2 A technical studentship project by Sebastian Göbel Some RF basics Continuous wave CW versus pulsed RF Average peak power and duty cycle Solid state versus tube based amplifiers High demand in LINACS ID: 934810
Download Presentation The PPT/PDF document "The principle of Pulse Compressor" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Slide2The principle of Pulse Compressor
2
A technical studentship project by Sebastian
Göbel
Slide3Some RF basicsContinuous wave (CW) versus pulsed RFAverage, peak power and duty cycle.
Solid state versus tube based amplifiersHigh demand in LINACS
3
Slide4IQ and narrowband theory
A narrowband signal can be described as two sine waves in quadrature with separate amplitudes Similar to complex impedance
4
Slide5Which signals are applicable?
Applicable
Not applicable
Signals that can be described as a slowly changing signal
For example a telecommunications carrier
Narrowband signals
For example a humans voice
Cannot be described a slowly changing signal
Very wideband signals
5
Slide6Complex description of a signal
The signal can be described as a signal with a slowly changing amplitude and phase
6
Slide7IQ description of signal
Possible to describe phase, frequency and amplitude modulationRepresents the instantaneous amplitude and phase of the signal
7
Slide8Modulation and demodulationModulation
DemodulationAdding information to a carrierComplex modulation enables two separate information streams I,Q.
Changing I,Q is equivalent to changing amplitude and phase.
Recovering information from a modulated carrier.
Enables separately receiving I,Q channels
Possible to measure amplitude, phase over time simultaneously.
8
Slide99
Slide10Basics about resonatorsOne mode of a cavity can be described as an RLC circuit
Q is bandwidth of resonance, high Q means narrowbandFill time, decay time depends on Q, beta
10
Slide11What is a pulse compressor?
RF equivalent of a capacitor bankStores RF energy to temporarily amplify an RF pulseEnergy stored in cavities or delay lines
Is in theory passive, input is the control
11
Slide12Basic input for compression
12
Slide13How does it work?Power leaking from cavity destructively adds with inputQuick input phase doesn’t change cavity phase
Causes cavity power and input to constructive add instead
13
Slide14Usage of pulse compressors
High power pulsed RF with low duty cycleCould allow for use of solid state over tube amplifiers
Reduces
, increases duty cycle
Faster pulses using Klystron
Klystrons are based around high Q cavities, causing slow reactions
14
Slide15Model of Pulse Compressor
Described as coupled RLC circuit
15
Slide16Simulation of model
Simplify by
Describe input current as sum of R,L,C currents and solve for second order derivative of cavity voltage
Solved numerically using
fourth order
Runge-Kutta
method
Stepping through time and approximating cavity voltage and derivative
16
Slide17Simulation results
Simulation using measured values from cavity
Focus on 1us pulse
Program capable of any values
17
Slide18Basic input
18
Slide19More advanced input is possible
19
Slide20Analysis of simulated dataAmplitude is easily readable from signal envelope, easily trackedInstantaneous phase is impossible to see from graph.
I,Q Demodulation enables easily extracting instantaneous amplitude and phase over time.20
Slide21IQ Demodulation in softwareUsed resonant frequency of cavity for LOGenerate two arrays with quadrature LO signals, emulates LO and quadrature hybrid.
Mix the two LOs with the input and receive two signals, filter with rolling average of 5 RF cycles.21
Slide22Test of a real Pulse CompressorInteresting to test a real Pulse Compressor and seeing if simulation is accurate.
Single cavity Pulse Compressor chosenMatching of cavities is critical and very temperature dependent.Dual cavity would require active controls to match cavities.
22
Slide23Equipment required
RF synthesiserFast amplitude and phase modulator(s)
High Q cavity
Q in the many thousands at least for shorter pulses
Circulator
Measuring device(s)
Preferably not just amplitude.
23
Slide24Cavity
Based around copper cavity from amplifier, low loss.Designed for 200 MHzCustom made by Cristiano GagliardiSimulated by Rolf WegnerFeatures a pickup
24
Slide25Cavity results
Parameter
Measured
Note
F0
197 MHz
Drifts
with temperature and not critical
Q0
7.8E3
Good
Beta
3.6
Good
25
β
is measured from the reflection Coefficient at resonance together with loaded Q,
.Q is 3dB-bandwidth of T, is quality factor of circuit connected to a 50 ohm system.Then Undampened Q, is extracted from ,
Slide26Phase modulatorConsisted of 3 CERN made phase shifters
Not quick enoughPhase rise, fall time around 600nsNot capable of 180deg phase inversionCaps at around 160degRF synth not quick enough eitherMeasured using oscilloscope and IQ demodulation
New solution was found
using
IQ
modulation
Speed less than 100ns
26
Slide27Test equipment
Type
of equipment
Equipment used
Note
RF synthesizer
Rohde & Schwarz
SMB100A
Good
Cavity
Custom made cavity
Good
Amplitude
modulator
IQ modulator
Controlled by arb. Signal generator.
Phase
modulatorIQ ModulatorControlled by arb. Signal generator.Arbitrary signal generatorTektronix AFG3022CTo control amp, phase modulators from software.CirculatorTDK CU281A, 200MHzMeasuring deviceTektronix DPO4104BOscilloscope, IQ demodulation used in software.27
Slide28Comparing simulation and lifeComparing simulated output pulse and measured output pulse
Using same computer generated phase functionAllows more complex phase, amplitude functionsGenerated in Matlab, transferred via USB to arbitrary signal generatorRead using oscilloscope and demodulated to display data similarly to simulated data
Frequency measured using curve-fit during input
pulse
Set reference phase with statistics
28
Slide29180 Degree step zoomed out
29
Slide30180 Degree step
30
Slide31Exponentially rising phase31
Slide32Stable amplitude and phase pulse
32
Slide33New amplitude and phase shifterIQ Modulation basedAllows phase and amplitude
Modern and widely used in high-speed communication systemsDevelopment kits available360 Degree capable
33
Slide34IQ Modulator
I,Q equivalent to amplitude and phase
Can change amplitude and phase
34
Slide35ADL5390ADL5390 was chosen
Two matched variable amplifiers with a combinerVoltage input 0-1V, controllable by arbitrary signal gen.FeaturesUp to 230MHz modulation BW, limited to 25MHz by signal generatorFull scale change in around 50ns
Amplifier based, increased isolation
Few external components
Quadrature coupler
Biasing
Power supply
35
Slide36Expected speed of the modulator
230 MHz BW, limited to 25 MHz by the control signal
Quicker at lower gain as full scale is 50ns
To flip 0 to 180, is to switch from full scale twice
Expected time for 180 phase inversion is less than 100ns, even at maximum gain
Speed could further be improved with external post amplifier, at the cost of SNR
36
Slide37Quadrature couplerInitially a splitter and phase shifter was tested
Also called 3dB hybridTransformer based which reduces size compared to transmission lineHand tuned to 197MHzOutput mismatch, <0.2 dB, <5degImproved amplitude stability during phase change, sign of better quadrature input
37
Slide38BiasingADL5390 uses differential signals biased around 0.5V
Complementary input has trim potentiometer allowing calibrationDC offset amplifier was tested, not satisfactory resultsCaused by big loop radius, custom PCB or alternative chip needed, was not worth effortWould have provided biasing internally, reducing setup complexity and allowed overvoltage protection
Not a must for functional prototype
Solution was to use offset from the arbitrary signal generator
38
Slide39Power supplyLab power supplies available are switch mode
Noisy and contains high frequency tonesPost regulator using LT3042Low noise, High PSRR and BW.Input low pass filter and ferriteQuick start-up to enable large reference capacitanceNoise floor lower than
oscilloscope
See datasheet for reference schematic
39
Slide40Simplified diagram of IQ modulator40
Slide41Finished prototype
41
Slide42Test of IQ modulatorPhase shifting speed
More complex combinations of I,Q data to see amplitude accuracy over complete rangeAmplitude accuracy affected by angle, amplitudeTest with spiral I,Q formPhase rotates 10 turns linearlyWhile amplitude sweeps linearly from 0 to 1160 Degree phase step
To compare against old setup
42
Slide43Test setup for IQ modulatorMeasures input for reference
Measures output directlyTrigger of I inputControlled by arbitrary signal gen.Signals read from USB stickIQ Demodulation software reused for reading output
43
Slide4444
Polar plot [spiral]
Slide45I,Q over time plot [spiral]
45
Slide4646
Amplitude, phase over time [spiral]
Slide47Phase step test comparison160 degree pulse tested to be comparable to old setupDone at half gain as max gain
Significantly fasterUsable for test of pulse compressor
47
Slide4848
160deg step comparison
Slide49Possible more tests that could be done
Single side band modulationSeeing ratio of wanted, unwanted sidebandIQ constellation diagram, seeing amount of possible positions in I,Q space without interference
49
Slide50Thoughts for improvement
Higher resolution controlHigher resolution arbitrary signal generator gives less noiseBetter quadrature coupler, done on PCB for good matchingHigher precision, more stable biasing
Post amplifier for use of lower gain and therefore higher speed
50
Slide51SummarySimulation of cavity is accurateOld phase shifters were too slow and not capable
New IQ modulator a success, significantly faster and 180 degree capable.Could be a solution in the future if tube and klystron based solutions become undesirable
51
Slide52Reflection
Project goal was to achieve an understanding of resonators and to simulate the behavior using lumped elementsImplement a phase function to achieve a compressed pulseOptionally focus on new hardware better suited for control of a pulse compressor
52
Slide53Lessons learnedCERN delivery system can be slow
Three weeks of waiting for componentsCrimping properly is importantFast control loop need be to very small, even on PCB level
53
Slide54ReferencesVector mod/demodulation diagram, slide 9
E.Jensen, CERN, CAS Chavannes 2013Pulse compressor diagram, slide 11X-band SLED type Pulse Compressor, made in CERN, I.
Syratchev
, 2013
ADL5390 block
diagram, slide 34
Analog devices datasheet for ADL5390
54