same e) -gt lower Qext
  3000 (? ! ?)
• 420 kW / cav (8)
• Preferred -gt more cavities with less V
• ( but higher impedance)

38
SPS 25 -gt 150 GeV (6)

• (Accelerate in 2 s 6 MV to keep bucket size,
  contd.)
• -gt 12 cavities of 0.5 MV can do the voltage-job
• SPS 80 GeV/s -gt 155 kV/cav -gt f180-18.1º
• 2x nominal current -gt lower Qext 3000 (? ! ?)
• 300 kW / cav (12) (OK)
• ( design limit ACN use again later)
• 3x nominal current -gt lower Qext 1500 (? ! ?
  ! ? ! ?)
• 410 kW / cav (12)
• Preferred -gt more cavities with less V
• ( but higher impedance)

39
Scaled 200 MHz superconducting 4-cavity module
(virtual reality)
40
SPS 25 -gt 150 GeV (7)Virtual option
superconducting (1)

• Scale 400 MHz modules to 200 MHz (cryost.2m
  diameter)
• R/Q45W 5.5 MV/m -gt V4 MV (2x cavity length
  !)
• Power coupler might be 2x larger -gt assume
  1200 kW rated
• (very optimistic never built cold)
• Acc. in 2 s -gt 80 GeV/c/s -gt 1.9 MV / turn -gt
  f180-18.5º
• 2x nominal current -gt 2.7 MW beam power to be
  supplied
• with some reserve (beam loading compens.,
  setting error)
• need 3 sc. cavities _at_ 2 MV/cav.

41
SPS 25 -gt 150 GeV (8) Virtual option
superconducting (2)

• 2x nominal current -gt Qext 30000 (no problem
  sc.)
• 1000 kW / cav (3 sc.)
• Problem 3 x 1 MW plants under ground ???
• (same factories as SPS 200 MHz TW in BA3)
• -gt diacrode (life time not really tested !!
  )
• Less daring option Couplers of 450 kW -gt need
  8 cavities
• (for which 0.75 MV are sufficient) sc. option
  not worthwhile
• sc. option _at_ 200 MHz
• only if high power couplers OK

42
SPS 25 -gt 150 GeV (9) Virtual option
superconducting (3)

• Total RF power about same in nc. and sc.
  3-3.5 MW
• ( 2x nominal current)
• But parasitic impedances lower for sc. (H.
  pow. couplers !)
• 3 sc. low Rsh,HOM cav. lt 12 nc.
  high Rsh,HOM cav.
• Rsh 4 MW (3 sc.) 6.9 MW (12 nc.),
  no f.b.
• (provided Qext3000 possible nc.)

43
SPS-HPS transfer (1)
Cavities do not see particle energy (speed c,
correct for frev) Same current SPS and HPS (I
hope at least ) for same acceleration -gt
identical RF system will do Beam transfer is
done at zero-acceleration -gt power reserve for
reactive beam loading compensation at slightly
higher voltage (ACN 200 MHz cavities can
withstand 1 MV for a short time !!) Elena 2
paths for transfer and further acceleration in
HPS a) accelerate with 400 MHz in HPS ( 200 -gt
400 MHz into HPS) b) accelerate with 200 MHz in
HPS ( 200 -gt 400 MHz into LHC)

44
SPS-HPS transfer (2)
Elena

• Transfer/capture at 150 GeV -gt 400 MHz in HPS
  needs
• voltage to make bunch short enough in SPS or
• capture long bunch _at_ 200 MHz transfer in
  HPS
• SPS 9 MV _at_ 200 MHz (6-7 MV done) (150!!!)
• HPS for free less MV(10) _at_400 MHz than
  needed for acc. (23)
• SPS 10 MV _at_ 400 MHz (6-7 MV _at_ 200 MHz done)
  (1/3 LHC)
• HPS for free less MV (10) _at_400 MHz than
  needed for acc. (23)
• SPS for free (4), HPS 8 MV _at_ 200 MHz
  (130 of SPS 200 MHz)
• b) Stay _at_ 200 MHz SPS -gt HPS 4 MV -gt 4 MV
• SPSHPS for free less MV than needed for
  acceleration
• need b) 13 MV _at_200 MHz instead a) 23 MV _at_
  400 MHz -gt 1 TeV
• -gt Only reasonable solution Continue _at_
  200 MHz Cu
• (sc. cavities cannot use full possible field
  power limitations)

45
HPS acceleration 200 MHz(ramp 3 s)
Keep bucket in HPS need 13 MV _at_200 MHz during
acceleration -gt 18 ACN cavities (0.75 MV) can de
the voltage-job 150 GeV -gt 1000 GeV in 3s,
frev43 kHz 6.6 MeV/turn -gt for 18 cavities
370 kV/cav -gt 500 kW beam power/cav (2x
nominal Ib) Play numbers 0.375 MV 0.19
MV(acc) -gt 300 kW/cav 35 cavities _at_ 200 MHz
_at_300 kW RF power Qext2000 (11 MW RF power under
ground -gt 20 MW wall plug, pulsed)

46
HPS acceleration 200 MHz(ramp 6 s)
Keep bucket in HPS need 7 MV during
acceleration -gt 10 ACN cavities (0.75 MV) can de
the voltage-job 150 GeV -gt 1000 GeV in 6s,
frev43 kHz 3.3 MeV/turn -gt for 10 cavities
330 kV/cav -gt 460 kW only beam power/cav (2x
nominal Ib) Play numbers 0.375 MV 0.19 MV(acc)
-gt 300 kW/cav 18 cavities _at_200 MHz _at_300 kW RF
power Qext2000 (5.5 MW RF power under ground -gt
10 MW wall plug, pulsed)

47
HPS acceleration 400 MHz(ramp 3s (6s), academic
exercise)
Keep bucket in HPS need 23 MV during
acceleration -gt 12 sc. cavities (2 MV) can de
the voltage-job 6.6 MeV/turn -gt for 12
cavities 550 kV/cav -gt 770 kW only beam
power/cav (2x nominal Ib) lower voltage -gt lager
detuning for bl. comp. -gt relat. more power 3s
ramp 40 sc. cav. _at_ 400 MHz (0.6/0.17 MV !) -gt
300KW 6s ramp 32 sc. cav. _at_ 400 MHz (0.75/0.11
MV !) -gt 300 kW

48
Conclusions (1)
Evident fact Only few cavities, copper or
superconducting, can easily supply the desired
voltage. Gradients have to be lowered voluntarily
since the power coupler cannot transmit the
corresponding RF power to accelerate high beam
currents and compensate react. beaml. Power
coupler capabilities have to be increased
considerably, Qext ?? for sc. cav. couplers RF
losses into liquid He, deconditioning

49
Conclusions (2)
For a 200 MHz system the existing RF power
factories for large power are very space
consuming -gt problem to house them close to
cavities under ground (loop delay !!) Study
compact RF power transmitter at 200 MHz

50
Conclusions (3)
To keep the superconducting cavity option open -
except copy the existing 400 MHz system as is
Re-launch superconducting cavity research
activity at CERN (The sputter activity Nb on
Cu is not yet dead, possible study for LHC crab
cavities )