1. Experience with high loaded Q cavity operation at HZB
Axel Neumann, TTC CW-SRF meeting
U Cornell, 06/11-06/14/2013
Ithaca, NY

2. Test set up: Horizontal test facility HoBiCaT
Testing fully equipped cavities including helium vessel, motor- and piezo tuner, CW modified TTF couplers, magnetic shielding, etc.
Temperature range down to 1.5 K, typically 1.8 K with K: 16 mbar ±30 mbar rms
Coupling variable, installations down to bc=1 possible
RF set up: 19 kW IOT, 400 W solid state amplifier driven by PLL or Cornell’s LLRF system
Two cavities tested in parallel or sample studies
Gun cavity tested with diagnostic beam- line

3. High QL Measurements done/planned at BESSY/HZB
Cavity type
RF requirements or
achievements
Design/operated QL
Project
Eacc=15-20 MV/m,
Q0= , sf=0.02 deg., sA/A=1.10-4
Operated:
BESSY FEL, SC CW,
low beam loading
Epeak=12-25 MV/m
Q0=
sf=0.02 deg., sA/A=
Operated:
Lead cathode, all SC gun for low current FELs
Eacc=20 MV/m,
Q0= , sf≤0.05 deg.
5.107
BERLinPro Linac, zero beam-loading
E0=30 MV/m,
Q0≤4.109
BERLinPro Gun prototype, 4mA
Eacc=10 MV/m see Cornell’s presentation
1.105
BERLinPro Injector,
100 mA, low QL
TESLA
Tested in HoBiCaT
1.6 cell Pb/Nb
hybrid, J. Sekutowicz
Tested in HoBiCaT
HZB
1.4 cell
Expected 2014
Cornell booster
Expected 2014

4. Mechanical oscillations of the Cavity:
CW operation: Field stability determined by Microphonics
Helium pressure fluctuations
Df/Dp = Hz/mbar, Gun:100 Hz/mbar
Heat transport dynamics
16 mbar
Field amplitude variation:
Dynamic Lorentz force, Df/DEacc² = 1Hz/(MV/m)²
2.5 mm Niobium walls
Stochastic
background
noise
Deterministic,
narrow-band
sources:
Vacuum pumps
Mechanical oscillations of the Cavity:
Microphonics
G. Bissofi
222 Hz
151 Hz
Response of the
Cavity-Helium vessel-Tuner
system:
(FEM simulations, e.g.:
Devanz et al. EPAC 2002)

5. X X Measurements at high loaded Q
Low beam-loading CW SRF linacs allow operation
at high QL narrow bandwidth (order of 10s Hz) X
CW operation: Microphonics, peak events?
Field stability at the presence of microphonics
High cw gradient: E.g. 20 MV/m  Ponderomotive instabilities by LF detuning
High cryogenic dynamic losses, helium bath stability
Beam transients during ramping to 100 mA, how to handle? (ERL)
Residual beam-loading due to beam losses  non- perfect recovery, time jitter?
Combine microphonics compensation with LLRF at high QL for a multi-cell cavity

6. Power requirements and parameter space
Studied within this work
Pros:
High QL: low forward power for
a given field level, reduction of thermal stress for RF transmission line/coupler
But:
Effective detuning is a
convolution of the detuning
spectrum with the
cavity response (bandwidth)
Cavity transfer function itself altered by controller settings (feedback gains) QL
7-cell cavity, 20 MV/m, Ib=0

7. Detuning spectrum versus bandwidth (TESLA cavity)
For two different tuning schemes (Saclay I and INFN Blade) open loop measurements of
microphonics vs. QL were performed
Both tuners showed to have different transfer functions and thus detuning spectra
QL,Saclay:
QL,Blade:
Blade: Mechanical eigenmode
at 300 Hz, vacuum pump freq.
Saclay: Excitation of 1st mechanical eigenmode sets in

8. Detuning characterization of a TESLA cavity, short-term
Characterisation: Measurement results
Detuning (Hz)
sf = 1.56 Hz
SFFT
Excited Eigenmode
He pressure-
variations
FFT
Detuning (Hz)
Time (s)
HoBiCaT: sf = Hz (rms)  2-13° phase error (Aim: 10-2 °) „open loop“ „closed loop“
He pressure variations: fmod < 1 Hz
Cavity specific: Lines at 30 or 41 Hz 8

9. Longterm stability: Peak events
Microphonics recorded at HoBiCaT with TESLA cavity for 48 hours
RMS Values around 1-5 Hz  Determines field stability and thermal loading of RF system (5 kW)
Peak values extend out to 17 σ!  Determines RF power installation (15 kW)
Peak events occur times a day! (This was partly improved by changes to the control settings of the under-press. pumps.)
Expected field stability: °
For „comfort“ want to reduce the microphonics
Gaussian sub-range
0.8 Hz rms 9

10. Time-frequency analysis by Wavelets^
t/s
s. w
Morlet
Df (Hz)
Variation up to Df =10 Hz on a ~100 ms time scale
Spectrum of He-pressure variations of stochastic nature
Adaptive, „learning“ (dynamic) compensation mandatory
Need for classic feedback control 10

11. Detuning compensation: Characterize the tunercavity action
TESLA Cavity
1.6 cell gun Cavity
Mechanical
resonances
Turbo pump
Mechanical
resonance
Helium
Activity?

12. Model based controller: Fit of the transfer function
TESLA Cavity
Fit: Parallel acting 2nd order systems
Evaluate response of higher modes at lower frequencies
>20 modes needed for fit
Systems complexity complicates use of model based feedbacks
(e.g. Kalman filter)
Transfer function as
look-up table
Kalman approach tested
within a Master thesis
(P. Lauinger), test in prep.
Relevant for tuning 12

13. A tested scheme: Least-mean-square based adaptive feedforward
Dl (nm)
t (s)
External mechanical
oscillations
FFT
t (s)
DU (V)
Compensating
signal
t (s)
Df (Hz)
Detuning
of the cavity
IFFT
∙H-1
Calculation
of optimal
FIR filter
parameters
For white noise excitation
The FIR filter would be
H-1piezoDf 13

14. Compensation results Single-resonance control: SFFT QL=6.4.107
Feedback only
sf = 0.89 Hz
Feedback
and Feed-
forward
sf = 0.36 Hz
Open
loop
sf = 2.52 Hz
SFFT
QL=
Multi-resonance control:
Piezo resolution seems
to limit control of
neighboring modes

15. LLRF studies with U Cornell: Limits of QL
log(sf)
Best results: ° ° °
QL=5.107, f1/2=13 Hz
QL=2.108 , f1/2=3.25 Hz
9 cell TESLA cavity
Eacc= MV/m
Tbath= 1.8 K
PI piezo loop
8/9-p filter optimized
QL=1.108, f1/2=6.5 Hz
LF detuning  IOT beam instable Cavity field trip
Areas with sf>0.1 were blanked out

16. LLRF studies QL sf (Hz) sf (deg) sA/A Pf (kW) 5.107 9.5 0.008 1.10-4
1.106
1.108
7.9
0.009
2.10-4
0.595
2.108
4.2
0.024
3.10-4
0.324

17. Gun cavity LLRF and microphonics studies
Insufficient
cooling of cathode
More about Gun Cavity in talk by
A. Burrill
@25 MV/m, quench
would occur after few minutes
A lot of power dissipated in LHe bath  effect on microphonics?

18. Results with the SC Gun Cavity
QL: , due to ponderomotive instability changed to
At 25 MV/m the cavity losses increase due to bad thermal contact of cathode plug and back wall via indium seal
Microphonics increase by factor of three, mechanical resonance excited
Strong line at 35 Hz appears: Eigenmode of the Helium bath?

19. On-going studies: Thesis P. Lauinger
Further studies with TESLA cavity planned as well as microphonics
compensation using Kalman filter approach
Microphonics (Hz)
Qcrit (W/cm²)
Pheater (Dfmax) (W)
Heater power (W)
Cavity driven by LLRF at Epeak=15 MV/m
Piezo compensation in PI loop mode with low-pass filtering
Additional power dissipated in LHe bath by heater within liquid
Microphonics recorded while heater is powered
TLHe (K)
T. Peterson, TESLA-Report

20. Summary
Microphonics as main error source for field stability extensively characterized for TESLA and 1.6 cell Gun Cavities at various QL
Microphonics compensation demonstrated with TESLA cavity at high QL, an order of magnitude feasible Needs to be implemented within operating LLRF system
LLRF studies showed a stable operation at up to QL=2.108, still needs to be demonstrated for fields larger than Eacc>12 MV/m
Experiments to correlated microphonics and helium heat transport dynamics were started, more results hopefully this summer
New microphonics compensation schemes will be tested soon
For both cavity types studied a field stability of at least sf≤0.02 deg and sA/A≤ was demonstrated
Thanks to, people involved:
S. Belomestnykh*, J. Dobbins, R. Kaplan, M. Liepe, C. Strohman (Cornell, *now BNL) for the LLRF system
J. Sekutowicz (DESY), P. Kneisel (JLab) for the 1.6 cell Gun Cavity
W. Anders, A. Burrill, R. Goergen, J. Knobloch, O. Kugeler, P.
Lauinger + HoBiCaT personell (HZB)