1 BPM Signal Processing with Diode Detectors Marek Gasior Beam Instrumentation Group CERN Outline Why using diode detectors for processing BPM signals Diode detectors in systems for which the signal amplitude is not that important two examples ID: 601669
Download Presentation The PPT/PDF document "8th DITANET Topical Workshop on Beam Pos..." 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
8th DITANET Topical Workshop on Beam Position Monitors, 16-18 January 2012, CERN
1
BPM Signal Processing with Diode Detectors Marek GasiorBeam Instrumentation Group, CERN
Outline:
Why using diode detectors for processing BPM signals ?
Diode detectors in systems for which the signal amplitude is not (that) important (two examples)
A compensated diode detector
Diode
ORbit
measurement system (DOR)
DOR lab measurements (resolution, stability, linearity, calibration)
DOR measurements with the collimator BPM and the SPS beam
DOR measurements with the LHC beamSlide2
BPM Signal Processing with Diode Detectors
2
Diode detectors can be used to convert fast beam pulses from a BPM into slowly varying signals, much easier to digitise with high resolution. In this way amplitudes of ns pulses can be measured with a lab voltmeter.As the diode forward voltage Vd depends on the diode current and temperature, the output voltage of a simple diode detector also depends on these factors.The detector output voltage can be proportional to the peak amplitude or an amplitude average of the input pulses.
Why diode detectors ?
Input (
V
i
) and output (
V
o) voltagesof a peak detector withan ideal diode
Input (
V
i
) and output (Vo) voltagesof a peak detector witha real diode
Input (
V
i
) and output (
V
o
) voltages
of an average-value detectorSlide3
BPM Signal Processing with Diode Detectors
3
Charge balance equation for the following assumptions:a simple diode model with a constant forward voltage Vd and a constant series resistance rconstant charging and discharging current, i.e. output voltage changes are small w.r.t. the input voltage
Simple diode detectors
A numerical example: LHC, one bunch.
For LHC
τ
≈
1
ns and
T
≈
89
μs
, so for
V
o
≈
Vi one requires R/r > T/τ. Therefore, for r ≈ 100 Ω, R > 8.9 MΩ.
For large
T to τ rations peak detectors require large R values and a high input impedance amplifier, typically a JFET-input operational amplifier.The slowest capacitor discharge is limited by the reverse leakage current of the diode (in the order of 100 nA for RF Schottky diodes)
V
i
V
d
n
bunchesSlide4
BPM Signal Processing with Diode Detectors
4
One diode detector for each BPM electrode.
Subtracting signals before the detectors (e.g. by a 180° hybrid) is no good, as the resulting signals would be:
smaller (→ larger nonlinearities);
changing signs when crossing the BPM centre.
The diode forward volt
age
Vd introduces a significant position error.V
d depends on the diode current and temperature.Simple diode detectors are good for applications when the signal amplitude is not that important.Two examples:
Tune measurement systemsAn LHC safety system: Beam Presence Flag
Simple diode detectors for BPM signalsSlide5
BPM Signal Processing with Diode Detectors
5
In tune measurement systems one measures frequency of beam oscillations.For proton machines often such a system must be optimised for sensitivity, e.g. for the LHC the system has to see μm turn by turn beam oscillations (emittance conservation, tolerances of the collimation system)Often the BPM signal amplitude is much too large to be processed by the front-end electronics, so it must be reduced (i.e. signal thrown away). For example, the LHC tune stripline pick-ups give up to some 200 V peak voltages.Diode detectors can work directly with such large voltages (one diode is good for some 70 V of signal).The large DC voltages corresponding to the beam offset are removed by series capacitors. Otherwise these voltages would waste the dynamic range of the following acquisition chain.
The resulting small signals corresponding to beam oscillations are amplified and filtered,
p
rior to be digitised with high resolution audio ADCs.
Diode detectors in tune measurement systemsSlide6
BPM Signal Processing with Diode Detectors
6
BPLXD6R4.B1BPLXB6R4.B2
LHC tune measurement system based on diode detectorsSlide7
BPM Signal Processing with Diode Detectors
7
FFT1 continuous system at βH βV 400 m, B2, 1 bunch of 5×109, fill #909, 13:35 11/12/09,
8K
FFT @ 1 Hz, 20 FFT average
LHC beam spectrum, one pilot bunch
Absolute amplitude calibration was done by exciting the beam to the amplitudes seen by regular LHC BPMs.Slide8
BPM Signal Processing with Diode Detectors
8
Beam spectrum observation on an electron machine (PSI)
Work done in the framework of feasibility studies for stabilisation of the magnets in the CLIC final focusing system.
Very many thanks to
our PSI colleagues
for their
great hospitality.
The measurement was done with the LHC tune system harware, only sightly modified:detectors for electron beams (diodes upside-down w.r.t. protons !)
bandwidth of the analoque front-end shifted from 0.5 – 5 kHz to 1 – 1000 Hz.Slide9
BPM Signal Processing with Diode Detectors
9
SLS beam spectrum measured with the diode detector system
Absolute amplitude calibration was done by exciting the beam to the amplitudes seen by regular SLS BPMs.Slide10
BPM Signal Processing with Diode Detectors
10
LHC Beam Presence Flag system based on diode detectors
BPF is an LHC safety system, optimised for reliability and robustness.
Very simple, the only inputs are beam signals and 230 V (in particular no computer interface, no timing signals).
In addition no
programmable devices, no adjustable components, no
ADCs.
4 independent system channels for each beam use signals from 4 BPM electrodes, resulting in
quadruple redundancy for the final “high beam intensity” enable.Slide11
BPM Signal Processing with Diode Detectors
11
Compensated diode detector
Compensated diode detector consists of two diode peak detectors, one with single, second – with two diodes.
All three diodes are in one package, for good thermal coupling and symmetry of the forward voltages
V
d
.
Two operational amplifiers are used to derive 2 Vd voltage and to add it to the output of the two-diode detector. This way the resulting output voltage is equal to the input peak voltage.This is the simpler and most promising scheme, found in a very popular text book on electronics.To get an “ultimate peak mode operation”, the discharge resistors can be omitted. In this case the discharge is done by the reverse leakage current of the diodes.
The asymmetry in the charging conditions becomes less important for larger input voltages.Slide12
BPM Signal Processing with Diode Detectors
12
Signal of each pick-up electrode is processed separately.The multiplexer is foreseen for calibration with beam signals.The low pass filters decrease the signal amplitude.The conversion of the fast beam pulses into slowly varying signals is done by compensated diode detectors.These slow signals can be digitised with high resolution, averaged and transmitted at slow rates.All further processing and calculations are done in the digital domain.
Simple and robust hardware, high resolution, no beam synchronous timing required
Diode
ORbit
Front-EndSlide13
BPM Signal Processing with Diode Detectors
13
Diode
ORbit
front-end connected to two LHC BPMsSlide14
BPM Signal Processing with Diode Detectors
14
DOR FE prototypes are still in the “state of the art”HAND MADE
in Rapid Development Technology
®Slide15
BPM Signal Processing with Diode Detectors
15
Inputs of all 4 FE channels parallel, simulating beam in the PU centre.Input signal: 10 MHz sine wave, FE gain 25 dB (max).Raw results, without any offset and gain calibration.24-bit 8-channel ADC sampling at 11.7 kHz, samples averaged in the microcontroller to 50 Hz equivalent sampling, then to 1 Hz for the plots.
Front-end channels have 10 Hz LP filters before the ADCs.
Amplitude changes due to temperature sensitivity of the signal generator.
The amplitude faster jumps may be due to some internal calibration of the generator or somebody (i.e. me) touching the input cables, as during the night signals were much smother.
DOR prototype: lab measurementsSlide16
BPM Signal Processing with Diode Detectors
16
Beam position calculated as for an LHC arc pick-upwith 49 mm electrode distance.”Natural” long term stability shown, no calibration.
DOR prototype: lab measurements
The plot repeated from the previous slideSlide17
BPM Signal Processing with Diode Detectors
17
All 4 inputs in parallel, simulating beam in the PU centre.Input: 10 MHZ sine wave with slow triangular modulation to simulate intensity changes, FE gain 25 dB (max).Raw results, without any offset and gain calibration.Beam position calculated as for an LHC arc pick-up with 49 mm electrode distance.ADC sampling at 11.7 kHz, samples averaged to 50 Hz equivalent sampling.Front-end channels have 10 Hz LP filters before the ADCs.
DOR prototype: lab measurementsSlide18
BPM Signal Processing with Diode Detectors
18
Now channels 2, 3, 4 are correlated with channel 1, resulting in the coefficients (offset, gain), in the ADC full scale (FS) units:ch2: -0.001720, 1.002262ch3: 0.000325, 0.998747ch4: -0.001247, 1.000359Channel difference improved from 103 to 10
5
level, by some 2 orders of magnitude.
P
osition error improved also by some 2 orders of magnitude.
For amplitudes larger than 20 % of the full scale the noise is not larger than some 50
nm peak-peak, i.e. some 10 nm rms. This is with the 25 dB gain of the high-frequency input amplifiers, 50 Hz ADC equivalent sampling and 10 Hz analogue bandwidth for the orbit changes.The final diode front-end will be equipped with a calibration circuitry capable of connecting the same signal of programmable amplitude to all inputs.
The input circuitry will allow the same with beam signals, i.e. the same beam signal can be connected to both front-end channels processing one pick-up plane.
DOR prototype: lab measurementsSlide19
BPM Signal Processing with Diode Detectors
19
Input signal made different with fixed attenuators, simulating an offset beam.Simulated beam offset is the same for both pick-up planes (both planes connected in parallel).Input signals, gain and sampling as before for the simulated centred beam.
Errors some 2 orders of magnitude larger than for the centred beam.
Correlating channels does not help significantly.
Position measurement dependency on signal intensity for large beam offsets will be addressed in the future development.
100 µm
DOR prototype: lab measurementsSlide20
BPM Signal Processing with Diode Detectors
20
DOR prototype: lab measurements with a calibration muxSlide21
BPM Signal Processing with Diode Detectors
21
wholecorrelation for 60 scorr. for the whole meas.corr. for 1 s every 1 h
DOR prototype: lab measurements with a calibration mux
as above, projected to a 49 mm apertureSlide22
BPM Signal Processing with Diode Detectors
22
DOR prototype: lab measurements with a calibration mux
Measurement with 4 equal signals and calibration mux (not switching), results scaled to 49 mm aperture.
Correlation coefficients used to calibrate the channels, the same coefficients for the whole measurement.Slide23
BPM Signal Processing with Diode Detectors
23
right upstream buttonleft upstream buttonright downstream buttonleft downstream button
RU
LD
LU
RD
Drawing from A.Nosych
DOR on collimator BPM: Measurements with the SPS beam
Main role of the embedded BPM system is to indicate when the beam is in the middle between the jaws, i.e. when the signals from the opposing electrodes are equal.
Geometrical factor, a constant for “static” BPMs, changes with the jaw distance, fortunately not dramatically, less than a factor of 2.
Signal amplitude changes with the jaw distance by a factor of 3.
Very interesting case from academic point of view (lab-type measurement possible with real beam, online calibration by BPM precise movement).
Measurements with one LHC nominal bunch Slide24
BPM Signal Processing with Diode Detectors
24
DOR on collimator BPM: beam tests on the SPSSlide25
BPM Signal Processing with Diode Detectors
25
LHC: standard vs. DOR processing comparison
BPM electrode signals are split and send to both systems
Beam position calculation using the same polynomial
Aperture 61 mmSlide26
BPM Signal Processing with Diode Detectors
26
LHC: standard vs. DOR processing comparisonSlide27
BPM Signal Processing with Diode Detectors
27
LHC: standard vs. DOR processing comparisonSlide28
BPM Signal Processing with Diode Detectors
28
Diode detector strengths:Simple and robust.“Perfect sampling” even for very short bunches without any precise timing.Slowly varying signals at the output, easy for signal processing and high resolution digitization.Low rates of the digital data.Natural noise gating (i.e. very low gain for noise between beam pulses).No limit for beam offsets, as signals from each electrode are processed separately.Excellent resolution.
Surprisingly good long term stability.
Limited dependence of the output voltages on the number of circulating bunches.
Conclusions
Diode detector weaknesses:
Limited dynamic range of the linear operation, variable gain amplifiers required to compensate important bunch intensity variations.
Only turn by turn, bunch by bunch not easy
(some ”reset techniques” required).
Large bunches favoured
(good for masking undesirable reflections).
Important dependence of the measured position on signal intensity for large beam offsets.
It will be addressed in the future development.