1. Experience with High Loaded Q cavity Operation at JLAB
Tomasz Plawski – Jefferson Lab , Newport News, VA

2. Outline 12 GeV Project High Q Cavity Operation Challenges
C100 RF system
Field Startup
RF Field Control
Operability
C100 One Hour Run

3. 12 GeV Upgrade Project New cryomodules New RF power sources (13 kW)
Refrigeration
Magnets
Additional arc-beamline
Extraction system
New experimental area Hall D

4. High QL Challenges Field stability (microphonics) Field startup
Fundamental frequency f0
1497 MHz
Accelerating gradient Eacc
> 20 MV/m
Input coupler Qext
3.2 x 107
Active length
0.7 m
r/Q
1300 Ω/m
Tunning sensitivity
0.3 Hz/nm
Pressure sensitivity
420 Hz/torr
Lorenz force frequency sensitivity KL
~2 Hz/(MV/m)2
Field startup
Field stability (microphonics)

5. RF System for C100 Cryomodule

6. Digital Self Exciting Loop (DSEL)
Field Startup - DSEL
Digital Self Exciting Loop (DSEL)

7. SEL -Resonance Control
Probe Phase
Discriminator (PZT Drive)
Frequency Discriminator
Four frequency ranges automatically selected
1-125 Hz (PZT)
125 Hz Hz (stepper, PZT)
1250 Hz- 50 kHz (stepper)
50 kHz- 500 kHz (stepper)

8. C100 GDR mode – Original/Modified Tuner
Phase noise
25.6 deg rms /14 Hz rms
Phase noise
7.5 deg rms /4 Hz rms

9. C100 Field Stability (Phase)
Tone mode/ open loop
GDR mode/ closed loop
(tan(det)*f0)/(2*QL)
For 2.5*10^ deg corresponds to 4 Hz rms detuning
Phase noise spectral density for Tone Mode (< 3MV/m) and GDR (IQ regulation at 21 MV/m) mode

10. C100 Gradient Stability
measurement with Audio Analyzer

11. “Coupled Cavities”
C100-4 Cavities 4, 6, 7, 8 responding to an applied PZT step voltage change from 4 to 3 volts in cavity 5
Adjacent Cavity
coupling is ~ 10%
between 1-4 and
5-8 cavities
Cavities 4 and 5
have a “quasi”
mechanical
support between
them.
Ringing is the
21 Hz mechanical
Mode
Cavity 5 PZT
moved 460 Hz
Cavity Gradients
10 MV/m
Locked in GDR Mode
Klystron had the
overhead to keep
cavities locked
Stepper Motor
kicks in to tune
the cavities
Basically used the PZT as a hammer stepping the voltage quickly.
Take away points are on the slide.
One that is not mentioned is how easy the 21 Hz mechanical mode is excited in the adjacent cavities.

12. Graph of gradient and detuning (Hz) as a cavity is faulting (blue)
Cavity Fratricide
Cavity Fratricide occurs when one cavity faults (Arc, waveguide vacuum, quench etc.) and the Lorentz force detuning of the faulted cavity detunes the adjacent cavities resulting in the cavity faulting too.
Adjacent cavity was operating at 5 MV/m so the klystron had the overhead to absorb the detuning
Cavity 5 is stepped 4 volts (2000 Hz) using the PZT which is much larger than a faulting cavity
Adjacent cavities feel the effect.
Closest cavity (#6) sees largest perturbation.
Cavity 4 sees smallest do to the center rod in the string between cavity 4 and 5 in the cryomodule.
Cavities were operated at 5 MV/m so that they would not trip off due to lack of power
A useful feature possible on digital LLRF control systems is the use of data buffers. The hardware allows the operator to catch real time data from the cavity-control system. This is extremely useful when diagnosing cavity faults or measuring microphonics. Figure 3 is a plot of a cavity fault. The top graph displays the cavity gradient of the faulted cavity and the adjacent cavity. The bottom graph shows each cavity’s detuning in Hz. The red curve on the bottom graph shows the sharp reaction (Lorentz contraction from the faulted cavity) of the non-faulted cavity. The adjacent cavity was operating a fairly low gradient, 5 MV/m, so the klystron had more than enough overhead to absorb the 77 Hz detuning.
Graph of gradient and detuning (Hz) as a cavity is faulting (blue)

13. C100 PZT Control Piezo compensation bandwidth: 1 Hz PI regulator
Cavity frequency control is provided by a mechanical stepper motor and a PZT. The stepper motor provides coarse tuning and can tune the cavity to ± 1 Hz of the reference. The drawback is that it is slow ≤ 0.1 Hz, and can disturb adjacent cavities.
The PZT provides “close in” fine tuning control (± 1000 Hz). The control bandwidth (speed) is 1 Hz and is limited by the mechanical resonances of the cavity. Even with this relatively slow control, the effect is dramatic on cavity tuning. Figure 4 shows the cavity detuning with and without the PZT turned on while locked in GDR mode. Presently the PZT control algorithm is a simple PI feedback. The PZT voltage is centered at 75 volts, for a full range of 150 volts. If the voltage goes above or below a certain voltage the stepper turns on to “re-center” the PZT back to 75 volts
Piezo compensation bandwidth: 1 Hz
PI regulator
Wider bandwidth causes mechanical mode excitation/ instabilities
Substantial improvement for slow detuning ( helium pressure drift or slow microphonics)

14. Patience

15. Operability: Cavity Faults
Cavity/Cryomodule: Fault, Mitigation and Recovery
At 20 MV/m and KL of 2 you are looking at a 800 Hz detuning when the cavity trips or ~ 13 bandwidths
Adjacent cavities will feel this and RF systems must react. Typically too fast for a PZT.
If we assume a cavity to cavity coupling of 10%, the RF system must have the power overhead to absorb a neighbor cavity faulting.
If not, it will set up a “domino” effect and you will be recovering the whole cryomodule.
CEBAF C100 cavities have observed this at 20 MV/m and a KL of 2.
One solution is to switch the adjacent cavities to SEL mode keeping gradient in the cavities. Then switch back to GDR mode when the faulted cavity is recovered.

16. C100 One Hour Run

17. Data from BPM during C100 Run
During this test, two C100 cryomodules were on, adding 154 MeV to the total of 2 GeV of the electron beam energy. Beam current uA (sum)

18. Summary - C100 Operation
Two C100s were operated during last years Nuclear physics run.
Training: Operations staff were trained to operate and recover faulted C100s during this period.
Software and Algorithms were refined and improved
Cavity/Cryomodule Fault recovery improved from ~ 50 mins to single minutes
One button fault recovery is in beta testing
SEL to GDR crossover improved to allow switch at 20 MV/m
Piezo Tuning algorithm fully functional
Data buffers operational allowing real time fault diagnosis
C100 Operations concluded with a successful run of SL25 at 108 MV and 465 uA for over an hour of nuclear physics beam.
Firmware is the key: No hardware modifications were necessary during commissioning , still weeks of effort was required to make system operable.
Overall Jefferson Lab’s technical and operations staff gained valuable
experience during this time period.

19. C100 RF Team Curt Hovater Trent Allison Rama Bachimanchi George Lahti
Clyde Mounts
Tomasz Plawski
Mike Wilson
and many others