Transcript PPT

Diamond LLRF Activities
Pengda Gu
For DLS RF Group
The Diamond RF Group
Chris Christou
Alek Bogusz
Pengda Gu
Matt Maddock
Peter Marten
Shivaji Pande
Adam Rankin
David Spink
Alun Watkins
Agenda
•
•
•
•
•
Introduction of DLS RF System
Probe Blip and analogue gain control
Non-IQ phase mesurement
Direct sampling of 500MHz
DLLRF requirement for DLS
RF Systems
Storage ring: 500MHz
3 x 1 cell Nb Cornell cavities
3 x 300kW IOT combined amplifier
Booster: 500MHz
1 x 5 cell Cu PETRA cavity
1 x 60kW IOT
Linac: 3GHz
2 x 5.2m Cu DESY structures
2 x 35MW klystrons
bunchers
4
General SR RF Plant
DAs and
IOTs
Sys#2
Sys#3
*
*
500 MHz
IOT
DA
IOT
DA
IOT
DA
IOT
Circulator
Waveguide
Combiner
Phi rotate
I Q
Feedforward
Stepper
Motor
*
Open/Close
Loop
I Q
Set
Bandwidth VME Refl
Control
IOC Pwr
LLRF electronics
Two cavities in operation, each with its own
RF amplifier, LLRF and control system
*
I Q
Fwd
IQ Demod
Splitter
Sys#1
Interlock
MO
Splitter
IQ
Mod
Splitter
DA
Waveguide
Superconducting
combiner,
cavity
circulator and
load
I Q
Probe
Quench detector
Phase
and
Gain
Liquid Helium
Refrigerator System
DeRF
On/Off tune
SR RF Systems
RFTF
Cryogenic
Plant
HV PSU
Circulator
and load
LLRF , DAs
and Aux
Supplies
Superconducting
Cavities
IOTs and Combiner
Performance of the RF System
RF Instability
Dual cavity
Vcav 1 = 1.5 MV
Vcav 2 = 1.55 MV
Cavity voltage and power oscillation
LLRF phase error causes excessive
synchrotron motion
Probe problem
1. Cavities suffer from probe blips.
2. Probe blips happen with and without beam.
3. Probe blips happen on the main pickup which isn’t copper
plated.
4. Probes don’t have blips at the same time.
5. Probe blips don’t always trip the beam.
6. Very high voltage.
Cavity Probe RF Signal
‘Blips’ on individual probes
Cavity Signal
Spare pickup
Another spare pickup
Test of SSRF Cavity
Signal plot from SSRF FBT Near Bottom
Photo of pickup with and
without centre antennae
DLS and SSRF Cavity pickups are the same .
RF cabling is different.
Suspected to be caused by charging up of the ceramic insulator
on the pickup.
Trip due to RF Probe Blip
A close up of the PM below shows a loss of field for ~2 µs after which the
signal returns but the amplifier has already tripped on reflected power
Power ramp due to ‘loss’
of cavity voltage
Problem:
• fast rise-time spikes causing loss of RF signal on cavity HF pickup (and spare
probes). Up to 4 blips/day/probe.
• LLRF interprets this as reduction in cavity volts and pushes up the forward power
to compensate, causing a beam trip in some circumstances.
Cavity Probe Blips
Actions taken:
• Addition of filtering in the LLRF to reduce the bandwidth from 1MHz to 50 kHz.
This has been successful at preventing trips at low loop gains. No probe blip trips
since January 2014.
• Trial of a additional circuitry to detect the blip and reduce the loop gain during
the period of the blip (~2 to 6us typically).
Gain
IC
Gain
Control
Probe
I or Q
Ref
High pass
Rectify
Filter
Comparator
Simplified schematic of probe blip detection circuit
Cavity Probe Blips
Probe blip
detector PCB
mounted in
LLRF module
Signal break-out
connector
Tests using cavity simulator and blip simulated by blanking probe signal for 10us:
V Cavity
V Fwd pwr
Blip Detector o/p
Trigger
Circuit disabled
Circuit enabled
Cavity Probe Blips
Storage ring tests @ 300mA with LLRF#3 modified to include blip circuit:
LIBERA box, Cavity , PFwd and beam (X)
motion signals:
Probe blip ->
Attack and decay timeconstants of gain
reduction need
optimisation to
minimise beam
disturbance.
Residual beam kick
from blip ±50 µm
Phase Measurement Unit
Two units finished and installed in RF hall.
Mf RF  Nf SF
Direct RF sampling
Phase advance between two
samples
Then calculate I
and Q
2
I
M
2
Q
M
  2
N
M
M 1
 y sin( i   )
i 0
i
M 1
 y cos(i   )
i 0
i
Xilinx ML605 Development Board
FMC150 module uses Texas Instruments
ADS62P49 dual 14-bit, 250 MSPS ADC
ADC clock generation constructed using
ADS9912 DDS
Algorithm Implementation
data_tready
Sy stem
Generator
 ++
Out
a
z-1
a=b
Counter
Gateway Out
Scope2
b
14
data3.mat
In
From File
Gateway In
data_tdata
Relational
data_tvalid
Delay
Constant
data_tdata
 15
Down Sample
FIR Compiler 6.2
x
mag
Out
Gateway Out-A
z -11
Scope
data_tready
y
atan
Out
CORDIC ATAN
data_tdata
data_tvalid
data_tdata
FIR Compiler 6.2 1
Test Results
Amplitude RMS Jitter
Amplitude RMS Jitter (%)
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
-20
-10
0
10
20
30
Signal Power Level set at the Signal Genrator (dBm)
Gateway Out-Phi
-KGain
Scope1
 15
Down Sample1
output1.mat
To File
A deliberate 0.06 deg phase modulation is clearly seen from 3dBm.
ADS62P49 channel A, RF level -3dBm at the signal generator with 0.06degree phase modulation
-149.3
-149.4
Phase
-149.5
-149.6
-149.7
-149.8
-149.9
0
200
400
600
800
1000
1200
Sample number
No averaging, smoothing etc
Sampling at ~ 241 MHz
Phase measurement ~ 16 MHz
1400
1600
1800
2000
IQ Plot of Demodulated Signal from -10 to 13dBm with 287degree Phase Modulatio
1.5
8
x 10 IQ plot of demodulated signal, power level from -10 to 13dbM at the signal generator, phase modulation 287degree peak-peak
Phase measurement at discrete
power levels.
1
Minimal distortion over dynamic
range
0.5
Advantages of this method:
• No imbalance between I and Q
channels often observed using
IQ demodulators
• No DC offset errors
0
-0.5
-1
-1.5
-1.5
-1
-0.5
0
0.5
1
1.5
8
x 10
Results from cavity 3 probe signal
Phase variation and the single bunch can be clearly recognized.
~ 1 deg
Bunch train
Hybrid
bunch
Phase data
Bunch gap
Close up shows clear phase shift across
bunch train
Direct Sampling and Demodulation of 500MHz Signal
TI ADC12D1800 sampling at 2.0 GSPS .
Sample by sample demodulation
 I   cos n  cos( n  1)   y n1 

   
  
 Q    sin n sin( n  1)   y n 
  90 0
Test in the Lab
Test during normal Operation
Transient of RF field can be seen clearly.
Phase variation can clearly be observed.
Beam loading can clearly be observed.
The single bunch in the gap can clearly be observed.
Signal from the Probe at FBT
Different from the LLRF probe due to relative amplitude of RF and beam signal.
LLRF basic Functionalities
 Vector control of cavity accelerating field
0.15 degree and 0.05% of amplitude (RMS value).
 Cavity resonance control
The cavity will be tuned to minimize the total forward power needed.
 Reduce the effective cavity impedance apparent to the beam
Increase the Robinson instability threshold. Limited by the loop delay and the
closed loop gain.
R
2 R
Rmin 
1 G

Q
0T
 Reduce noise from different sources in the system
Switching noise of IOT high voltage power supply.
 Ramping of cavity field
 Interlocks and equipment protection
 Probe blip blockage
LLRF desirable functionalities
 Post-mortem function or triggered data output
 Testing and self-calibration
 Baseband network analyzer
 Conditioning mode
 Feed forward algorithm, comb filter (one-turn delay feedback)…..
 Amplifier linearization
Summary
• DLS LLRF system has been in operation for more than 9years.
• No major problems experienced.
• Digital LLRF is planned for SRF cavity and NC cavities.