Improvements of the LLRF system at FLASH
Mariusz Grecki, Waldemar Koprekand LLRF team
Agenda
• GUN linearization• Adaptive feed-forward at ACC1• Beam load compensation at ACC1• Klystron nonlinearity compensation• Detuning measurement and Piezo control
Gun linearization
Where is the problem?
RF-Gun setup
High power chain linearizationPfor Q Pref I Pref QPfor I
Set-PointTable
Feed-ForwardTable
+
FeedbackGAIN
FPGA CONTROLLER
CalibrationOffset
DAC1 ADC5
RotationMatrix
RotationMatrix
I Q
I
Q
I
Q
I Q
I Q
I Q
I Q
DAC2
VECTOR MODULATOR
Timming & ControlModule
DigitalInput 1
DigitalInput 2
TIRGGER 1MHz
Offset
x
CalibrationOffset
ADC6
CalibrationOffset
ADC7
CalibrationOffset
ADC8
I Q
+
I Q
Filtery=N*xn+(1-N)*yn-1
I
Q
IntegratorGAIN
x
+
I
QI
Q
∫
-+
Linearization
VME
DOOCS SERVERFF table calculationAmplitude SP
Phase SP
Corr.tables
Measurement devices of GUN forward power
waveguide
Klystron
to RF-Gun
diode
A/φ detector
I/Q detector
PforPref
High power chain linearization based on amplitude detector measurement
phase [deg]
norm
aliz
ed k
lyst
ron
pow
er v
aria
tion
Temperature behaviour during linearization
High power chain linearization based on diode measurement
phase [deg]
norm
aliz
ed k
lyst
ron
pow
er v
aria
tion
Temperature behaviour during linearization
Temperature behaviour during linearization
Phase of diode vs forward power measured in SIMCON
norm
aliz
ed k
lyst
ron
pow
er v
aria
tion
phase [deg]
Phase scan of forward power
Rotation matrix
cos(α) -sin(α)sin(α) cos(α)=
I2
Q2
I1
Q1x
x
y
V1
V2
α
A11*cos(α) -A12*sin(α)A21*sin(α) A22*cos(α)=
I2
Q2
Iin
Qinx x
y
I1I2
Q1
Q2
Ir1 Ir2
Qr1
Qr2
Non-orthogonal I & Q
I
Qφ
Iout = Iin + Qin* sin(φ)Qout = Qin* cos(φ)
New DOOCS control panelsA11*cos(α) -A12*sin(α)A21*sin(α) A22*cos(α)=
Irot
Qrot
Iin
Qinx
Iout = Iin + Qin* sin(φ)Qout = Qin* cos(φ)
Input calibration panel Advanced input calibration panel
Phase scan of forward power
Before and after input linearization
Adaptive FF at ACC1
CAVITY 1-8
Set-PointTable
Feed-ForwardTable
Exceptiondetection
-
DAC
SIMCON 3.1 BOARD – FIRMWARE
ADC1-8
I/Q detector
RotationMatrix
VECTORSUM
Klystron correction tables
VECTOR MODULATOR
Timming & ControlModule
DigitalInput 1
DigitalInput 2
TIRGGER 1MHz
ExceptionHandling
BIS AND/OR RF GATE
Offset
DigitalOutput 1
Error signal
IRQ to VME
AFFtable+
MIMO+
Beam loadingcompensation
ADC9
+
TOROID
DAQ VME40 signals from SIMCON
Loop delay regulator
PowerPC
Averaging
Errortable
IRQ
Xilinx Virtex II Pro
AFF vs FB
FB, Gain=25, AFF OFF FB, Gain=5, AFF ON
AFF used for beam load compensation
AFF OFF AFF ON
AFF status
• implemented in FPGA and PowerPC• two versions of AFF tested• stable operation over minutes• control through virtual RS-232 port• DOOCS control panel not ready
Beam load compensation at ACC1
CAVITY 1-8
Set-PointTable
Feed-ForwardTable
Exceptiondetection
-
DAC
SIMCON 3.1 BOARD – FIRMWARE
ADC1-8
I/Q detector
RotationMatrix
VECTORSUM
Klystron correction tables
VECTOR MODULATOR
Timming & ControlModule
DigitalInput 1
DigitalInput 2
TIRGGER 1MHz
ExceptionHandling
BIS AND/OR RF GATE
Offset
DigitalOutput 1
Error signal
IRQ to VME
AFFtable+
MIMO+
Beam loadingcompensation
ADC9
+
TOROID
DAQ VME40 signals from SIMCON
Loop delay regulator
PowerPC
Averaging
Errortable
IRQ
Xilinx Virtex II Pro
20 ns
Sampling of toroid signal
70 nst
30 bunches measured by SIMCON
before – clock=50MHz – local clock after – clock=54MHz – clock from MO
Beam load compensation
Klystron nonlinearity compensation
(more general:High Power Amplifiers
nonlinearity compensation)
Current high power amplifiers diagnostic hardware status
GUN
Vector Modulator
LLRF amplif ier
20dBmax out 0dBm
Preampli fier
max out
400W
Klystr onDirec tional c oupler
LO
ADC
atten uator
LO
attenuat or
Direc tionalcoupler
LLRF CONTROLERDSP/FPGA
ADC
ADC DOOCS Servers
Power s pl itter
(3 dB)
attenuator
ADC
LO
attenuato r
ADC
LO
Circ ulator(~2dB)
Available DOOCS signals:TTF2.RF/ADC/KLY3_VM/CHANNEL.TDTTF2.RF/ADC/KLY3_PAMPL1/CHANNEL.TDTTF2.RF/ADC/KLY3_PAMPL2/CHANNEL.TDTTF2.RF/ADC/KLY3_OUT/CHANNEL.TD
Directional coupler
Current high power amplifiers diagnostic hardware status
Acc1
Vector Modulator
LLRF amplif ier
20dBmax out 0dBm
Preampli fier
max out
400W
Klystr onDirec tional
c oupler
LO
ADC
att
LO
att
Direc tionalcoupler
LLRF CONTROLERDSP/FPGA
ADC
ADC DOOCS Servers
Power s pl itter
(3 dB)
att
ADC
LO
Power spl itter
(3 dB)
att
ADC
LO
Circulator(~2dB )
Available DOOCS signals:TTF2.RF/ADC/KLY2_VM/CHANNEL.TDTTF2.RF/ADC/KLY2_PAMPL1/CHANNEL.TDTTF2.RF/ADC/KLY2_PAMPL2/CHANNEL.TDTTF2.RF/ADC/KLY2_OUT/CHANNEL.TD
Current high power amplifiers diagnostic hardware status
Acc2_3Available DOOCS signals (temporary location):TTF2.RF/ADC/ACC3.TOTAL/CH02.TD – DAC output ITTF2.RF/ADC/ACC3.TOTAL/CH03.TD – DAC output QTTF2.RF/ADC/ACC3.TOTAL/CH04.TD – after RF gateTTF2.RF/ADC/ACC3.TOTAL/CH05.TD – after 1 st preampTTF2.RF/ADC/ACC3.TOTAL/CH06.TD – after 2 nd preampTTF2.RF/ADC/ACC3.TOTAL/CH07.TD – after klystron
Vector Modulator
Power splitt er (3 dB)
LO
LLRF ampli fier
20dBmax out 0dBm
Preampli fiermax out
400W
Klys tronDirec tional
coupler
attenu ator
LO
ADC
at ten uator
LO
attenuator
Directional coupler
LLRF CONTROLERDSP/FPGA
ADC
ADC
ADC DOOCS Servers
Power spl itter (3 dB)
attenuator
ADC
LO
Current high power amplifiers diagnostic hardware status
Acc4_5_6 Available DOOCS signals (temporary location):TTF2.RF/ADC/ACC5.TOTAL/CH02.TD – DAC output ITTF2.RF/ADC/ACC5.TOTAL/CH03.TD – DAC output QTTF2.RF/ADC/ACC5.TOTAL/CH04.TD – after RF gateTTF2.RF/ADC/ACC5.TOTAL/CH05.TD – after 1st preampTTF2.RF/ADC/ACC5.TOTAL/CH06.TD – after 2nd preampTTF2.RF/ADC/ACC5.TOTAL/CH07.TD – after klystron
Vec tor Modulator
Power splitt er (3 dB)
LO
LLRF ampli fier
20dBmax out 0dBm
Preampli fiermax out
400W
KlystronDirectional
c oupler
attenuator
LO
ADC
attenuato r
LO
attenuator
Directional coupler
LLRF CONTROLERDSP/FPGA
ADC
ADC
ADC DOOCS Servers
Power splitter (3 dB)
attenuator
ADC
LO
TO BE INSTALLED
HPA AM/AM and AM/PM characteristics measurements.
• The set of amplitude-amplitude and amplitude-phase characteristics measurements have been done for the klystron 5 and klystron 2 during last two years.
• The constellation diagram method have been used for the characterization results visualization.
• Basing on achieved characteristics the correction coefficients have been calculated.
High power chain non-linearities characterization
Non-linearities and saturation phenomena:
• increasing the driving power ⇒non-linear amplifier behaviour
• constant increasing of driving power ⇒ saturation
• different saturation level for a different operation conditions
Complex representation of the HP chain devicesExample for kly. 5 (each axis unit is an ADC voltage)
Signal parameters:Pulse length – 1200 us,Number of steps – 50,Signal range – 0 up to max. available level
I
Qtime
I max
tp
Results example – klystron 5
Constellation diagram:Measurement for one phase (Q=0). Klystron output characteristics for different HV levels.
Constellation diagram:Grid measurement with 20 steps resolution
KLYSTRON 5
1st preamp 2nd preamp Klystron output
DAC output VM output
Results example – klystron 2
Constellation diagram:Grid measurement with 50 steps resolution
Constellation diagram measurement:Measurement for one phase (Q=0).Klystron output characteristics for different HV levels.
HPA's Linearisation algorithm principles.
linear char.
corr amp.
req amp
controler output signal
max amp.
real char.
Input amp.
Output amp.
max
Output phase[deg]
Input amp.max
controler out. signalcorr. amp
Phasecorrection
From the nonlinearity measurement the AM/AM (amplitude to amplitude) and PM/AM (phase to amplitude) of the high power chain can be achieved. NOTE!! The nonlinearity is only function of input amplitude.
Driving signal representation:Z = Id + Qd = |Z| * [cos(phi) + i * sin(phi)]
Correction signal:C = Ic +Qc = |C| * [cos(th) + i * sin(th)]
From the linearisation both amplitude and phase correction are achieved. Can be realised using the complex multiplication.
C*Z = Idc + i*Qdc C*Z = ||Z|*|C||*[cos(phi+th)+i*sin(phi+th)]
Linearization tool implementation in LLRF field controller (Simcon).
• The linearization tool realization is based on the set of look-up tables • The amplitude of the controller driving signal (calculated in the FPGA) is used for
addressing the look-up tables with correction coefficients for I and Q. There are 32 word 18bits tables with corrections calculated in the MATLAB from the characteristics achieved during the characterization. In order to minimize the tables size (save the FPGA resources) the tables with linear interpolation between the knots.
• The contents of the tables is updated – according to the changes of the HV level (for instance adjusted by an operator).
Amplitudecalculation
Rotationmatrix
I lin.
I corr. I corr. Rangetable
I curve slope range table
Amp. Range32 posit.tables
Proportional feedback controller
Q corr. Q corr. Rangetable
Q curve slope range table
Amp. Range32 posit.tables
I dQ d
Ic
Qc
Q lin.
Tool tests results in ACC2&3 and MTS.
• The set of in-situ tests were performed in the MTS and the FLASH in order to evaluate performance of different configuration of linearization tool.
• During the characterization phase the nonlinearities of the amplitude and phase characteristics have been determined for the different HV level of the klystron modulator.
• Achieved data was processed and used for the correction coefficients calculation. Depending on tested variant of configuration the tables with 4, 8,16 or 32 positions were up-loaded to the FPGA.
• Although the most often used configuration was 32 positions tables with linear interpolation, others were also possible but required the LLRF controller reconfiguration (recompilation).
Examples of tool tests results in MTS.
Study of the RMS error of vector sum error signal in function of LLRF feedback loop gain. Black traces – with linearisation blue and red trace – without linearization
Amplitude and phase characteristics of MTS HPC nonlinearities (blue trace) and linearized characteristics (red trace). Characteristics for modulator HV level 9,4kV
Klystron 5 HPC linearisation results• Linearisation test had been performed using Simcon(FPGA)
controler,• Correction tables were „on” • HV level – 10800 (value on PLC) about 110kV• Two iteration of the linearization were performed.
Current work – linearization tool implementation in ACC1 LLRF
feedback controller.• Linearization tool will be installed in the ACC1 LLRF
control loop controller. • Appropriate modification of the controller structure,
dedicated controller DOOCS server, and Matlab scripts have been done.
• The offline tests of the tool performance is planned for July and August 2007.
• In-situ tests before regular operation will be performed during August-September 2007 accelerator study period in FLASH.
Conclusions.
• Diagnostic setup installation ready for the high power chain amplifiers examination has been prepared for most of the FLASH modules.
• Linearization method that can be implemented in standard LLRF feedback loop controller have been developed and tested.
• Successful tests of the linearization tool have been performed in the MTS and the HPC of ACC2&3.
• Tool implementation in ACC1 field controller is in progress.
Detuning measurementand Piezo control
Detuning compensationcontrol system
FPGA
DAC
ADC
ADC
PIEZO DRIVER
PIEZO SENSOR
RF PROBES
ACTUATOR
PIEZO STACK
SENSOR
downconverter
Simcon3.1L
Detuning block
Detuning computation block
atan atan
Forwardcalibration
Probesignal
divisionsubstraction
Ampprobe Phsforwa
sincos
Phsforwa-Phsprobe Ampforwa/Ampprobe
detuning
6 clk
4 clk
4 clk
FIR
Phsprobe
substraction
1 clk
1 clk
3 clk
19 clk
Ampforwa
Data acquisition block
MT1 MT2
2 ms 98 msII readingDAQ recording
10 Hz of RF pulse repetition rate
Arbitrary access of II to DAQ internal memory
Measurements in ACC7/CAV6
0 500 1000 1500 2000 25000
5000
10000
15000
0 500 1000 1500 2000 2500-1.5
-1
-0.5
0
0.5
1x 105
0 500 1000 1500 2000 25000
000
000
000
000
000
000
000
000
000
000
0 500 1000 1500 2000 2500-14
-12
-10
-8
-6
-4
-2
0
2
4
6x 104
400 600 800 1000 1200 1400 1600 1800 2000
-1
0
1
2
3
4
x 104
amplitude and phaseForward power
amplitude and phaseProbe signal
Detuning
0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22-4000
-3000
-2000
-1000
0
1000
2000
3000
0.097 0.098 0.099 0.1 0.101 0.102 0.103 0.104 0.105
-1200
-1000
-800
-600
-400
-200
0
200
Probes debug Pulse debug
200 400 600 800 1000 1200 1400 1600 1800 2000
0
100
200
300
400
500
600
700
X: 544Y: 226.2
time[us]
detu
ning
[Hz]
ACC7/CAV6 - online detuning measure
X: 1205Y: 52.8
Detuning measurement
filling
flat
decay
FT detuningequals to 180 Hz
Amplitude and Phasemeasurement
0 500 1000 1500 2000 25000
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
time[us]
grad
ient
[MV
/m]
ACC7/CAV6 - Amplitude of probe signal
0 500 1000 1500 2000 2500-180
-160
-140
-120
-100
-80
-60
-40
-20
time[us]
degr
ees
ACC7/CAV6 - Phase of probe signal
Probe signalafter coodrinate conversion
Microphonics
100 200 300 400 500 600 700 800 900 1000
0.5
1
1.5
2
2.5
3
3.5
4
4.5
x 104
X: 131.3Y : 4.372e+004
X: 431.3Y : 1.486e+004
X: 731.3Y : 6241
freq [Hz]
ampl
itude
[a.
u.]
abs(fft)
mechanical cavity resonancearound 130 Hz
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35-295
-290
-285
-280
-275
-270
-265
time[s]
ampl
itude
[a.u
]
Piezo driver results (1)
0
20
40
60
80
100
120
0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 1,1 1,2
Umi [V]
Umo [V]
0,00
50,00
100,00
150,00
200,00
250,00
300,00
350,00
400,00
Imo [mA]Gu dla Cpiezo = 2,47uF Gu dla Cpiezo = 3,37uFGu dla Cpiezo = 5uF Imo dla Cpiezo = 2,47uFImo dla Cpiezo = 3,37uF Imo dla Cpiezo = 5uF
Half a sine
Piezo driver results (2)
y = 0,3067x2 + 0,909x + 26R2 = 0,9012
0
20
40
60
80
100
120
140
0 2 4 6 8 10 12 14 16 18
N
θ[°C]θ(N)2nd order polynomial aproximation θ(N)
rising number of pulsespiezo 5 uF
Imo = 510 mA
Piezo driver results (3)
A
B
C
A
B
C