1. Ben Woolley CLIC Workshop 2016 (LCWS16) CERN, Switzerland 18/01/2016
X-Band test-stand operation: LLRF, controls, conditioning and operational algorithms, stability, XBox3 schedule
Ben Woolley
CLIC Workshop 2016 (LCWS16)
CERN, Switzerland
18/01/2016

2. Outline General test stand specifications Control systems
≤ 250 ns pulse, 50 MW, ≥ 50 Hz pulse repetition rate.
Adequate diagnostics.
E.g Xbox-2.
Control systems
Down mixing and digitization.
LLRF signal generation and up mixing.
Safety and interlocks.
PXI crate.
Automation algorithms
Power control: Pressure and BDR.
Pulse compressor tuning.
System stability
Xbox-1.
Xbox-2.
Xbox-3
Layout.
LLRF crates.
Schedule.
B. Woolley
18/01/2016

3. Test Stand Specification
For most CLIC accelerating structure designs the required input power to achieve an average unloaded accelerating gradient is MW.
The CLIC pulse shape has a 160 ns flat top with a ~70 ns ramp. However for conditioning of the structures a flat pulse ≤ 250 ns can be used.
Recent studies suggest that it is the number of pulses that dictates the speed at which a structure conditions, therefore high pulse repetition rates are preferred ≥ 50 Hz.
Use of a single 50 MW klystron and pulse compressor can be used to test one or more structures (Xbox-1/2).
Or 2+ klystrons can be combined (Xbox-3 or nextef).
B. Woolley – CLICWS 2016
18/01/2016

4. Diagnostics of the DUT RF Load Dipole Device under test Magnet
Incident Power
Reflected Power
Uppsala Dark Current Spectrometer
60 dB directional coupler
RF Load
Transmitted Power
RF In From Pulse Compressor
WR90 Waveguide
Ion gauge readout
Ion gauge readout
Input coupler
Output coupler
Screen
Dipole
Device under test
Beam-pipe
Beam-pipe
Magnet
Upstream Faraday cup signal
Radiation Monitor
Downstream Faraday cup signal
Collimator Readout
B. Woolley – CLICWS 2016
18/01/2016

5. Example: Xbox-2 Scandinova Modulator
Klystron (50MW, 1.5us pulse) For Crab cavity test: SLAC XL5 klystron
Pulse compressor (250ns, ratio ~3)
Stainless steel load
B. Woolley – CLICWS 2016
03/11/2015

6. Down-mixing and Digitization
RF Input 1
Amp
400 MHz LPF
splitter
IF output 1
2.9 GHz PLL
RF Input 2
Amp
400 MHz LPF
X4 freq.
IF output 2
RF Input 3
Amp
400 MHz LPF
IF output 3
RF Input 4
Amp
400 MHz LPF
IF output 4
12 GHz signals are sent to a mixing crate containing a CERN built 2.9 GHz PLL (Stephane Rey).
x4 multipliers are used to produce 11.6 GHz.
This is used to down-mix the signals to 400MHz IF.
NI 5772 digitizer module samples at 1.6GSPS and FPGA IQ demodulation is used to obtain amplitude and phase.
B. Woolley – CLICWS 2016
18/01/2016

7. LLRF Generation and Up-mixing
To TWT
Amp
12 GHz BPF
2.4 GHz input from PXI IQ modulator
2.4 GHz PLL
X4 freq.
splitter
splitter
To Reference input
12 GHz BPF
NI 5793 IQ modulator with 200MHz bandwidth and 250MSPS IQ voltage generator is used to produce an arbitrary signal which can be varied in amplitude and phase at 2.4GHz.
The signal is mixed up to 12 GHz using a mixer with a 9.6GHz input.
A reference 12GHz signal is also crated which is locked to the signal generated by the IQ modulator.
B. Woolley – CLICWS 2016
18/01/2016

8. Safety, Interlocks and control Algorithms
Main personal safety issue is X-ray radiation during operation.
Interlocks on the bunker door and klystron/modulator access doors stop modulator pulsing if opened.
Modulator interlocked is radiation levels are too high inside or outside of the bunker.
Machine protection issue is from high vacuum and reflected power to the klystron.
Double interlocking of the vacuum; 10-5 mbar interlocks the modulator and 10-7 mbar interlocks the LLRF driver.
Reflected power is monitored by log detectors which stop the LLRF output if a certain level is breached.
Interlock hardware
Modulator control system is used for radiation and access interlocks.
NI 6583 module attached directly to an FPGA in the PXI crate is the main trigger and interlock module at Xbox-2&3. This is used for all other interlocks; vacuum, RF, high temperatures, etc.
B. Woolley – CLICWS 2016
18/01/2016

9. PXI Crate
Allows up to 17 expansion cards to be used. Digitizers, RF generators, DMMs, serial interface etc. Plus powerful quad core controller.
OS is NI’s LabVIEW Real-Time, easily programmable in ‘almost’ standard LabVIEW language, with the stability and speed that real-time programming allows.
Acquisition and interlock cards use FPGA interface for added reliability and speed.
B. Woolley – CLICWS 2016
18/01/2016

10. Pulse and Power Control Algorithms
Pulse length and LLRF frequency are set:
Fast  pulse to pulse
PID loop on the incident power to the structure
Medium  seconds-minutes
increase power by 10kW every few minutes if no BD
reduce power by 10kW if successive BDs too close in time
Slow  hours
BDR measurement and stop power increase if it is too high
BDR measurement performed across a moving window of approx. 1M pulses.
Use PID loop using the system pressure as a process variable.
Increase in pressure results in a reduction of power and vice versa.
OR 
B. Woolley – CLICWS 2016
18/01/2016

11. Xbox-1 Pulse Flattening
PC Cavity
Encoder read-out
Stepper control
Temperature changes lead to:
Total volume offset  decrease in pulse flatness.
Relative volume offset  Increased reflection to the klystron.
Used to use stepper motors attached to pistons to tune the pulse compressor.
B. Woolley – CLICWS 2016
18/01/2016

12. Xbox-1 Pulse Flattening
PC Cavity
Chiller temp control
Chiller 1
Chiller 2
Temperature of each cavity is changed depending on the flat-top slope and the ‘klystron reflection ratio’ defined above.
B. Woolley – CLICWS 2016
18/01/2016

13. Xbox-2 Pulse Flattening
Better design of pulse compressor (Igor Syratchev), results in good thermal stability. No temperature control needed.
Instead phase is controlled to flatten the top of the pulse:
ɸn+1 = ɸn + g(ASP – A)
ASP amplitude set point
A current amplitude
g feedback gain
B. Woolley – CLICWS 2016
18/01/2016

14. Xbox-1 Up-time (TD26CC) 59% uptime for the 2015 dogleg run (TD26CC)
Long periods of downtime due to modulator issues; solenoid PSUs and IGBT faults.
General inconsistency of running due to pulse compressor detuning caused by excessive BDs in the waveguide network.

15. Xbox-2 Up-time 65% uptime for the 2015, T24 OPEN structure.
Long periods of downtime in October/November linked to interventions to the area in preparation for Xbox-3 (water, electricity etc).
Commissioning of Xbox-3 modulators required shutdown of Xbox-2 because of low cooling water flow rate.
Some inconsistency of running caused by radiation interlocks which cannot be automatically reset.

16. Xbox-3 Layout Klystron/modulator Directional coupler
RF /vacuum gate valve
Directional coupler
RF/vacuum pumping port
3 dB hybrid
RF pulse compressor
RF load
WR90 waveguide network
7 MW x 4µs x 400 Hz
14MW x 4µs x 200 Hz
56 MW x 250 ns x 100 Hz
B. Woolley – CLICWS 2016
18/01/2016

17. High Power RF Network
Can be run as two separate units with a pair of klystrons each:
Klystron/modulator
Directional coupler
RF pulse compressor
RF load
14 MW x 4µs x 200 Hz
6 7 MW x 4µs x 400 Hz
RF /vacuum gate valve?
3 dB hybrid
56MW x 250 ns x 200 Hz
TS = TEST SLOT
- Phase & amplitude  Structure RF Signals INC, REF, TRA. (3 /TS)
- Phase 250MSPS  Phase but not speed is key KLYin, KLYout, PCin (3/TS)
- Slow Amplitude only. Log  Reflected signals/machine protection (3/TS)
-Faraday Cup Signals  (2/TS)
B. Woolley – CLICWS 2016
18/01/2016

18. Xbox-3 LLRF Racks
Each test stand will get it’s own temperature stabilized rack with some shared resources (PLL crate and power meters).
The channels are split, one of the channels is MUX’d to a power meter for calibration and the other is sent to the mixing crates or log detectors.
The mixing crates are based on the Xbox-2 designs but contain fast multiplexers which switch in time with the power sent to the DUTs. Saving on ADC channels.
B. Woolley – CLICWS 2016
18/01/2016

19. Progress of LLRF crates
Log detector crate
Up/down mixing plate (1 of 2 per crate)
Manufacturing of the crates is progressing well.
All components have arrived at CERN and are ready to be integrated.
Commissioning and testing of the crates will start within the next few weeks.
All crates to be completed + tested by end of February.
Interlock/trigger crate
B. Woolley – CLICWS 2016
18/01/2016

20. Xbox-3 Schedule Klystron and modulators are tested and ready to go.
Unit A has had a long RF test and reached above nominal power at short pulses (800 ns) and longer pulse width (3.5us) at lower power. (Limited by waveguide network).
Cooling system flow-rate upgrade complete; can now cool Xbox2+3.
New solid state driver prototype successfully tested, with 4 production units on order. Should arrive before end of February.
All parts for the LLRF and control systems have arrived at CERN.
High quality RF cables are yet to arrive, were due before end 2015.
LLRF sub-systems are being assembled. Expected completion before the end of February.
Waveguide components have been manufactured with visual inspection of the components on going. Should be installed by end of February.
First pair of cavities have been manufactured but one was not leak tight. Repairs are ongoing. However, cold RF test was successful.  Manufacture of 5 more pulse compressors launched.
Expected delivery of all 6 pulse compressors will not be for another few months.  Xbox-3 will be commissioned without pulse compressors.
Racks have been installed in the experimental area and the waveguide supports are almost complete.
B. Woolley – CLICWS 2016
18/01/2016

21. Summary
Over the past ~3 years we have greatly increased our understanding of the operation of high power X-band test stands.
Control system upgrades and increased automation have allowed us to develop tools to more safely and quantitatively condition structures.
Apart from some major interventions the up-time of the Xboxes has been very good.
Work on Xbox-3 is progressing steadily and most/all of the hardware should be in place by the end of February.
B. Woolley – CLICWS 2016
18/01/2016

22. Thank you!
B. Woolley – CLICWS 2016
18/01/2016

23. Extra slides
B. Woolley – CLICWS 2016
18/01/2016

24. 2-Way Combination Demonstration
10 mW amplifier
RF pulse compressor
Co-ax Hybrid
RF loads
KREF1
KREF2
KLYOUT1
KLYOUT2
PCIN1
TRA1
TRA2
PCREF1
WG network
12 GHz FAST Signals (x5)
LLRF CRATE
400 MHz IF (x5)
Down mixing Stage (Aq.)
12 GHz
LO PLL 2.9 GHz
LO PLL 2.4 GHz
Up mixing Stage1 (Gen.)
Up mixing Stage2 (Gen.)
12 GHz
2.4 GHz
2.4 GHz
PXI CRATE
RFSG 1 (IQ gen. & VM)
12 GHz SLOW Signals (x3)
RFSG 2 (IQ gen. & VM)
1.6 GSPS ADC x5
Diode/Log detector CRATE
Diode/log detectors x4/x8
DC coupled 50 MHz (x3)
250 MSPS ADC x1 (4ch.)
PXI and LLRF crates are locked by 10 MHz reference
B. Woolley – CLICWS 2016
18/01/2016

25. Results
Concept of phase switching proved in 2013 with modified version of the Xbox-2 LLRF system.
Low power hybrid and the Xbox-2 pulse compressor were used.
Switching at 400 Hz was demonstrated (lower left).
Ch1
Ch2
0.5ms /div
B. Woolley – CLICWS 2016
18/01/2016