Ring BPM Electronics

1
Ring BPM Electronics
2
Outline

• Schedule
• July 2002 Design Review
• Requirements
• Issues
• Approach
• Signal Processing
• Status
• Acceptance Criteria, Brochure and ICD information

3
BPM Schedule

• Electronics for the RTBT Ring on site starting
  April 2004
• Majority of Pickups complete

4
2002 DAC Review Concerns

• BNL staff voiced concern about the long term
  reliability of the pressure contact used between
  the BPM electrode and the vacuum feed through
  connection. We suggest BNL and SNS staff resolve
  these concerns ASAP.
• A review of the manufactures specifications and
  calculation of the maximum mechanical and
  electrical stresses on the bellows contact have
  proven that the contact will be operating well
  within its specification limits.
• The committee was provided with the BNL
  specification prepared for the BPM front-end
  electronics vendor and found numerous
  shortcomings
• The specification was revised to include the
  improvements suggested by the committee. A
  decision was made to proceed with an in-house
  design to meet the requirements..
• We are concerned whether the described design of
  the calibration system is sufficient for
  operationsit seems desirable to also use the
  calibration system for cable and/or electrode
  centering checks,etc.
• The calibration system concerns have been
  addressed and will be discussed in the
  presentation for the RF electronics.
• There was little detail provided of the BPM
  signal cabling system We have special concern for
  the last few feet of the cables connecting to the
  BPM electrode connectors.
• The cabling system always has been part of the
  baseline design. Short lengths of flexible
  coaxial cable are included to connect to the PUE
  and the electronics.

5
2002 DAC Review Concerns 2

• If nonlinear correction factors for beam
  positions not close to the center need to be used
  to meet accuracy requirements then those
  quantities need to be included in the software
  specifications
• Field maps will be provided for each size BPM.
  The software will be developed in conjunction
  with testing of the electronics and pickups so
  that nonlinearities can be compensated..
• There appears to still be some confusion and/or
  lack of consensus about the beam position and
  beam current ranges over which the accuracy
  requirements apply
• We have reached a consensus with the AP group.
• Dead time due to gain switching should be
  quantified and passed on to AP personnel.
• Gain will be switched during the gap and will
  require 130nS.
• How the BPM gain setting/read-back will be
  addressed was not presented, nor was how many
  samples are lost during the gain switching
  process.
• Gain read-back will be made available to the
  operator with each data set. Switching the gain
  will cause no loss of turn data.

6
2002 DAC Review Concerns 3

• Electron cloud effects may have a big impact on
  the operation of the BPMs in the ring and RTBT as
  there is a potential for electron interaction
  with the Electrodes. Work towards understanding
  the impact of electron currents on signal
  integrity and overall reliability of the
  electronics.
• The SNS is a stable, DC machine. The position of
  the beam will be known before electron cloud
  effects become apparent each cycle.
• Protection of the electronics input will be
  provided
• There continues to be talk of using BPM
  electrodes to generate electron clearing fields.
  Other than the insulated pickups electrode design
  there is no provision for this in the system
  design
• Baseline does not include electron clearing
  fields. If required, ORNL will provide
  additional hardware.
• There was discussion about implementing a dual
  analog front-end scheme.
• Dual band capability will be part of the design.

7
BPM Ring - RTBT Requirements

• Intensity 5e10 to 2e14 Protons
• Range /- 100 mm
• Accuracy 1 of half aperture
• Resolution 0.5/1 of half aperture
  averaged/turn-turn
• Data Structure turn-by-turn (except during gain
  switching)
• dual plane PUE at each quad

8
Quantities

• HEBT
• 12cm bpm 37
• 21cm bpm
• TOF systems 6
• Ring (Floated stripline)
• 21cm bpm 28
• 26cm bpm 8
• 30cm bpm 8
• RTBT (Shorted stripline)
• 21cm bpm 15
• 36cm bpm 2

Electronics LANL BNL BNL
Cables ORNL Specs BNL
9
Ring BPM Electronics Design Issues

• Particles diffuse into a single bunch after a few
  turns. There is little base-band energy in the
  first turn.
• BPM PUE is an open stripline and has little
  sensitivity at base-band, higher sensitivity at
  RF.
• Nominal beam current dynamic range is 30001,
  required resolution is 0.5 to 1 of half
  aperture (0.1mm for a 105mm radius) requiring an
  additional 10001 for a total dynamic range on
  the order of 60dB.
• Calibration must deal with the dynamic range of
  the machine.
• A dual path approach will be applied to address
  these issues.

10
Normalized Spectrum of a Pulse 645ns duration
945ns Period
Bunch Shape
11
S/N Ratio Estimate 50 Ohm Noise
12
Bandwidth Selection

• S/N is relatively constant with bandwidth for a
  given beam current (see earlier slide).
  Therefore we require an additional criteria to
  establish system bandwidth.
• Require settling within beam GAP time about 250ns
  max
• A Gaussian filter will take about 6 time
  constants to settle
• For 200ns this sets the filter time constant to
  33.3ns
• This is a bandwidth of 4.777MHz
• A system bandwidth of 5MHz has been selected

13
Analog Processing

• Method - RF detection of first turns followed by
  Baseband detection for remaining turns.
• Use protected amplifiers with inter-pulse gain
  switching to cope with the large dynamic range
  requirements.
• Linear AFE with data processing in LabVIEW for
  flexibility.
• Allows choice of processing method.
• Beam Based Alignment calibration.
• Possible use for other Diagnostics.

14
Ring BPM Signal Conditioning Amplification
RF/MIXER/BASE-BAND BOARD
BASE-BAND BOARD
LP
IN
10dB
0dB
HP
x
0dB
Protected amp
BW10MHz
BW10MHz
CAL
10dB
ANTI-ALIAS FILTER 8.5 MHz 5 pole
MIX
CAL enable
X
ADC
LO 400MHz
LO enable
20 to 50 dB Gain Range
Base-band -20dB to 20dB Gain Range RF/MIXER
2.5MHz out 0dB Gain
15
Processing Techniques

• RF Incorporate in Base-band system for early
  turns and increased sensitivity
• Base-band
• Amplify, Filter, digitize at 64 times revolution
  freq.
• Process signals using mean square to provide an
  average signal for an entire mini-pulse.
• Use Log-Ratio approach to improve linearity with
  displacement and eliminate gain errors after
  Beam-based alignment.

16
Block Diagram

SSRAM
SIGNAL DIGITIZERS
PCI BUS

ALTERA FPGA
ADC Clock.
Amplifiers and Signal Conditioning
T 0 TRIGGER
PUE
Control I/O
Event Link
EL RTDL Receivers
RTDL
Power Supply
17
BPM Board Arrangement
Riser Board
PCI Connectors
PCI / Digitizer / Baseband Board
RTDL EL Inputs
HF Board
Shield - if needed
Motherboard
SMA Inputs
18
Status

• Analog Front End undergoing Testing
• Initial testing revealed PC noise near 14MHz.
  This was removed by changes to the anti-aliasing
  filter.
• Noise floor estimates for the first prototype
  indicate about 2mV equivalent output noise or
  about a s/n of 53dB for a 1 volt peak signal
• PCI board, with timing decoder incorporated, is
  in layout
• Includes an Altera FPGA with greatly increased
  capacity and flexibility compared to the current
  LANL design.
• Event Link RTDL receivers on board
• Can include signal pre-processing
• RF board design in shop for artwork/prototype
  construction

19
Diplexer Testing
Low - Pass
High - Pass
SWR with LP connected to 8753, HP terminated, and
damping resistors 330 Ohms across the grounded
coil in HP case, and the first coil in LP case.
20
Test Setup
21
Prototype Base-band Board Testing
30dB Gain, anti-aliasing Filter BW17MHz
50dB Gain, anti-aliasing filter BW5.5MHz
22
Comparison of Channels
Comparison of channels 1 (white) and 2 (Yellow).
Gain of channel 1 is 50dB, gain of channel 2 is
30dB. Signal attenuated 20dB for channel 1 to
yield similar signal levels on each channel.
Channel 1 anti-aliasing filter set to 8.5MHz and
Channel 2 filter set to 17MHz. PC noise near
15MHz is reduced for channel 1. Calculations of
s/n indicate near 16dB for channel 1.
23
15 mA Beam, 50 dB gain
24
Acceptance Criteria

• Acceptance tests
• Simulate input signals
• Measure noise and record
• Measure bandwidth and record
• Calibrate Gains record all gains
• Demonstrate Control using EPICS interface
• Burn in and final test
• Include test equipment certification(s) with the
  Traveler
•  

25
BPM System Brochure
Number of PUEs   Ring 21 cm Open
Stripline 250 mm 70 degree quantity 28
Ring 26 cm Open Stripline 250 mm 70 degree
quantity 8 Ring 30 cm Open Stripline 250 mm 70
degree quantity 8   RTBT 21 cm Shorted
Stripline 250 mm 70 degree quantity
15 RTBT 30 cm Shorted Stripline 250 mm 70
degree quantity 2 AP requirements Intensity
5e10 to 2e14
Protons Pulse Length 0.3 to 1000
uS Range /- 100
mm Accuracy 1 of half
aperture Resolution 0.5 / 1
of half aperture averaged / turn-turn Data
Structure Turn by turn (except during gain
switching) Target requirements AP requirements
as above plus Bandwidth 5 MHz. Update
rate 6 Hz. Calibration On board cal.
Signal to PUE and Beam Based Alignment Sample
rate 64 x Revolution Frequency No loss of
turn data during gain switching  
26
BPM ICD Information
Process Variables X position averaged over the
macropulse (mm) Y position averaged over the
macropulse (mm) X position array, holding
position over each minipulse (mm) Y position
array, holding position over each minipulse
(mm) X position array, holding position over each
minipulse (mm) Y position array, holding position
over each minipulse (mm) Length of the averaged
period (microseconds) Delay time from Tzero to
beginning of average period (microseconds) Sample
period of each TBT array element ( of
turns) Calibraton control (s) Modes RTDL Event
Link ( Required events and event timing to be
added) Power Requirements 115 VAC, Single
Phase, 300 Watts max
27
Back - Up
28
Protected Low-Noise Amplifier (20dB)
G -10

• High Voltage Input Stage
• Fault protected high voltage switch
• Gain controlled low-noise stage
• Transformer coupled impedance matched for
  improved noise figure

G20
G -0.5
Attenuator 0.2
HSMS2812 Shottky
THS4021
/-15V supply
T4
Protects next stage against preamp failures
124K
Zi402
Zi100
OPA642
/-5V supply
MAX4632
Transorb
4420
25V Fault protected /-15V supply
Expected noise about 1.5nV/rt. Hz _at_ 1MHz
29
Filters-Two Gaussian and One ChevyshevAnti-aliasi
ng and transient response
Simulated dual filter response 2 each 5-pole 7MHz
Gaussian followed by 1 ea. 5-pole 10MHz 0.01dB
Chebyshev
Pulse response, showing the two Gaussian filters
with good transient response, and the over-all
response with 6 overshoot due to the Chebyshev
filter.
30
BPM Processing Linearity to Displacement
A plot of sensitivity along the axis of a pair of
pick-up elements for a 70 degree stripline
designed BPM with a half aperture of 105mm. The
sensitivity is shown to be 0.31dB per mm.
Linearity is shown to be reasonable over a
range of /- 65mm.
A plot of sensitivity along the axis of a pair
of pick-up elements for a 70 degree stripline
designed BPM with a half aperture of 105mm. The
sensitivity is shown to be 0.018 per mm.
Linearity is shown to be reasonable over a
range of /- 20mm.
31
Processing Simulation - 1
Approximate the signal (response to exponential)
Modify to include response to pulse
Provide Scaling
Current Pulse
32
Processing Simulation - 2
Simulate noise sources and signal
33
Processing Simulation Comparison 15ma Beam 1mm
displacement 210mm diam BPM
S/N Estimate Table 15ma beam - 5 MHz Bandwidth
S/N 50 Ohms 49dB Amp noise 10dB Cable
atten. 2dB Addl atten. 6dB Switches
filter 6dB RESULT S/N 25dB
Signal simulations confirmed by comparing results
of LabVIEW, PSPICE, MathCAD, and EXCEL
spreadsheets computing Shafers equations. See
back-up slides for description of simulation
effort used to create this chart.
34
Compute Output and Calculate Errors
Peak Signal - Difference Over Sum
Mean-Square Difference Over Sum
With noise subtraction
For no noise
Log Ratio of Mean-Square signals
35
Mean Square processing
The output is approximately
(S is the rms signal, with beam at (r,?), and BPM
half aperture b and angular width ?)
Remember a is
The minimum resolvable displacement is
approximately
36
Comparison of Estimated Resolution to a stdev
calculation
37
Calibration - Approach

• Offset Compensation
• Apply Beam based Alignment
• Compensate for mechanical, electrical, and
  channel gain offset contributions
• Initial measurements can be approximately
  compensated using channel gain measurements
  discussed later.
• Trouble shooting
• All cabling checked with TDR during installation
• Obtaining a signal on each channel with beam will
  assure connectivity
• Calibration
• Inject a pulse on each channel, measure response
  on same and adjacent channels after cable delays
  (TDR).
• Confirms connectivity without beam
• Confirms channel integrity
• Checks Gain stability
• Pulse edges Ring a high Q filter due to the
  reflected pulse, checks RF section

38
Calibration

• Beam-based alignment
• Talman Malitsky (BNL SNS Tech Note 116)
• Requires dedicated beam time.
• Introduces a systematic fractional quadrupole
  strength change to a group of 8 in series.
• Orbit smoothing algorithm corrects for fractional
  change yielding kicker values that minimize the
  badness.
• Individual quad misalignments are inferred from
  the kicker strengths and the quadrupole
  strengths.
• All BPM offsets are recorded to enable subsequent
  use of the BPMs as secondary standards.
• Readily includable in the present operational
  control software
• For a change of 1 in quad strength the
  algorithm is essentially unaffected by changing
  the chromaticities from their natural value, and
  is little affected by inclusion or exclusion of
  magnet imperfections.
• With a 1 change, the steering accuracy is 100
  times worse than the measurement accuracy.
• To achieve 0.1mm steering accuracy will require
  about 1?m measurement reproducibility. (Short
  term, not even long term relative accuracy.
  Perhaps achievable by using a very low frequency
  excitation with lock-in detection.)

39
Beam Based Alignment

• Paper by Talman Malitsky, SNS/BNL Tech Note
  116 indicates
• To achieve 0.1mm steering accuracy requires about
  1?m reproducibility.
• Earlier estimates indicate a S/N of 25dB is
  achievable for a 15ma beam.
• To achieve a 15ma 1 ?m resolution requires about
  a S/N77dB (see chart)
• A 38ma beam would provide about a 25833dB S/N
• Additional noise reduction of 44dB is required
• Averaging over 1000 turns improves noise by 31.5
  or 30dB
• Extra improvement must come from a larger signal
  (14dB more signal or 5 injected turns (191ma)) or
  more turns. A minimum of 5 injected turns is
  required to average over a single macro-pulse.
  Alternatively a single turn 38ma beam requires
  about 25119 turns to be averaged (26
  macro-pulses, 0.42 seconds).
• Channel gain ratio is a constant and subtracts
  when the BPM offset is subtracted.

40
First Turns Processing
Low Frequency Processing
Diplexer
CPL -20dB
1st 3-Turns
ADE-1 (L7)
400MHz BP Helical Filter TDW2436A-400M
5MHz LPF SCLF-5
3dB Pad
50
7dBm
Cal
Cal Signal
50
3.3V
FoN/RFref
Split 4-way 0-deg
Silicon Labs RF/IF Synthesizer Si4133
PWDN
Lo Ch.2
SDATA
Lo Ch.3
SCLK
Lo Ch.4
SEN
22dB
Ref Freq 16Frev (16MHz)
AD4PS-1
41
Si4133 Synthesizer
62.5 to 1000MHz (with divider)
42
BPM Calibration Concept
Ch.1
Altera FPGA
a11
CPL -20dB
Cal Pulse (tw100ns)
MPU Calc Cal Coeff
Ch.2
a12
CPL -20dB
PCI
Ch.3
a13
Reflected Pulse (a11)
CPL -20dB
Open Ckt Stripline 21cm
Ch.4
a14
CPL -20dB
Example Inject on Ch.1, measure Response on
all 4-channels