fp 420 bpm workshop signal processing for bpms marek gasior cern-ab-bi

FP 420 BPM Workshop 1M. Gasior, CERN-AB-BI

FP 420 BPM Workshop

Signal Processing for BPMs

Marek GASIOR

CERN-AB-BI

2Comparison among signals processing for BPM

BIW

-00 M

ay 8

- 1

1

Comparison among signals processing Comparison among signals processing for BPMfor BPM

G. Vismara G. Vismara

Introduction Introduction

Signal analysisSignal analysis

Design parametersDesign parameters

System familiesSystem families

System descriptionsSystem descriptions


BIW

-00 M

ay 8

- 1

1

IntroductionIntroduction

Very large evolution since early days Processing choice depends on machine

parameters No unique solution Wide range of signal processing:

Individual

Multiplexed (MPX)

Difference-over-Sum ()

Normalization (Phase & Time)


BIW

-00 M

ay 8

- 1

1

Signal analysisSignal analysis

Beam currentIb = [Qb*Nb]/trev

Qb = charge/bunch, Nb = number of bunches, trev = revolution period

Ib or Qb ? Ib measurements over several revolution cycles

Qb measurements on individual bunches

Induced signalV (t) = Zt*Ib (t) Zt is the PU’s transfer impedance

Bunch shape: (longitudinal charge density) Gaussian for leptons, (Cosine)2 for protons

Bunching factor: BF = (Bunching period)/(Bunch widthfwhm)


BIW

-00 M

ay 8

- 1

1

Beam structure (rf bucket filling)Beam structure (rf bucket filling)

Un-bunched beam Un-structure beam. No rf (Protons & Heavy Ions machines).

Very difficult to be treated. All

Beam bunches in all rf bucket. Optimized for maximum Ib

(SPS) The easiest to be treated. Almost monochromatic freq.spectrum

Few Beam bunches in few rf bucket with longitudinal symmetry.

The highest bunch density (LEP). Variable

Particular structure (no longitudinal symmetry); it includes single bunch filling and single passage (transfer lines)


BIW

-00 M

ay 8

- 1

1

Beam structuresBeam structures


BIW

-00 M

ay 8

- 1

1

Signal processing methodsSignal processing methods

The beam positionbeam position is uniquely related to the amplitude ratioamplitude ratio of the induced signals on opposite electrodes.

Processing methods for position calculation:

Difference over Sum () Analog and Digital process

Amplitude to phase/ time Passive analog process

Log-ratio (logA-logB) Active analog process

Transfer Function

-1

-0.5

0

0.5

1

-1 -0.5 0 0.5 1

Normalized

Position (U)

Computed

Position (U)

D/ SAtn(a/ b)loga-logb


BIW

-00 M

ay 8

- 1

1

Reference parameterReference parameter

Position = Kx,y* (A-B)/(A+B) = Kx,y*Np

Kx,y = scaling factor; A, B = induced signals; Np = Normalized

Position

Np is dimensionless and varies between 1U passing through 0 for a

centered beam. 1U is the “Normalized half aperture” Na

The “Normalized half apertureNormalized half aperture”” should be the reference parameter when specifying a processing system.

This will make possible comparisons among systems

r = Kx,y

A -1U B 1U

D

C

0


BIW

-00 M

ay 8

- 1

1

The ability to minimize the beam position errors Error sources:

mechanic, magnetic and electronics causes The offset for a centered beam should be minimized

Beam based alignment techniques Electronics error sources:

Impedance mismatching on interconnecting cables Electromagnetic interference and noise on the

input stage Non-linearity and beam intensity dependence Channels gain differences and calibration errors Digitizer granularity

Parameters: AccuracyParameters: Accuracy


BIW

-00 M

ay 8

- 1

1

Parameters: ResolutionParameters: Resolution

Important in colliding machines for luminosity Minimum position difference that can be resolved

Single shot: Stdev of individual measurements referred to the

normalized aperture Averaged:

as above but integrated over several revolutions Limiting factors:

At low level, it depends on the input noise and the BW For large signals, on the ADC resolution and the time

jitter State of art resolutions :

Single shot: < 0.02% of Na (few micron)

Averaged : < 20 ppm of Na (sub-micron)


BIW

-00 M

ay 8

- 1

1

Parameters: StabilityParameters: Stability

The measurement’s uncertainty will affect the global resolution of system.

The position measurements should be independent of the beam intensity, the bunch shaping and the rate. They should be stable vs. temperature and time, at least during the time interval between two calibrations

Stability versus input signalStdev from a series of digitized positions measured over the whole dynamic range.

Position temperature coefficientSlope of the position drift versus temperature

Long term position stabilityStdev of a series of digitized positions versus time


BIW

-00 M

ay 8

- 1

1

Parameters: Sensitivity & DynamicParameters: Sensitivity & Dynamic

Sensitivity: The minimum input level at which a beam position measurement

still fulfills the accuracy specifications (> 107 p/b) Dynamic

It determines the capability of the system to absorb very different beam intensities conditions

It’s defined as the difference, expressed in dB, between the maximum input level before a large non-linearity on the output signal appears (saturation) and the minimum input level at which a pre-defined signal to noise ratio (S/N) is reached

Processors using a discrimination level will not be limited by the S/N ratio, the lower limit being determined by the discriminator's threshold.


BIW

-00 M

ay 8

- 1

1

Parameters: Acquisition timeParameters: Acquisition time

The time required for the signal processor to store a full set of data into the memory

The importance of this parameter is related to the capability of resolving individual bunches and the absolute resolution of the processor

Several elements contribute to build-up this time: The LP and BP filters The switching and acquisition time (MPX processors) The PLL’s time to synchronize (synchronous detector) The AGC’s set-up time (constant sum) The S&H circuit and the ADC’s conversion time


BIW

-00 M

ay 8

- 1

1

Processing system familiesProcessing system families

Advantages Weakness

Large Bandwidth Limited dynamic

Long term stability No turn by turn

Center stability Gain switching

Amplitude No intensity independent information

Multiplexed

Individual

PassiveNormalization

ElectrodesA, B


BIW

-00 M

ay 8

- 1

1

No intensity information Reduced N°of digitizing

bit

Advantages Weakness

Long term stability Gain matching

Large dynamic Limited linearity

Large Bandwidth Time matching

Simplicity Phase matching


ConstantSum

Amplitude

to phase

Amplitude

to time

Logarithmic conversion

Normalizers


BIW

-00 M

ay 8

- 1

1


Definition Types

Individual bunches - Track & Hold separated by >10 ns - Log amplifiers to single - Amp to Time Normalizer

Turn-by-turn - Heterodyne

or individual bunches - Amplitude to separated by >100 ns Phase normalizer

Non consecutive - MPX turns measurements

WideBand

NarrowBand

SlowAcquisitio

n

Acquisition

Time


BIW

-00 M

ay 8

- 1

1


Legend: / Single channel

Wide Band

Narrow band

Normalizer

Processor

ActiveCircuitr

y

Heterodyne POS = (A-B) Synchronou

sDetection

AGCon

MPX

ElectrodesA, B

PassiveNormaliz.

POS = [log(A/B)] = [log(A)-log(B)]

DifferentialAmplifier

Logarithm. Amplifiers

IndividualTreatment

Limiter,t to Ampl.

Amp.to Time POS = [A/B]

POS = [ATN(A/B)] Amp.

to Phase

.Limiter, to Ampl.

POS = HeterodyneHybrid

HomodyneDetection

POS = or = (A-B)/(A+B)

Sample,Track,Integr. & Hold

Switch. gain

Amplifier


BIW

-00 M

ay 8

- 1

1

MPX schematicsMPX schematics

MPXPre-Ampl

Active

Matrix

A

B

C

D

X

Y

AGC

BPFilter

BPFilter

VCOPLLLimiter

BPFilterIF.

AmplMixer

VLSI

Freq.Synt.

Mixer

MPX

A

B

C

D


BIW

-00 M

ay 8

- 1

1

MPX descriptionMPX description Conceived for closed orbit of stable stored beams

The input signals are sequentially multiplexed into a single receiver

Multi-stage configuration of GaAs switches (Channels isolation >50 dB)

A BP filter selects the largest line of the spectrum

Pre-amplifier with AGC. Large input dynamic (>80 dB) and gain control (>50 dB) Noise Figure difficult to optimize.

Active mixer, driven by a frequency synthesizer, down convert to standard IF

IF amplifier with AGC and synchronous detection, by comparing the phase of a sample of carrier signal with a reference signal via a VCO in a phase lock loop (VLSI)

BP filter to suppress side-bands (100 kHz > BW< 1MHz).

De-multiplexer, Track & Hold and active matrix produce 7 signals (A, B, C, D, Sum, X, Y) store theirs values in four analog memories


BIW

-00 M

ay 8

- 1

1

Reduced number of channels (x4)

Identical gain for all the channels

No need for gain selection (AGC)

Large dynamic range (>80dB)

Excellent position stability

No temperature dependence and components aging.

Reduced N° of bits at equivalent resolution (Normalization)

MPX performancesMPX performances

Stable beam during the scanning

No turn by turn acquisition

Slow acquisition rate (MPX)

Reduced Noise Figure (front end matching & MPX insertion losses, AGC pre-ampli.)

Reduced linearity, for non-linear PU’s since the is not constant

Large engineering

No intensity information (AGC)

Advantages Limitations


BIW

-00 M

ay 8

- 1

1

LPFilter

LPFilter

A

B

0° 0°

180°0°

Pre-Ampl

Pre-Ampl

Sample& Hold

Sample& Hold

Wide band processing

Difference over Sum (Difference over Sum () )

BPFilter

BPFilter

A

B0°

0°0°

180°

Sample& Hold

Sample& Hold

Narrow band processingMixer

Mixer

PreAmpl

PreAmpl

Limiter


BIW

-00 M

ay 8

- 1

1

Difference over Sum (Difference over Sum () description) description

The and signals are obtained from a passive four port 180° hybrid

Wide-bandIt offers wide-band response (from kHz to GHz over >3 decades), very

large dynamic only limited by the electronics, isolation among ports (>30 dB)

Programmable gain amplifiers (Ga As switches) and track or peak & hold circuits

Wide bunches may be directly digitized by FADC (>1 GS/s)A & B signals can be treated separately by suppressing the 180° hybrid

Narrow-band BP filters are used to select the largest line in the spectrum. Programmable gain amplifiers and homodyne detector (a fraction of

signal is limited and used as local oscillator). Track & Hold and an externally triggered ADCs, digitize the

andsignals


BIW

-00 M

ay 8

- 1

1

Difference over Sum ( Difference over Sum ( ) ) performancesperformances

[W.B.] & {N.B.}

The central position independent on input intensity.

Intensity measurement available

Excellent Noise Figure

[Wide band allows measurements on multiple bunches (t <20 ns)]

{ Large dynamic > 90 dB}

Programmable gain amplifiers

Multiple calibration coefficients

The absolute position is f(gain)

{Tight phase matching(at all the gainsrequired by the synchronous detection (5°) }

{ Pedestal error on }

LimitationsAdvantages


BIW

-00 M

ay 8

- 1

1

LPFilter

LPFilter

A

B

LogampI to V

Converter

Diff.Ampli

Position = K * VoutLogamp

I to VConverter

Logarithmic amplifier schematicsLogarithmic amplifier schematics

Each signal is compressed by a logarithmic amplifier, filtered and applied to a differential amplifier.

The position response is Pos. [log(A/B)] = [log(A)-log(B)]

(Vout) where Vout is the voltage difference between the log-amplifiers outputs


BIW

-00 M

ay 8

- 1

1

10 dB

Full-waveDetector

Ampli-limiter

+Out-

+Log. Out -

+I n-

10 dB

Full-waveDetector

Ampli-limiter

10 dB

Full-waveDetector

Ampli-limiter

10 dB

Full-waveDetector

Ampli-limiter

10 dB

Full-waveDetector

Ampli-limiter-80 -60 -40 -20 0 dBm

.4

.8

1.2

1.6

2.0

0

1.0

-1.0

2.0

-2.0

dB

V

Logarithmic amplifier descriptionLogarithmic amplifier description

New generation circuits use several cascaded limiting amplifiers, with fix gain and wide bandwidth. Full wave rms detectors are applied among each stage and by summing theirs output signals, a good approximation to a logarithmic transfer function is obtained. Typical parameters are:

Input dynamic range : >90 dB Input noise: < 1.5 nV/Hz Non conformance lin.: < 0.3 dB Limiter Bandwidth: D.C. to >2 GHz Video Bandwidth: D.C. to 30 MHz


BIW

-00 M

ay 8

- 1

1

Logarithmic amplifiers performancesLogarithmic amplifiers performances

Possible applications in the time and frequency domain (NB & WB)

Very large dynamic range (>90 dB) without gain adjustment

Wide input bandwidth No bunch shape dependency Simultaneous digitization of

individual + and - charges Auto-triggering capability Simple engineering

State of art performances are not simultaneously available

Poor position stability vs.. input level, for particular conditions

Limited linearity ( few % of the normalized aperture)

Limited long term stability Temperature dependence


BI Review - Rhodri Jones (CERN - SL/BI)

28

INPUT OUTPUT

A A

BB

T1 = 1.5 ns

T1 = 1.5 ns

The Front-End Electronics


29

-2.5

-2.0

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

Time [ns]

Am

plitu

de A

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Am

plitu

de B

1.5ns

A

B

B + 1.5ns

A + (B + 1.5ns)A B

Beam

The Wide Band Time Normaliser


30

-2.5

-2.0

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

Time [ns]

Am

plitu

de A

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Am

plitu

de B

1.5ns

A

B

A + 1.5ns

B + (A + 1.5ns)

A + (B + 1.5ns)A B


t depends on position


31

-2.5

-2.0

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

Time [ns]

Am

plitu

de A

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Am

plitu

de B

A B

A+(B+1.5ns)

B+(A+1.5ns)+10ns

System output


Interval = 10 1.5ns


32


CAL. and TESTGENERATOR ADC

LOWPASS

FILTER

CALIBRATOR

PICK-UP

50 CABLE

Intensity

Measurement

Trigger

AutoTrigger

50 CABLE

LOWPASS

FILTERCALIBRATOR

NORMALISER INTEGRATOROPTICAL

LINK

TUNNEL SURFACE

DAB


33

-5%

-4%

-3%

-2%

-1%

0%

1%

2%

3%

4%

5%

1E+08 1E+09 1E+10 1E+11 1E+12

Number of Charges per Bunch

Per

cent

age

Err

or w

.r.t.

Hal

f Rad

ius

[%]

Linearity - High SensitivityLinearity - Low SensitivityNoise - High SensitivityNoise - Low sensitivity

Pilot Nominal Ultimate

WBTN - Linearity v IntensityFor LHC Arc BPMs 1% ~ 130m


34

WBTN - Linearity v Position

-8%

-6%

-4%

-2%

0%

2%

4%

6%

8%

-1 -0.5 0 0.5 1

Normalised Position

Per

cent

age

Err

or w

.r.t.

Hal

f Rad

ius

[%]

Calibration Point

Measured Value

Calibrated Value

Linearised Value

For LHC Arc BPMs 1% ~ 130m


35

Accuracy and ResolutionBunch Type Pilot Bunch Bunches of Nominal IntensityMode of Operation Trajectory

(singleshot)

Orbit(224 turnaverage)

Trajectory(single shot,

single bunch)

Trajectory(single shot,

average of allbunches)

Orbit(average of all

bunches over 224turns)

Resolution(rms) 200m 20m 50m 5m 5m

EL

EC

TR

ON

ICS

Accuracy(rms) 150m

AlignmentError (rms) 200m

ME

CH

AN

ICA

L

Residual afterk-modulation

(rms)<50m

3 = 750m20% of 4mmClosed Orbit ‘budget’(Spec = 500m)


BIW

-00 M

ay 8

- 1

1

Amplitude to Time Normalizer Amplitude to Time Normalizer performancesperformances

Reduced number of channels (x2)

No need for gain selection Input dynamic > 50 dB Signal dynamic independent on

the number of bunches ~10 dB compression of the

position dynamic (recombination)

Acquisition rate > 40 MS/s Auto-trigger Reduced N° of bits at

equivalent resolution (Normalization)

Mainly reserved to bunched beams

Tight time adjustment Propagation delay stability and

switching time uncertainty are the limiting performance factors

No Intensity information

Remark: A specifically designed monolithic Ga-As chip will allow for a large speed breakthrough



BIW

-00 M

ay 8

- 1

1

a

b

2nd analogfrequencytranslator

2nd analogfrequencytranslator

Evolution of radio systems: as processing and sampling technologies improve,digital moves from the baseband end (a) towards the pick-up sensor (b).

Pick-up or sensorSupplying rf signal

Analogdemodulatorand filter

Digitalfrequencytranslator

Digitaldemodulatorand filter

Digital signal

processor(DSP)

output

outputADC

ADC1st analog frequencytranslator

Analogfrequencytranslator

AnalogDigital

Digital receiver (basic)Digital receiver (basic)

Digital receiver is a new approach of the heterodyne receiver

The basic functionality is preserved but implemented differently

Present situation allows to place the digital transition just after the IF amplifier


BIW

-00 M

ay 8

- 1

1

Programmability

Narrow and wide band processing

Identical gain for all the channels due to possible permanent calibration

Resolution may be improved by over-sampling techniques

Excellent linearity (ADC)

Large dynamic range (AGC)

Reduced N° of bits at equivalent resolution (Normalization)

Digital Receiver performancesDigital Receiver performances

No single shot measurement

No “plug and play” system

Large engineering

All problems related to a new un-experienced processing system

The present advantages alone do not justify the man power investment, but I consider this technique as one of the most promising for the future



Conclusions for FP 420 BPMs

No obvious solution to satisfy all FP 420 BPM requirements, especially

– Required „normalized accuracy” in the order of 10 um / 100 mm, i.e. 10-4

– Bunch by bunch measurement

Already „bunch by bunch” resolution in the order of 10-4 is difficult to achieve

Propositions

– Measurement with a single multiplexed channel for a few bunches

– Something special, which employs specific features of the FP 420 BPM system

What can help to relax the difficult requirements?

– Required accuracy concerns relative distance beam – SI detector

– Required measurement range is much smaller than the vacuum chamber diameter


An idea of a PU with reduced „working aperture”

jaw 1 jaw 2

electrode 1signal

electrode 2signal

optical measurementof relative jaw position

Sidetector

A „collimator arrangement” to reduce the PU „working aperture”

If the jaw distance can be reduced to some 10 mm, then the „normalized accuracy” drops to 10-3, which is much more reasonable

One PU electrode on one unit with the Si detector, giving excellent relative positioning

Jaw relative position measured with optical means

Possible dynamic jaw positioning with respect to the beam

Some know-how could be quickly transferred from the collimator people, especially if they could think about using a similar idea for the collimator system


A version more robust for scattered particles

ima

ge

be

am

probe 1signal

probe 2signal

This version may be also interesting for collimators, for symmetric jaw positioning with respect to the beam

Signal quality not very demanding for simple signal equalization from both probes


Spare slides


Wall Current Monitor (WCM) principle

The BEAMBEAM current is accompanied by its IMAGEIMAGE A voltage proportional to the beam current develops on the RESISTORSRESISTORS in the beam pipe gap The gap must be closed by a box to avoid floating sections of the beam pipe The box is filled with the FERRITEFERRITE to force the image current to go over the resistors The ferrite works up to a given frequency and lower frequency components flow over the box wall


WCM as a Beam Position Monitor

For a centered BEAMBEAM the IMAGEIMAGE current is evenly distributed on the circumference The image current distribution on the circumference changes with the beam position Intensity signal () = resistor voltages summed Position dependent signal () = voltages from opposite resistors subtracted The signal is also proportional to the intensity, so the position is calculated according to / Low cut-offs depend on the gap resistance and box wall (for ) and the pipe wall (for ) inductances

LR

fL π2

L

RfL π2


A new design: Inductive Pick-Up (IPU)

An eight electrode “tight” design to avoid resonances in the GHz range

The electrodes cover 75 % of the circumference

The electrode internal diameter is only 9 mm larger then the vacuum chamber of 40 mm and it is occupied by the ceramic insertion (alumina)

The transformers are as small as possible to gain high frequency cut-off with many turns

The transformers are mounted on a PCB

The connection between the electrodes and the cover is made by screws

Electrode diameter step is occupied by the ceramic tube

The tube is titanium coated on the inside


Active Hybrid Circuit – Performance

The CMRR at 100 MHz is as high as 55 dB (datasheet 42 dB)

The CMRR for frequencies below 10 MHz is limited by the measurement setup

signal high cut-off frequency about 200 MHz

2 3 5 2 3 5 2 3 5 2 3 5

F re q u en cy [H z ]

-6 0

-4 0

-2 0

0N

orm

aliz

ed a

mpl

itud

e [d

B]

1 0 0 k 1 M 1 0 M 1 0 0 M

C M R R = c o m m o n s ig n a l / d if fe re n tia l s ig n a l

d if fe ren tia l m o d e s ig n a l

c o m m o n m o d e s ig n a l

C M R R = -5 5 d B @ 1 0 0 M H z


F req u en cy [H z ]

-1 0

-5

0

Nor

mal

ized

am

plit

ude

[dB

]

1 k 1 0 k 1 0 0 k 1 M 1 0 M 1 0 0 M1 0 0

s ig n a l

s ig n a l

IPU and AHC – Frequency Characteristics

A wire method with a 50 coaxial setup which the IPU is a part

signal – flat to 0.5 dB within 5 decades, almost 6 decades of 3 dB bandwidth (no compensation)

signal – 5 decades (four decades + one with an extra gain for low frequencies)

BW: 1 kHz – 150 MHz (> 5 decades)

BW: 300 Hz – 250 MHz ( 6 decades)


IPU and AHC – Displacement Characteristics

[mm] 05.078.9position vertical

[mm] 01.061.9positionhorizontal

V

H

-1 0 -8 -6 -4 -2 0 2 4 6 8 1 0W ire d isp lacem en t [m m ]

-1

-0 .8

-0 .6

-0 .4

-0 .2

0

0 .2

0 .4

0 .6

0 .8

1

Rat

io

/

D isp lacem en t m ax . = 2 0 m m

-6 -4 -2 0 2 4 6H o rizo n ta l (H ), v e rtic a l (V ) d isp lac em en t [m m ]

-0 .2

-0 .1

0

0 .1

0 .2L

inea

rity

err

or [

mm

]H e rro r = V e rro r =

D isp la c em en t m a x . = 2 0 m m

s ig n a l is c o n s ta n tw ith in re so lu tio n o f th e m e a su re m e n t o f 0 .1 %

F req u e n c y = 1 M H z

9 .6 1 / + 0 .0 1 [m m ]9 .7 8 / + 0 .0 5 [m m ]

x = H

x =V

x - w ire d isp lac e m e n t [m m ]H

x - w ire d isp lac e m e n t [m m ]V

H

V

A thin wire forming a coaxial line was displaced diagonally across the pick-up aperture. The measurement was done with a network analyzer: signal was applied to the wire and hybrid signals were observed.

fp 420 bpm workshop signal processing for bpms marek gasior cern-ab-bi

Documents