Caspers/Mostacci:Proposal for the

1
Proposal for the SPL Chopper StructureF. Caspers
and A. Mostacci

• Potential difficulties related to earlier designs
  (radiation aspects, out-gassing, size)
• What could we do in order to reduce the cost for
  the puls generator or RFamplifier?
• Are there other aspects, which would make an
  alternative design interesting?
• Discussion of layout and results
• Conclusion

2
Potential difficulties related to earlier designs
(radiation aspects, out-gassing, size)

• In previous reports a beam transmission
  efficiency around 95 over the chopper section
  has been quoted
• Translating these numbers for the SPL (average
  beam power in the chopper section 4 kW) we would
  get about 200 Watt beam losses
• In the CERN-SPL chopping line (L6 meter) 1.5
  meter are occupied by chopper structures (present
  version)
• The substrate material proposed and used (Rogers
  RT duroid 6002) contains teflon out-gassing?!
• The radiation limit for teflon in vacuum is 105
  Gray (1Gray 1 Joule/Kg)

3
How to reduce the cost for the driver?(impuls
generator or wide-band amplifier)

• For a 2 ns rise- and fall-time (10-90) the
  external drivers could become very costly for a
  deflection voltage of 0-1 kV (peak, uni-polar)
  per deflecting plate.(1 kV into 50 Ohm requires
  20 A ! for 500 Volt we still get 10A)
• If we could produce deflecting structures with a
  width of, say, less than 5 cm, they would rather
  easily fit into dipole or quad magnets and we
  could simplify the design of this section.
• We gain linear (in terms of kicker output power)
  with the length of the deflecting structures
  used.
• Less voltage per kicker module would simplify the
  driver design (less signal combination thus
  better rise/fall time) and lowers the heat loss
  in the meander.

4
An alternative design for a meander structure on
aluminum oxide (1)

• Using alumina (eps9.6) on a massive metallic
  support (e.g. aluminum) would increase the heat
  transfer capability, allow for less width and has
  infinite radiation resistance
• A recent printed version on organic substrates
  with separators had a width of 90 mm for ?0.065
  (2 MeV). For ?0.08 (3 MeV) a width of 75 mm or
  somewhat less appears possible with that
  technology.
• With an alumina substrate for ?0.08 a width of
  45 mm can be reached.
• However Can we get acceptable values for
  reflection, loss and dispersion when taking
  alumina?
• Are there technological limits to produce printed
  meander structures on 50 cm long alumina strips?

5
An alternative design for a meander structure on
aluminum oxide (2)

• The concept of separating ridges would be very
  hard to implement for a ceramic substrate. Thus
  we have to check the influence of mutual coupling
  from adjacent lines on the dispersion (variation
  of group delay vs frequency)
• We have to find an optimum for the thickness of
  the alumina substrate. If it is too thin, it will
  become very fragile and the conductor losses
  increase due to the small width of the lines. If
  its too thick, the mutual coupling from adjacent
  lines increases.
• Reducing the length of the coupling section
  between adjacent lines should help to reduce
  dispersion.
• Thus a design using a symmetric double meander is
  proposed ( a simple meander was proposed earlier)
  

6

• We are considering first a double meander
  structure 20 cm long as shown below

1982mm (33 cells)
50 mm
Substrate
W
3 mm (50
)
3 mm thick
e

9.6
0.45 mm
6 mm
W
(100
)
42.5 mm
7

• Results for the 20 cm long structure (no
  additional electro-deposited metal after firing,
  0-1GHz)DC resistance near 5 Ohm( too high)

8

• 50 cm long structure, transmission measurements,
  0-600MHz, time -domain display
• The metal thickness of the meander has been
  increased to about 40 micron by electro-chemical
  deposition
  

1.65ns
1.98ns
9

• 50 cm long structure,transmission measurements,
  0-1000MHz, time-domain display (notice the
  increased ringing due to larger bandwidth)

10

• 50 cm long LANL structure, measured time domain
  response, EPAC paper 2000 by S.Kurennoy,

1 ns filter
2 ns filter
11

• 50 cm long structure,transmission measurements,
  0-1000MHz for comparison, on the right, meas
  data from LANL, Kurennoy, EPAC2000 also 50 cm
  long but on RT duroid substrate with separating
  fins

12

• 50 cm long structure,transmission measurements,
  0-1000 MHz, phase after delay correction

13

• 50 cm long structure,transmission measurement,
  0-1000 MHz, group delay variation vs frequency

14

• 50 cm long structure, reflection measurement, S11
  (step-response) in the time domain (0-1000 MHz
  bandwidth)
• notice, that the char. impedance is slightly too
  low (about 46 Ohm instead of 50 Ohm)

15

• 50 cm long structure, field coverage factor (data
  provided by S.Kurennoy) peak value on axis 0.79
  for the 50 Ohm meander in upper color bar.

Relative color scale up to 100
16

• 50 cm long structure, field coverage factor (data
  provided by S.Kurennoy), 100 Ohm structure, peak
  value0.74

17

• Conclusion
• This prototype of a 50 ohm double meander printed
  on 3 mm thick alumina is a competitive candidate
  for the SPL chopper
• Losses and dispersion are comparable to the LANL
  prototype
• The structure can stand the radiation and heat
  load (if water cooled) and should have good
  vacuum properties
• Wideband impedance matching can be achieved
• The field coverage factor is slightly lower(79)
  as compared to the LANL prototype (89 ) for 50
  Ohm impedance.
• 100 Ohm char. impedance is also possible, but at
  the expense of further decrease in field coverage
  as well as in field homogeneity