LLRF Control System

1
LLRF Control System

• Outline
• Scope
• Requirements
• Design Considerations
• Evaluation
• System drawings
• How this fits into beam-based longitudinal
  feedback
• Conclusions

2
Scope

• This document summarizes the design of the LCLS
  LLRF control system design including its
  interface with the beam-base longitudinal fast
  feedback.

3
Scope

• The low level RF controls system consists of RF
  phase and amplitude controls at these locations
• Laser
• Gun (Klystron 20-6)
• L0-A, a.k.a. L0-1 (Klystron 20-7)
• L0-B, a.k.a. L0-2 (Klystron 20-8)
• L0 Transverse cavity (Klystron 20-5)
• L1-S (Klystron 21-1)
• L1-X (Klystron 21-2)
• L2 - (Klystrons 24-1,24-2,24-3) to control avg
  phase/ampl of L2
• L3 Transverse cavity (Klystron 24-8)
• L3 - 2 sectors of klystrons, S29S30

4
Requirements (1)

• Meet phase/amp noise levels shown below
• Table 1. RMS tolerance budget for lt12 rms
  peak-current jitter (column 3) or lt0.1 rms final
  e- energy jitter (column 4). The tighter
  tolerance is in BOLD, underlined text and both
  criteria, DI/I0 lt 12 and ?DE/E0? lt 0.1, are
  satisfied with the tighter tolerance applied.
  All tolerances are rms levels and the voltage and
  phase tolerances per klystron for L2 and L3 are
  ?Nk larger, assuming uncorrelated errors, where
  Nk is the number of klystrons per linac.

5

• ParameterSymbol?I/I0 lt 12??E/E0? lt 0.1Unitme
  an L0 rf phase (2 klystrons)?00.100.10S-band
  degmean L1 rf phase (1 klystron)?10.100.10S-band
  degmean LX rf phase (1 klystron)?x0.50.5X-band
  degmean L2 rf phase (28 klystrons)?20.070.07S-band
  degmean L3 rf phase (48 klystrons)?30.50.15S-band
  degmean L0 rf voltage (1-2 klystrons)DV0/V00.100.
  10mean L1 rf voltage (1 klystron)DV1/V10.100.10m
  ean LX rf voltage (1 klystron)DVx/Vx0.250.25mean
  L2 rf voltage (28 klystrons)DV2/V20.100.10mean
  L3 rf voltage (48 klystrons)DV3/V30.50.08BC1
  chicaneDB1/B10.010.01BC2 chicaneDB2/B20.050.05Gu
  n timing jitter?t00.80.8psecInitial bunch
  chargeDQ/Q02.04.0

6
Requirements (2)

• Achieve 120 Hz feedback to maintain phase/amp
  stability
• Adhere to LCLS Controls Group standards RTEMS,
  EPICS, Channel Access protocol
• Begin RF processing of high-powered structures
  May, 2006

7
Local feedback loop requirements

• At each of these locations, the klystrons phase
  and amplitude will be monitored and controlled
• When beam is present, control will be done by
  beam-based longitudinal feedback (except for
  T-cavs) when beam is absent, control will be
  done by local phase and amplitude controller (PAC)

8
Design considerations

• Through end of January 2005, various solutions
  were evaluated, from 100 COTS modules to hybrids
  of in-house designed boards.

9
Options considered Jan/05 (2)
10
Options considered Jan/05 (3)
11
Options considered Jan/05 (4)
12
Options considered Jan/05 (5)
13
Options considered Jan/05 (6)
14
Options considered Jan/05 (7)
15
Options considered Jan/05 (8)
16
Narrowing down the options May/05

• Later, the options were narrowed down to two an
  Off-the-shelf solution and an in-house solution.
• This subset of options was presented at the
  Lehman Review, May 10-12, 2005. Ref Low Level RF

17
Off-the-shelf solution May/05
18
In-house solution May/05
19
Evaluation

• The Off-the-shelf solution is
• Expensive (25K per instance 10 instances)
• Noisy. ADCs are up to 150 from what they
  measure so analog noise levels and ground loop
  problems would need to be dealt with
• The in-house solution is
• Possibly longer to develop due to board design
  and fabrication time.

20
Evaluation (2)

• Characteristics of the Off-the-shelf solution
  were seen as requiring more effort than those of
  the in-house solution
• Potential offered by the lower cost of the
  in-house solution to replace 250 klystron
  controllers in the remainder of the LINAC is
  attractive
• Hardware people were available as of 22aug2005 to
  work on board design if µcontroller was decided
• Turned to the EPICS community for ideas and chose
  a µcontroller

21
Evaluation (3)

• Lower cost alternatives to the 15K VME chassis
  and IOC were discussed in the session on hardware
  at the EPICS Collaboration Meeting. April 27-29,
  2005
• Of the options available, only the Coldfire
  uCdimm 5282 processor had the communication speed
  and power to meet our data requirements. Cost is
  150 per processor plus the development of the
  board it sits on

22
Evaluation (4)

• By choosing the Coldfire processor, we are able
  to make use of the port of the operating system,
  RTEMS, which has already been done.
• RTEMS is the standard for the real-time operating
  system chosen for LCLS by the Controls Group
• EPICS, the standard for the control system
  software for LCLS runs on RTEMS
• With these choices, the LLRF control system will
  be fully integrated into the rest of the LCLS
  EPICS control system and can speak to other
  devices and applications such as control panels,
  alarm handlers and data archivers, using Channel
  Access protocol, the standard communication
  protocol for this project.

23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
How this fits into global feedback (Gun)
27
How this fits into global feedback (L0)
28
How this fits into global feedback (L1)
29
How this fits into global feedback (L2)
30
How this fits into global feedback (L3)
31
How this fits into global feedback
32
Conclusions

• This solution
• meets the spec for speed and noise
• avoids signal noise problems
• avoids ground loop problems
• meets LCLS control system requirments and
  standards running EPICS on RTEMS
• provides a low cost path for future upgrade in
  the rest of the LINAC when the rest of the
  klystron control is replaced

33
Conclusions

• At 120 Hz, the LCLS LLRF raw signals must be
  processed, the phase and amplitude corrections
  must be sent out, applied and achieved
• When there is beam, this system will integrate
  with the beam-based longitudinal feedback by
  accepting the latters RF phase and amplitude
  corrections and passing them on.