Design, Assembly, Optimization and Commissioning of a Rotating Coil,
Multipole Magnet Measurement System for the
Thomas Jefferson National
Accelerator Facility
By
Kenneth S. Baggett Jr.
Thesis submitted to the Graduate Faculty of
Christopher Newport University in partial
fulfillment of the requirements
for the degree of
Master of Science
2005
Approved:
Edward Brash, Chair
_______________________________
David Doughty
_______________________________
John G. Hardie
_______________________________
Copyright by Kenneth S. Baggett 2006
All Rights Reserved
ABSTRACT
Title of Thesis:
Design, Assembly, Optimization and Commissioning of a
Rotating Coil, Multipole Magnet Measurement System for the
Thomas Jefferson National Accelerator Facility
Degree Candidate:
Kenneth S. Baggett Jr.
Degree and Year:
Master of Science, 2006
Thesis directed by:
Edward Brash, PhD, Associate Professor of Physics,
Department of Physics, Computer Science, and Engineering
A rotating coil, multipole magnet measurement system has been built and
commissioned to carry out high precision magnetic field measurements at Thomas
Jefferson National Accelerator Facility (JLab). The facility’s existing magnetic field
measurement system was in need of replacement due to its lack of replacement parts.
The new system was designed to have a repeatability of 1 part in 10,000. New
magnets currently being constructed at JLab exceed the capabilities of the existing
multipole measurement system and require this higher level of precision.
The system was constructed of several data acquisition components which were
purchased or constructed in accordance with the measurement precision requirements
noted above. Software created using Lab Windows/CVI synchronizes the hardware
and automates the data collection process.
The components used to create the
measurement system, the results of the analysis, the procedures used to acquire the
data, and the sources of measurement error will be discussed in detail.
TABLE OF CONTENTS
1 INTRODUCTION
1
1.1
Harmonic Coil Magnet Measurement
2
1.2
Measurement System Components
3
1.3
Magnet Powering
3
1.4
Problem Statement
4
2 RELATED LITERATURE
5
2.1
Rotating Coil Measurements
5
2.2
Magnet Degaussing
8
2.3
Hysteresis, Ramp Rate, and Overshoot
9
2.4
Harmonic Measurement Probes
11
3 MEASUREMENT SYSTEM DESIGN
13
3.1 Signal
3.1.1
Digital Integration
3.1.2
Signal Multiplexing
3.1.3
Mulitplexer Chassis Box
3.1.4
Wire and Cabling
13
13
14
15
17
3.2 Motion
3.2.1
Power Drive
3.2.2
Stepper Motor Control
3.2.3
Motor and Encoder
3.2.4
Triggering
18
19
19
20
20
3.3 Data Transfer and Analysis
3.3.1
Hardware Communication
3.3.2
Host Computer Requirements
21
21
21
3.4 Magnet Powering
3.4.1
Epics
3.4.2
Three Piece Ramp
3.4.3
Jlab Control Software.
22
22
23
25
3.5 Supplementary Components
3.5.1
Signal Processing
31
31
4 MEASUREMENT SYSTEM ASSEMBLY
ii
33
4.1
Magnet Powering Instrumentation Control
33
4.2
Bucked Signal Configuration
35
4.3
The Measurement System
36
5 OPTIMIZATION
38
5.1
Cycle Analysis
38
5.2
Cycle Options: Forward versus Forward-Reverse Averaging
41
5.3
Cycle Averaging Optimization
43
5.4
Improving Statistics
44
6 COMMISSIONING
46
6.1
Reference Signal Repeatability
47
6.2
Function Generator Reference Signal Setup
48
6.3
Continuous Rotation Testing with Reference Signal
50
6.3.1
Five Cycle Averaged Rotation Testing with Reference Signal
53
6.4
Signal Analysis Algorithm Comparison
57
6.5
Probe Reproducibility Testing
61
6.6
Measurement Uniformity Across Gain Settings
62
6.7
Individual Position Uniformity
65
6.8
Signal Loss – Probe Input to PDI Input
68
6.9
Short and Long Term Repeatability
69
6.10 Optimized Repeatability
72
6.11 Absolute Strength
76
7 CONCLUSIONS
79
APPENDIX A METROLAB PDI 5025
81
APPENDIX B FFT ALGORITHM COMPARISONS
84
iii
APPENDIX C QUADRUPOLE AMPLITUDE DATA – COILS 1,2,4
86
APPENDIX D EXPERT CONTROL PANELS FOR THE PDI
SOFTWARE
92
APPENDIX E MEASUREMENT SYSTEM ASSEMBLY DETAILS
98
REFERENCES
110
iv
List of Figures
Figure 2-I: Cross section of a rotating coil moving through a quadrupole magnet 5
Figure 3-I Top view of the multiplexer chassis box interior.
16
Figure 3-II: Stages of the three piece ramp protocol used to set magnet current.
24
Figure 3-III: PDI software start screen. .
27
Figure 3-IV Parameter file creator software panel.
28
Figure 3-V Expert parameter file software panel.
29
Figure 3-VI PDI software runtime panel.
30
Figure 3-VII PDI software runtime plots.
31
Figure 4-I Bucked signal wiring configuration.
35
Figure 4-II Measurement system rack components
36
Figure 4-III Multipole measurement stand
37
Figure 5-I Simulated frequency convergence – averaged.
38
Figure 5-II Simulated frequency convergence - continuous cycles.
39
Figure 5-III Software function used to calculate the magnet’s harmonic content. 40
Figure 5-IV Forward only vs. Forward Reverse measurement repeatability.
42
Figure 5-V Changes in standard deviation as measurements Added.
44
Figure 6-I Main harmonic amplitude reproducibility for each coil using 5
sequential cycles.
53
Figure 6-II Main harmonic amplitude reproducibility using the average of 5
cycles.
56
Figure 6-III Measurement repeatability across gain settings 1 – 20.
63
v
Figure 6-IV Measurement repeatability across gain settings 50 -1000.
64
Figure 6-V uV*s deviations from average of all runs every 1.8 degrees.
67
Figure 6-VI Long term repeatability using generated signal
70
Figure 6-VII Long term system repeatability using temperature controlled
generated signal (8 hours).
71
Figure 6-VIII Initial versus Optimized Repeatability for Probe P2B.
73
Figure 6-IX Short term optimized repeatability of system using rotating coil
(P2B).
74
Figure A-I Schematic of MetroLab’s PDI 5025 Integrator Circuit
83
Figure B-I CAMAC FFT Algorithm
84
Figure D-I PDI Expert Control Panel
93
Figure D-II Three Piece Ramp Expert Panel
94
Figure D-III Motion Control Expert Panel
95
Figure D-IV Epics Control
96
Figure D-V HP-34970A DAQ Switch Unit
97
Figure E-I PDI 5025 Microswitch Configuration for GPIB
98
Figure E-II Signal Chassis Box Wiring Breakout Configuration
102
Figure E-III IP MAX axis configuration for motor controller and encoder
104
Figure E-IV MAX instrument initialization
105
Figure E-VI IP Configuration for GPIB ENET/100
109
Figure E-VII GPIB ENET/100 Setup
109
vi
List of Tables
Table 2-I: Changes in magnetic field resulting from current overshoot.
10
Table 5-I A proposed method to improve the standard deviation of measurement
samples.
45
Table 6-I Initial system repeatability using a simulated signal - Direct connection
/ continuous revolutions.
50
Table 6-II Initial system repeatability using a simulated signal – Coil 1 location /
continuous revolutions.
52
Table 6-III Initial system repeatability using a simulated signal – Coil 1 location /
five averaged cycles.
54
Table 6-IV: Phase Comparison between CAMAC and PDI Algorithms
59
Table 6-V: Amplitude Comparison between CAMAC and PDI Algorithms
60
Table 6-VI Initial probe reproducibility at 5 A as a percentage of signal strength.
61
Table 6-VII Voltage signal loss from probe input to PDI input.
68
Table C-I Measurement amplitude repeatability for probe P2A-Coil 1
86
Table C-II Measurement amplitude repeatability for probe P2A-Coil2
87
Table C-III Measurement amplitude repeatability for probe P1A-Coil 1
88
Table C-IV Measurement amplitude repeatability for probe P1A-Coil 2
89
Table C-V Measurement amplitude repeatability for probe P2B-Coil 4
90
Table C-VI Measurement amplitude repeatability for probe P1C-Coil 4
91
Table E-I GPIB instrumentation channels
vii
107
viii
1
Introduction
Higher demands on the measurement resolution and repeatability of multipole
magnets at JLab are driving a need to push the limits of magnet measurement
systems. For the JLab accelerator and Free Electron Laser (FEL) to operate
optimally, measurement of absolute magnet strength to a precision of 1 part per
1000 Gauss and measurement reproducibility of 1 part per 10,000 Gauss is
essential. An upgraded rotating coil measurement system has been developed to
achieve these specifications. In this system, integrated flux measurements are
carried out using an integrator and subsequently analyzed to determine the
magnetic field strength, the magnetic center, the skew components, and the higher
order harmonics that may degrade magnet performance.
Uniform magnet powering is also essential for magnetic field repeatability. The
specification for new magnets being built at JLab, such as the QX quadrupole
magnets in the FEL, is to attain the 1 part per 10,000 gauss reproducibility.
Deficiencies in magnet powering uniformity may adversely affect the
reproducibility of the absolute field.
The FEL QX quadrupole magnets are
resistive and when combined with longer lead lengths can result in approximately
3.2 Ohms or more of resistance in certain instances. Existing bulk power supplies
operate at a nominal 28 Volts thus causing a problem with standard hysteresis
cycling between 9.9 and -9.9 Amps. In some cases the hysteresis limits are not
reached, resulting in differences in the absolute magnetic strength at specific
currents. This problem renders the current versus field maps useless for setting
1
magnets to a precise field value and could prevent the success of the FEL UV
beam-line upgrade. To match actual installation conditions, a magnet power
supply typical of those used in the FEL is now used to power magnets in the
magnet test facility. Also, a 200’ lead length can be used between the magnet and
power supply to mimic conditions in the FEL.
1.1
Harmonic Coil Magnet Measurement
A classic approach used to measure quadrupole magnets involves using a rotating
coil.
A single coil rotational harmonic measurement probe may be used to
establish the strength and magnetic center of the magnet. In this method, the coil
is rotated in the aperture of the magnet. As the coil rotates, a digital integrator
records the induced voltages and angular intervals. At the completion of a cycle,
the integrator can then transfer data to the host computer for processing. A Fast
Fourier Transform (FFT) analysis of the data can be completed on the host
computer to determine the harmonic content. To establish a higher level of
sensitivity to higher harmonics, a bucked coil probe configuration may used. In
this configuration a probe contains two coils such that all sides of both coils are
off axis and the bucked signal is produced by subtracting the voltage integral of
the inside coil from that of the outside coil. By properly selecting the radii and
turn counts of the two coils, the dipole and quadrupole signals will be bucked out.
2
1.2
Measurement System Components
Jefferson Lab selected the MetroLab PDI 5025 digital integrator as the core of the
new measurement system. The PDI 5025 is a digital integrator that utilizes
counters, buffers, and data registers to manage input data independent of external
computers. This independent hardware system removes the need for a real time
operating system. While the operation of the MetroLab Integrator is somewhat
slower than other existing measurement techniques, its signal to noise ratio of a
few parts in 108 makes it ideal for signal processing.
At low fields, noise
becomes the limiting factor for measurement accuracy and therefore the origin of
the sources of noise must be established and their effects minimized. Each
rotating coil probe must be rigid and accurately designed for the system to
produce accurate and precise results. Other instrumentation design considerations
included the availability of a stepper motor control together with the MetroLab
Integrator encoder synchronization, optical triggering, signal multiplexing, and
power supply control automation.
1.3
Magnet Powering
A fundamental part of the measurement process is the power supply ramping
protocol. To properly analyze system repeatability, ramp and hysteresis protocols
have been properly designed to return each magnet to its proper strength versus
current on the magnet’s hysteresis curve. Linear ramping, three part ramping, and
bang-bang ramping methods are used to provide both accurate and time effective
3
results. Magnet powering for the rotating coil stand is discussed in the design
chapter.
1.4
Problem Statement
A new rotating coil multipole magnet measurement data acquisition system is
required to meet the specifications of the Thomas Jefferson National Accelerator
Facility’s FEL and 12GeV upgrade. Furthering the need for the project is the fact
that the existing system has been in operation beyond its life expectancy and has
become problematic and undependable. Constraints on absolute magnet strength,
magnet reproducibility, and magnet set ability make the need for measurement
system optimization imperative. Determining absolute magnet strength to 1 part
per 1,000 at full field and reproducibility to 1 part in 10,000 is now considered
crucial to successfully guide both the accelerator’s and FEL’s beams at the higher
energies through the machine’s geometric design. To meet these demands the
rotating coil measurement system will utilize a high-resolution digital integrator
to record sensitive induced voltage signals from the coil. System noise level will
be evaluated and minimized to improve the resolution of the measurements.
4
2
Related Literature
2.1
Rotating Coil Measurements
Rotating coils are commonly used to map multipole magnets used in particle
accelerators, where precise field integral measurements are required [3]. In this
method, a coil is wound onto and then glued to a core made of a mechanically
stable material such as glass, kevlar, or graphite composites. The wire is stretched
during winding to assure a well-defined geometry and coil stability. Rotating
cylinders are often used as the frame for the wire. During the measurement, the
coil is rotated and the resulting induced signal is transmitted to either a frequency
analyzer or integrator for harmonic analysis.
The method of rotating coil
measurement has been refined since its inception in 1954 [1-2].
The rotating coil consists of a number of turns of thin wire mounted on a rigid
frame that can be rotated inside the bore of a magnet, as illustrated in Figure 2-1.
Figure 2-I: Cross section of a rotating coil moving through a quadrupole
magnet
5
The rotating coil is positioned within the bore of the magnet such that it extends
beyond the magnet length at each end. As the coil rotates about the longitudinal
magnet axis, the magnetic flux through the coil changes, which causes a voltage
to be induced in the coil. According to Faraday’s law, this induced voltage, Vind,
is proportional to the rate of change of the magnetic flux:
Vind   N
where the magnetic flux,  =
d
,
dt
(2.1)

  
B  A , B is the magnetic field, A is the surface
area of the coil, and N is the number of turns of wire on the coil.
The magnetic flux through a rotating coil probe is given by the normal component
of


B integrated over the area, A , of the coil, according to
 
   B  dA
(2.2)
A
Using a Fourier analysis of the flux distribution, the harmonic content can be
determined, thus providing an accurate depiction of the magnetic field according
to:
 r
Br (r ,  )  B0  

n 1  rref





6
n 1
(bn cos n  an sin n ) ,
(2.3)
where Bo is the amplitude of the main field harmonic, rref is any reference radius,
bn and an are the harmonic coefficients of the normal and skew components
respectively, and  is the angle of the rotating coil (Figure 2-I). In this notation,
b0 describes the normal dipole coefficient, b1 the normal quadrupole coefficient,
etc. The corresponding skew field components are described by the coefficients
a1, a2, etc. [2-3].
An advantage to using an integrator instead of a simple voltage measurement is
that the amplitude of the induced signal is independent of the rotational speed.
For an integrator, the integrated voltage is given by:
t
 V (t )dt ( )  ( ) ,
0
(2.4)
0
where 0 is the angular reference position at t = 0, the point at which the
integrator is reset, and  is the measured angle relative to this reference angle[3].
The coil is rotated forward about the longitudinal axis through the zero
reference point. At this point the measurement is triggered initially. The
rotation continues and the measurement proceeds through 360 degrees
where the integration ends. At regular angular intervals the integrator is
triggered electronically to collect data. Once the forward measurement is
7
completed, the coil is then rotated backwards so that data can be collected
from the two different rotational directions and averaged. At the end of
the measurement cycle, the integrator outputs the integrated voltage at
each encoder location in V·s (Webers) according to:
 Vdt  L
eff
B( )  Radius
(2.5)
where L is the effective length of the field, R is the reference radius, B is the flux
density, and  is the reference angle [8].
A Fast Fourier Transform (FFT) is
then performed on the data to extract the amplitude and phase of the signal. It
should be noted that the efficiency of calculating the FFT of the data may be
increased by selecting a binary number of measurement points [3]. Subsequent
analysis of a complete rotating coil measurement leads to the extraction of
characteristic magnet parameters including integrated field gradient, effective
magnetic length, harmonic contents, magnet hysteresis effects, and field
repeatability [2].
2.2
Magnet Degaussing
Magnet powering, and occasional overpowering, leads to observable residual
magnetic fields in electromagnets. To restore the magnet to a “degaussed” state,
one where the residual magnetic field is as close to zero as possible, a controlled
degaussing method must be used. The degaussing routine is typically used at
JLab is to cycle the magnet between multiple positive and negative current values.
8
The first value is typically either the upper limit of the measurement range or a
point at which the magnet is overpowered. The routine starts at the positive value
and then ramps to the negative value. At this point the value is decremented by 1
Amp and repeated. This continues until the value reaches zero. At this point the
magnet is considered degaussed.
Other laboratories have proven other degaussing methods to be equally effective.
One procedure involves ramping the magnet current up to the upper limit and
maintaining it for ten minutes. At this point, the power supply is abruptly turned
off. The magnet is then allowed to remain un-powered for approximately five
minutes until the residual magnetic field is seen to become stable. Once the stable
field is restored the power supply leads are reversed and the current is ramped to
three-tenths of the current that was used as the upper current limit. This current is
maintained for three minutes and the power supply is again cut off. After ten
minutes, the residual field of the magnet is approximately zero [4].
2.3
Hysteresis, Ramp Rate, and Overshoot
It has been shown that homogeneity of the magnetic field in electromagnets is
somewhat dependant on current ramp rates, but the effects are minimal. Harris
and Cobb [4] conducted a series of magnetic measurements at various ramp rates
and concluded that the use of a slower ramp rate resulted in better field uniformity
and that the slower ramp speeds resulted in a lower field. However, because these
9
differences were only at the level of 1 to 2 parts in 105 at transverse positions of -2
and +2 inches, ramp rate was determined to be inconsequential for most practical
purposes.
Greater effects have been traced to power supply overshoot. In this case the
magnet field as well as the magnet current was monitored to determine correlated
effects attributable to power overshoot. The test involved degaussing the magnet
and then setting the current to a desired value at a set ramp rate. The next step
was to slowly cycle the current between an overshoot amount and the previous set
current value. After this hand dialed overshoot, the magnetic field was measured.
Table 2-1 shows overshoot effects on magnetic field values at multiple currents
[4].
Initial
Set
Current Value
(I1)
Initial
Field
Value
(Gauss)
% Increase
in Current
% Increase
in Field
% Change in Field
When
Current
Returned to I1
301.59
4998.4
0.24
0.240
0.010
599.95
9939.3
0.24
0.235
0.010
920.02
14593
0.23
0.160
0.013
920.05
14598
0.62
0.400
0.040
899.97
14373
2.20
1.500
0.090
Table 2-I: Changes in magnetic field resulting from current overshoot.
The table shows the relationship between the current and field strength along with
the effects of overshoot. As the current is slightly increased, the magnetic field
also increases. When the current is returned to its initial value there are small
differences in the field strength of the magnet due to hysteresis effects. Even
though these differences seem small, they could have a significant effect as the
beam propagates through an accelerator. Also, this table represents larger dipole
10
magnets operating at much higher fields than the quadrupoles magnets that will be
measured by the JLab measurement system. Because of this, it will be important
to quantify the overshot effects on quadrupole magnets at lower fields to see if
these percentages hold true.
2.4
Harmonic Measurement Probes
There are five distinct harmonic coil probes that are available for use in the
rotating coil measurement system at JLab. Each coil is fabricated for use with
magnets of varying lengths and bore sizing. Initial testing was done using the socalled P1A measurement probe. This probe has a radius of 1.120” and has two
coils which can be bucked, or canceled, against one another, to suppress the main
field component. In this way the resulting signal can then be used to evaluate the
field quality with increased sensitivity.
Accuracy in absolute magnetic field strength measurements cannot be achieved
without precise physical magnet alignment of the probe and magnet. In magnets
where there exists a horizontal field gradient, error in measured field strength will
result from any error in the horizontal probe position. For example, in a magnet
with 10% horizontal field gradient, measurement of the field strength to 1 part in
104 requires positioning the probe with 0.1 mm accuracy. Such precise alignment
is possible through the use of mechanical fixtures referenced to precisionmachined surfaces on the magnet [5].
11
Another issue effecting measurement accuracy is measurement system
construction.
The typical manufacturing and measurement uncertainty for
magnetic measurement apparatus is ~0.001” (25 μm) which translates into a field
error of 3 parts in 103 for a probe of 10 mm maximum radius. To meet the goals
of the new magnet measurement system, 1 part in 103 is required. This level of
precision can only be achieved using a calibration magnet.
In this case the
calibration (or standard) magnet absolute strength must be known as well or better
than the required precision of the subject magnet measurement [5]. Currently,
JLab does not have a standard magnet for absolute calibration.
There is a
possibility that an electromagnet could be precisely measured at an external
facility and then used as the standard magnet for the JLab measurement system.
Ideally, a permanent magnet will be purchased to be used as the standard magnet.
12
3
Measurement System Design
The PDI data aquisition system has been constructed from a collection of
components selected to attain the desired measurement precision. Components
needed for the measurement system were separated into two groups, signal and
motion. In the signal group, the components were selected to transfer and record
low level voltages to meet the reproducibility requirements of the system. The
motion group components were selected based on power and incremental step
size resolution.
Other peripheral components were selected to transfer data,
verify voltage settings, and synchronize instruments.
3.1
Signal
The measurement system is responsible for precisely integrating voltage signals
under 5V. Induced voltages of less than 1V at full field will be expected for many
magnets that will be measured routinely by the system.
Components were
selected and purchased to accurately measure magnetic fields over these ranges.
The resolutions of the components were evaluated to verify that the desired
measurement precision could be met.
3.1.1
Digital Integration
At the center of the measurement system is the MetroLab PDI-5025 Precision
Digital Integrator (PDI). The PDI, which was developed specifically to perform
magnet measurements, integrates the induced voltage received from a rotating or
translating coil relative to increments received from the motor encoders or timer
events. To MetroLab PDI was selected because of its high signal to noise ratio
13
and successful use at other magnet measurement facilities. Output buffering,
which allows up to 5200 results to be accumulated during a measurement
sequence, exceed the requirements of the existing measurement process. This
specification provides for up to 26 rotations at 200 points per revolution to be
collected in a continuous revolution data collection mode.
Results of the
integrator are transmitted to the host computer in units of 108 V·s.
3.1.2
Signal Multiplexing
A Keithley 7001 multiplexer was used to select the input signal for a specific
measurement set. The multiplexer has two available slots for signal banks. Each
bank is divided into four sections that provide 10:1 multiplexing. One bank is
dedicated to the rotating coil multiplexer stand, two other banks are reserved for
an existing translating dipole stand, and the fourth is reserved for a stretched wire
translational stand. Bucked signals require that the voltage inputs from two coils
be routed to a single channel. To do this, the positive sides of the two coils being
bucked are routed to a new channel and their associated negative leads are tied
together creating a complete circuit. This configuration prevents the bucking
process from being done electronically through the Keithley multiplexer. Bucked
signals were routed to a new channel in the required configuration prior to the
signal being delivered to the multiplexer as a separate input. The assembly
chapter also describes an alternative configuration.
14
3.1.3
Mulitplexer Chassis Box
A chassis box was designed and constructed at JLab to allow the measurement
system’s input signals to be routed through the Keithley multiplexer and returned
to a single output BNC connection. From the chassis output, the signal is then
routed to the PDI signal input. The chassis box provides a convenient place in the
measurement rack to manipulate the signal wires. The chassis box works in
conjunction with the Keithley multiplexer and is able to handle the signals from
up to four different measurement stands. The inputs from each stand are routed
through twisted pair cable to the chassis box. The cables attach to the chassis
through 25 pin D connections.
Inside the chassis box each signal wire is
separated and attached to a DIN rail connection point as shown in Figure 3-I.
This provides a convenient place for a computer engineer to troubleshoot signal
problems that may be associated with the measurement coils.
15
Figure 3-I Top view of the multiplexer chassis box interior.
Each terminal strip corresponds to one multiplexer bank.
These interior connection points in the chassis box are also used to route signal
wires from the coils to create a bucked coil wiring configuration prior to the
signal exiting the chassis box. All signals are then routed to corresponding 25 pin
D outputs connections at the rear of the chassis box. These signal outputs are
connected to the multiplexer. For each multiplexer bank, there are 10 input
16
signals and one output signal. The output signal from the multiplexer is routed
back to the chassis to the corresponding multiplexer chassis terminal strip. This
output is finally connected to a single bank output BNC connection and a four
bank sum BNC output connection on the front of the chassis box.
3.1.4
Wire and Cabling
The precision requirements and the low voltage signals make it important to
consider the types of wire used for data and voltage transfer when attempting to
accurately determine the magnetic field of an electromagnet. The induced voltage
from the measurement probes must be accurately transferred to the PDI for the
measurement to be reliable. In the case of a complex measurement system, the
coil signals may be routed to a multiplexer and then the appropriate signal will be
forwarded to the integrator. Additionally, signals are routinely bucked. When a
signal is bucked, there are two coils present. One coil is designed in such a way
that it cancels a portion of the other coil, usually the main harmonic, when their
output voltages leads are connected. The remaining signal is integrated, leaving
a strength difference that can be used to evaluate higher harmonic content or to
assess environmental differences that may occur during the magnetic
measurement. Bucked measurements result in weak signals, typically less than
10mV. This requires that system noise be minimal and that proper wire gauge is
used to propagate the signal to the integrator accurately. At each point that the
signal is routed, such as the multiplexer, the system must be checked for any
measurable noise contribution to the measurement. [7].
17
The transmission lines used in the system consists primarily of 20 and 22 gauge
twisted-pair cable. The properties of twisted pair cable make it a good choice as
the transmission media for the induced voltage signal. Electrical noise and crosstalk in the cable is minimized in this type of wire. A thicker wire gauge was used
in the PDI system than was used in the CAMAC system to ensure voltage drops
caused by resistance would be minimized. A series of tests measuring the voltage
differences at each end of the wire lines was conducted to ensure wire resistance
would not affect the measurement system. The results of these tests are discussed
in the commissioning chapter.
The insulated cables are used in the probe
connections and transfer cables on the measurement stand. The transfer cables
run from the probe inputs to the signal chassis box. The connections between the
chassis box and the multiplexer are also composed of twisted pair cable as is the
connection between the chassis box signal output and the PDI signal input.
3.2
Motion
Motion control is required by the measurement system to spin the measurement
probe while the PDI collects voltage samples. The PDI reads the encoder pulses
and takes measurements at programmable intervals preset prior to the
measurement. To accomplish this, a stepper motor with an attached incremental
encoder was selected along with a stepper motor motion controller components
and motor power amplifier.
18
3.2.1
Power Drive
The National Instruments MID-7602 power drive is a two axis power drive and
system interface used to operate the JLab multipole stand stepper motor. The
drive is used to connect motors, encoders, limit switches, I/O, and other motion
hardware to the National Instruments FW-7344 motion controller.
The MID-7602 was selected because it provided adequate power (output current
up to 1.4 Amps) and is compatible with the motor and encoder used in the new
system.
The CAMAC system collects 200 data points per revolution at every
1.8° as does the PDI system. Because both the MID-7602 and the FW7344
motion controller are built by National Instruments (NI), they can be easily used
together to control the stepper motor.
Also, in the case of a configuration
problem, NI support can be utilized to help develop a solution.
3.2.2
Stepper Motor Control
The FW-7344 motion control device is a fully programmable motion controller
for up to four independent coordinate motion axes. It uses a Firewire (FW)
connection, as opposed to a PCI card, for communication between itself and the
PC. It provides motion control for both servo and stepper motors and I/O for limit
and home switches. The JLab multipole stand uses a 4,000 count/revolution
encoder together with a 50,000 count/revolution motor to spin the harmonic
probe. The FW-7344 provides a straightforward and complete control system for
motion control of the JLab system as well as future additions to the system.
Configuration of the FW-7344 is done through National Instruments
19
Measurement and Automation Explorer (MAX) software. Flex Motion control
software, required for operation of the controller, is included in the MAX
installation with versions 3.1 and higher. Specific configuration settings are
described in Appendix E.
3.2.3
Motor and Encoder
A Maxon stepper motor is used to control the rotation of the harmonic probes on
the multipole measurement stand. The HEDL-6540 encoder is used to control the
motor rotation. The encoder offers a two channel quadrature output and a third
channel used for the index pulse.
The two output channel square waveforms (A and B) are accessed through a cable
and 10 pin female connector. The signals are then routed both the MID-7602
motor amplifier and the PDI unit. The index signal is not used in the current
measurement system configuration. Instead, an optical sensor is used to send an
index pulse at the start of a measurement.
3.2.4
Triggering
The HOA7720 optical trigger is used to send an index pulse to the PDI to signal
the start position of the measurement. The trigger consists of an infrared emitting
diode facing an Optoschimitt detector. A single TTL output is connected to the
PDI trigger input. Using inverting logic, the output emits a low signal when the
optical path is clear and a high TTL signal when the path is interrupted.
20
3.3
Data Transfer and Analysis
3.3.1
Hardware Communication
The National Instruments GPIB-ENET/100 is an Ethernet-to-GPIB controller.
GPIB communication is used to control and synchronize the various hardware
components in the measurement system.
The ENET/100 allows networked
computers to communicate with and control IEEE 488 devices from anywhere on
an Ethernet-based TCP/IP network. The decision to use the ENET/100 was made
to eliminate the need for a PCI-GPIB card to be installed on the host computer as
well as enabling wireless control of the measurement system. This will make
replacing and upgrading host computers significantly easier.
3.3.2
Host Computer Requirements
A PC using either a Windows2000 or XP operating system is required with an
Ethernet network connection for GPIB communication and Firewire cable
connection for motion control is required for operation of the measurement
system. Appropriate permissions to access the network drive for the magnet test
group at JLab are also required to execute the aquisition software version due to
the directories the software must access to read and write data during execution
and initialization.
21
3.4
Magnet Powering
While different magnets have different powering requirements, the most common
quadrupole magnets at JLab are powered using a trim card powering system. In
this system, a trim card is used to independently access voltage from a bulk power
supply. In this way, multiple trim cards can be attached to the bulk supply and
fed to various magnets. The trim card system is controlled by Experimental
Physics and Industrial Control System (EPICS), a set of collaboratively developed
software tools, libraries, and applications used to create distributed real-time
control systems.
3.4.1
Epics
EPICS is used extensively in particle accelerators throughout the world and is the
main control system used at JLab. The specific function of EPICS, as it relates to
the measurement system, involves the use of its JLab standard protocols to run
hysteresis loops and set currents.
Because EPICS is run from UNIX based
machines through TCP-IP and UDP based communications, a Windows based
EPICS Client has been set up on the Windows based Data Acquisition System
(DAQ) in the magnet measurement area at JLab. This system has also been tied
into the automated DAQ software used for magnet measurement.
Even though this will be a totally different method than that utilized by the
magnet measurement group at JLab in the past, EPICS offers distinct advantages
22
over prior test lab based powering methods. First, EPICS is a real time control
system capable of operating at a 10Hz resolution. Second, and more importantly,
EPICS is the primary means by which all magnets in CEBAF and the FEL are
cycled and set. Using the same current powering control system to map the
magnets that will be used to set their current once they are installed in the
accelerator or FEL should provide the most realistic field map. By utilizing
EPICS, the magnet measurement hysteresis and current setting can match exactly
that which is used to power the magnets in the FEL and CEBAF accelerator.
Also, the effects of overshoot have been handled in the magnet powering protocol
of the EPICS control system. This eliminates the need to quantify differences in
the magnet powering techniques used in the magnet test facility versus the FEL.
3.4.2
Three Piece Ramp
To provide a means by which certain magnets can be sent to off-site facilities for
supplemental measuring, a secondary powering system utilizing National
Instruments FieldPoint Analog Output control is also used.
In this case, a
separate but standard protocol “a three-piece linear ramp” ensures consistency in
ramp rate for a range of power supplies. The first step of the ramp is run at 1 A/s.
At 1 A from the set point, the ramp speed is decreased by a factor of 10, and at 0.1
A from the set point the rate is decreased further by a factor of 10. Figure 3-II
shows the different stages of the three piece ramp protocol when setting a current
from 0 to 10 A. Using this method, tests have shown that there is no measurable
overshoot of the current at its set point.
23
12
10
Current (A)
8
6
4
2
0
0
5
10
15
20
25
30
Time (s)
Figure 3-II: Stages of the three piece ramp protocol used to set
magnet current. The first step of the ramp is run at 1 A/s. At 1 A from
the set point, the ramp speed is decreased by a factor of 10, and at 0.1 A
from the set point the rate is decreased further by a factor of 10.
At JLab, a Danfysik power supply is used to both mimic EPICS magnet powering
protocols and implement custom protocols such as specialized hysteresis and
ramp rate algorithm control. Using a +/- 10 volt analog input signal as its control,
the Danfysik will output +/- 20 Amps to the magnet. An FP-AIO-610 Ethernet
module provides 12 bit resolution and operates between 0V – 10V or -10V +10V making it ideal for computer automated control.
In the event that
supplemental magnet measurements are desired from an external measurement
24
facility, a powering algorithm dynamic link library (DLL), a software control, can
be written and packaged together with a specific power supply so that it will be
accurately powered. This algorithm can then be modeled in either EPICS or on a
microcontroller so that once the magnet is installed in the FEL or accelerator, it
can be set accurately.
3.4.3
Jlab Control Software.
The measurement process is controlled by a software program written in C using
LabWindows/CVI as the compiler and IDE. SLAC provided control software that
is currently used at their facility for their PDI measurement system. This software
was used as a foundation from which the JLab measurement system software
would be created.
The goal of the software is to create a hybrid of the two existing software models
that can be used by a magnet technician with an intermediate skill level to operate
the PDI measurement system. Also, there is a desire to create data output that is
consistent with the output from the existing CAMAC measurement system
software.
The SLAC software was written to be operated by a programmer or magnet
engineer with programming ability. Changes to the parameters specific to each
magnet measurement must be changed inside the code and the code recompiled
prior to the measurement. Interaction with the user is done through a DOS
25
window command line prompt although several informative GUI graphs could be
made visible during a measurement by setting code parameters.
After a review of both the SLAC and CAMAC code models the decision was
made to make changes to the code directly to the existing SLAC code. Even
though several of the components used in the SLAC measurement system are the
same as those used at JLab, several components are different and needed to have
code written for instrumentation control. The initial focus was to implement code
for the different hardware used at JLab and tie it into the SLAC code in place of
their similar components. Specific instrumentation controls have been developed
for the motion hardware and data acquisition switch unit as well as controls built
to handle magnet powering during the measurement.
To decouple the need for a programmer to be involved in setting up the
parameters for a magnet measurement, a GUI interface, parameter file, and
parameter file editor were developed for the new PDI software. The start up
screen for the new PDI software is shown in Figure 3-III. From this screen the
user can select an appropriate parameter file, set specific job information, and
initiate a measurement run. Expert control screens for the various component in
the system can also be accessed from the start panel for system debugging,
creating new parameter files, and general component operations.
26
Figure 3-III: PDI software start screen. This panel allows the user to
enter measurement information, comments, and load parameter files
before the start of a magnet measurement. It also allows the user to access
the various instrumentation control panels for high level instrumentation
communication and debugging.
The CAMAC parameter file was used as a template to create the PDI parameter
file. The CAMAC parameter file contained some unused information which was
excluded from the PDI parameter file in an effort to minimize complexity for the
user. Two GUI interfaces were written to create parameter files that would be
used to configure the PDI measurement. This parameter file utility panel, shown
in Figure 3-IV, is used to configure the typical parameters used for a magnet
measurement. It allows the user to build a working parameter file using drop
down menu selections. It is especially useful when defining an extensive list of
currents settings to be set during the measurement process.
27
Figure 3-IV Parameter file creator software panel. This panel allows
the user to enter standard parameters for a magnet measurement and
creates a parameter file that is loaded at runtime.
A secondary editor can be used to make high level changes to the PDI parameter
file. In this screen the user can open either an existing PDI parameter file or
template and then make changes directly to the file parameters.
When this
interface is used, the original file is displayed in a bottom window allowing the
user to compare the changes that have been made before saving the file. Changes
are made directly to the parameter text file. When the new file is saved, the
28
original file is automatically backed up with a decremented extension (*.999).
Figure 3-V Expert parameter file software panel. This panel allows the
user to work directly with a parameter file. The user can open a file and
edit individual parameters on the top window. The bottom window
displays the original opened file so that the user can review changes.
The PDI Runtime Panel, Figure 3-VI, is displayed during a measurement along
with two graphing panels developed at SLAC, Figure 5-VI, that plot data relevant
to the measurement in an easy to interpret format. The main panel displays the
harmonic results from the FFT as well as provides feedback to the operator about
the current progress of the measurement and magnet current. The graphing panels
29
display the output from the PDI unit as well as a breakdown of the harmonic
strengths from the FFT. The magnet strength, relative to the main harmonic, is
also displayed for each measurement current.
Figure 3-VI PDI software runtime panel. This panel displays
measurement progress, general information, and runtime analysis.
30
Figure 3-VII PDI software runtime plots. These panels display various
aspects of the magnet measurement in a graphical format.
Additional expert control panels are available for each component for testing and
configuration. The expert panels and their descriptions are listed in Appendix D.
3.5
Supplementary Components
3.5.1
Signal Processing
The Agilent 34970A data acquisition switch unit was selected to handle various
input voltages and signals that may be used to monitor measurement conditions
during the measurement process. Its primary function will be to read the voltage
received from the current shunt while a magnet is being measured. The Agilent
34970A provides 6.5 digit multimeter accuracy, stability, and noise rejection
making it a suitable choice for this measurement requirement. Additionally, the
31
unit has up to 60 channels which can be separately configured to handle various
types of signals. Voltages, thermocouples, RTDs, and pressure sensors could be
routed to the data switch and read during a magnet measurement.
Communication with the instrument is made through its GPIB interface.
32
4
Measurement System Assembly
Assembly of the measurement system involved a variety of electronic and
mechanical design considerations. To accommodate the need for the existing
CAMAC components to remain functional as the new stand is built and
commissioned, additional fixtures were constructed to support the new motor and
optical trigger. These fixtures allowed the existing motor used with the CAMAC
system to be easily interchanged with the motor used in the PDI system. The
CAMAC encoder and triggers could also remain in place when switching between
systems. Magnet powering required the integration of two new magnet power
supply options. A 20 amp bipolar Danfysik power supply would be integrated for
magnet facility testing and EPICS powering would be integrated to create better
uniformity between the magnet measurement facility and the powering after
installation. Each communication, TTL, and induced voltage signal was routed to
their appropriate hardware interfaces and synchronized through software. To
evaluate each magnet’s harmonic content the bucked wire configuration would be
completed to deliver the appropriate signal to the integrator. Specific assembly
and wiring configurations are detailed in Appendix E.
4.1
Magnet Powering Instrumentation Control
Testing showed that the existing EMI Boss power supply used to power many of
the quadrupole magnets during their measurement would not maintain a stable
current without allowing for a significant settling time (5-10 minutes) after setting
33
the current.
A new Danfysik power supply was purchased to provide both
improved stability and increased output current capability.
Intrumentation
control for the Danfysik is accomplished through analog signaling. A National
Instruments Field Point Analog Input Output (FP-AIO) control module was
purchased to control the power supply. A 10V analog input signal corresponds to
a 20 amp output from the Danfysik power supply. The FP-AIO controller has a
12 bit DAC, or 4096 (212) steps over the measurement range. For +/- 10V, this
provides about 5mV resolution. This level of resolution provides a means by
which a magnet can be set to a specific current without causing current overshoot.
To provide the same means of powering magnets as is done in the accelerator, a
trim card slot was assigned to the magnet measurement facility and a power lead
run from the supply to the measurement stand. Instrumentation control of the trim
cards is accomplished through EPICS channel access (CA) control. The PDI host
PC was configured to utilize CA for EPICS [11]. The EPICS PC configuration
allows the computer to issue directives to EPICS over the command line. To
integrate EPICS into the PDI software, a series of .bat files were written that
could be called and edited by the PDI software at runtime. A current shunt is
used to verify each set current command is executed correctly during the
measurement process.
34
4.2
Bucked Signal Configuration
Some rotating coil probes contain two wire windings with specific geometry that
enable the two signals to be subtracted from each other (bucked). To collect
bucked signal data the positive lead of the induced voltage signal from each
winding are routed to a separate channel and the return leads are wired together.
The multiplexer closes the bucked channel and opens the direct channel for each
of the other signals creating a single closed circuit for the bucked signal. The
bucked wiring, as seen in Figure 4-I, can be configured inside the chassis box
before being routed to the multiplexer.
Figure 4-I Bucked signal wiring configuration. The positive leads of
the induced voltage signal from each winding are routed to a separate
channel and the return leads are wired together to cancel the quadrupole
signal and allow higher resolution harmonic analysis.
35
4.3
The Measurement System
The measurement system assembly was constructed in two parts. The first part
consisted of joining all of the rack-mountable components in the measurement
rack and establishing communication between each component and the host
computer. During the first stage of development, the rack was located in an office
setting away from the measurement stand. This provided a better atmosphere in
which to complete the detailed programming, test component wiring, and
configure hardware settings. A function generator was used to create a voltage
signal in place of a rotating coil. The component rack layout is shown in Figure
4-II
MetroLab PDI
Agilent DAQ/Switch Unit
NI ENET-100
Keithley 7001 Mux
Generic Power Supply
Mux Chassis Box
Drawer
NI FW-7344 Motion
Controller
NI MID-7602 Motor Amplifier
Figure 4-II Measurement system rack components
The second part of the measurement system assembly consisted of moving the
component rack into the magnet measurement lab, establishing communications
with the DAQ computer, and wiring the motor, encoder, and probe to the system.
36
Multi-stand parts were fabricated to quickly switch between the older CAMAC
system and the new PDI system. This was done so that, during commissioning,
the existing CAMAC components could be swapped out easily if multipole
measurements needed to be completed. The current layout for the measurement
stand is shown in Figure 4-III.
F
i
g
u
r
e
4-III Multipole measurement stand
37
5
Optimization
5.1
Cycle Analysis
A simulation was completed using System View and MatLab analysis programs
to analyze the differences between continuous and discrete rotation methods,
independent of the PDI measurement system. This program was used to generate
a ten period waveform, simulating five continuous forward probe rotations, and a
two period waveform, simulating one forward probe rotation. An FFT was then
carried out on the two data sets. The simulation frequency was set at 10 Hz,
sampled at 1000 Hz and the signal set at 1 Volt, with 1% Gaussian noise added.
The 1,000 Hz sampling rate is equivalent to the PDI data acquisition rate of 200
samples per revolution. Figure 5-I shows a two period waveform and the FFT of
the average of five, two period cycles.
Figure 5-I Simulated frequency convergence – averaged.
horizontal axis indicates the frequency resolution in Hz.
38
The
When five continuous cycles were used, the number of frequency intervals
increased which resulted in a better frequency resolution.
When measuring
magnets, a complex waveform is produced by the induced voltage picked up by
the rotating probe with the number of samples per revolution corresponding to a
larger number of data points. Spinning the probe continuously provides more
data points which, in turn, produce an increasingly accurate representation of the
harmonic content of the magnet. Figure 5-II shows a ten period waveform and
the FFT of that waveform.
Figure 5-II Simulated frequency convergence - continuous cycles. The
horizontal axis indicates the frequency resolution in Hz. The continuous
cycle method produces a more accurate representation of the magnet’s
harmonic content.
The continuous rotation method produces more zero crossings increasing the
ability of the FFT routine to resolve the frequency of the signal. As the number of
zero crossings increased, the uncertainty, a consequence of the complexity of the
39
waveform, decreased resulting in a clearer overall representation of the induced
signal as seen by comparing Figures 5-I and 5-II.
The PDI software used the real and imaginary components calculated by the
LabWindows/CVI ReFFT function to extract the desired harmonics. To do this,
the real and imaginary values from the FFT data were extracted at multiples of the
number of continuous rotations.
For example, if the probe was spun for 5
revolutions in a quadrupole magnet, the quadrupole term would correspond to
(n=2) * (5 revolutions) = 10. The function used to determine the harmonic
content from the FFT data is shown in Figure 5-III.
void vtcoil_calc_vthar(int num_samp, double vtfft_re[],
double vtfft_im[],
int num_rev, int num_har, double vthar_re[], double
vthar_im[])
{
F
i
g
u
r
/* The harmonics are at 1, 2, 3, ... cycles per
revolution */
vthar_re[0] = 0.;
vthar_im[0] = 0.;
for (i = 1; i <= num_har; i++)
{
vthar_re[i] = vtfft_re[i * num_rev];
vthar_im[i] = vtfft_im[i * num_rev];
}
/* Done */
return;
}
5-III Software function used to calculate the magnet’s harmonic
content.
40
5.2
Cycle Options: Forward versus Forward-Reverse Averaging
Tests were conducted to quantify and compare the repeatability of measurements
taken in the forward direction only with measurements taken in the forward and
reverse direction and then averaged. Forward-Reverse averaged measurements
have been traditionally used in the CAMAC system.
Each measurement probe is attached to the motor using hard rubber couplings
which employ slots to lock the probe into position as it spins through the magnet.
A small degree of play was observed when attaching the probe to the coupling.
Coupling play will induce backlash error when spinning the probe in the reverse
direction with respect to the encoder position.
Data was taken using both
measurement methods and compared to quantify the differences between the two
methods.
Both measurements methods were alternated to verify external
environmental factors would not skew the results.
Figure 5-IV shows the results of the two methods. A quadrupole magnet, known
as JLab QX062, was used for this test and the motor acceleration and velocity
parameters were set to a velocity of 20,000 steps/s and acceleration of 20,000
steps/s2 where 50,000 steps/s at a constant velocity would produce a one
revolution per second rotation rate. 200 samples per revolution were collected
during data acquisition. Each measurement represents a single probe cycle using
each method with no cycle averaging. The forward and reverse averaged data
shows less stability and exhibits a strength offset from the forward only data.
41
Future plans for the measurement stand include purchasing mechanical probe
couples will replace the existing couplings to minimize backlash error.
12130
Forward Only
Integrated Probe Voltage (uV*s)
12125
12120
12115
12110
12105
Average Fwd-Rev
12100
12095
12090
12085
12080
0
2
4
6
8
10
12
14
16
18
20
Measurement Number
Figure 5-IV Forward only vs. Forward Reverse measurement
repeatability. The forward - reverse averaged measurements show larger
deviations from the mean as well as a larger magnet strength offset. This
may be attributed to mechanical backlash in the probe connection
couplings.
42
5.3
Cycle Averaging Optimization
To determine the optimum number of cycles that should be taken during data
collection three runs of 20 consecutive measurements were taken in the forward
direction. The quadrupole term of each measurement was added to array of data
samples. The standard deviation of the sample was calculated as each point was
added according to
( x  x) 2
s n 1 
n 1
(5.1)
.
The changes in the standard deviation as more measurements were introduced
into the sample are illustrated in Figure 5-V. The standard deviation between
measurements begins to stabilize at approximately 5 measurements. This should
be the minimum number of cycles that should be taken and averaged during a
measurement. The third data set showed a jump in the standard deviation at the
twelfth data point which was caused by an anomalous data reading at that
position. The data beyond this point is not shown in the third data set. A method
to improve statistics in the case of data outliers is discussed in the next section.
43
1.6
Standard Deviation (uV*s)
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0
5
10
15
20
Measurement Number
Figure 5-V Changes in standard deviation as measurements are added
to sample. For each data set, the standard deviation approaches an
equilibrium value after approximately five measurements. Thus, a
minimum of five measurements should be taken to provide an adequate
sample size.
5.4
Improving Statistics
Environmental factors cause occasional data outliers to appear in the
measurement process.
A method could be automated into the measurement
process that can detect measurement outliers and flag the system to take
additional measurements.
To do this, the standard deviation of the default
measurement sample would be calculated. The standard deviation of the sample
set could then be compared to a standard deviation tolerance derived from the
magnitude of the sample values. Each point would then be compared to identify
the maximum outlier and another measurement would automatically be taken and
44
replace the worst outlier or added to the overall measurement set. An example of
the process is shown in Table 5-I.
Meas. Number
Sample Val
Diff from
Avg
Meas.
Number
Sample Val
Diff from
Avg
1
2
3
4
5
12119.973
12119.763
12118.455
12118.591
12121.284
0.3598
0.1498
-1.1582
-1.0222
1.6708
1
2
3
4
6
12119.973
12119.763
12118.455
12118.591
12119.715
0.6736
0.4636
-0.8444
-0.7084
0.4156
Average
Tolerance
StDev
12119.6132
1
1.15431677
New
Average
Tolerance
StDev
12119.2994
1
0.71697754
Table 5-I A proposed method to improve the standard deviation of
measurement samples.
45
6
Commissioning
A series of system tests have been conducted on the MetroLab PDI multipole
measurement system. These tests were conducted for commissioning purposes
and are intended to quantify the repeatability of the PDI system under various
conditions including:
1.
Quantifying the repeatability of the quadrupole term of a reference signal
simulating 5 continuous forward probe rotations when
i. The reference signal is plugged directly into the PDI unit
ii. The reference signal is plugged into each of the three measurement
probe coil locations
2.
Quantifying the repeatability of the quadrupole term of a reference signal
simulating 5 discrete forward rotations (averaged) when
i. The reference signal is plugged directly into the PDI unit
ii. The reference signal is plugged into each of the three measurement
probe coil locations
3.
Quantifying the repeatability of the quadrupole term at a 5 Amps for each
rotating coil probe
i. P1A – 1 inch Halbach style probe
ii. P1C – 1 inch Single coil probe
iii. P2A – 2 inch Halbach style probe
iv. P2B – 2 inch Single coil probe
v. P3A – 3 inch Halbach style probe
46
4.
Comparing two signal analysis algorithms for differences
i. FFT algorithm from CAMAC code
ii. FFT algorithm from the National Instruments function library
6.1
Reference Signal Repeatability
For normal data acquisition operations, the voltage signal from the measurement
probe is routed through a series of couplings, cables, and hardware upstream of
the PDI unit. An HP 33120A function generator was used to create a +/-300 mV,
1.596 Hz reference voltage signal. The 300 mV amplitude of the reference signal
is comparable to the voltage induced in the one inch, single coil probe, P1C, when
measuring a QA magnet at 3 Amps.
Initial measurements were conducted with a reference signal input directly to the
PDI unit, in an effort to quantify the ‘best case’ repeatability of a simplified
system using the previous CAMAC production parameters [9]. A comparison
was made between averaging five two period cycles, representing five discrete
forward rotations of the measurement probe in a quadrupole magnet, and
analyzing a single 10 period cycle, representing five continuous forward rotations
of a measurement probe in a quadrupole magnet. Both sets of data exhibited short
term (less than four hours) reproducibility across measurement sets at or less than
0.015% for the 1.596 Hz, 300 mV reference signal. The setup and results of these
measurements are detailed below.
47
6.2
Function Generator Reference Signal Setup
The signal from an optical trigger was split and used to synch the PDI unit and the
function generator as the motor rotated through 360 degrees. The reference signal
was set such that one 360 degree rotation of the motor was coincident with two
signal periods, simulating the rotation of a measurement probe in a quadrupole
magnet. The PDI uses encoder information from the motor to integrate incoming
signals; therefore synchronization between the motor and the reference signal was
essential. Two hundred data points were collected for a single 360 degree rotation
of the motor.
Discrete Forward Rotation
The current method of measuring multipole magnets on the rotating coil stand
involves collecting data on the forward, 360 degree, revolution of the
measurement probe. The probe rotation is then reversed and data is collected
during the reverse 360 degree rotation.
The forward and reverse data are
averaged, and this entire process is repeated five times, and concludes when the
five data sets are averaged to represent the magnet induced waveform.
In the simulated signal testing, however, it was not feasible to average data when
simulating
a
rotating
measurement
because
the
complexities
in
the
synchronization of the waveform in the reverse direction could introduce
48
additional measurement error. Therefore, only data simulated for the forward
rotation were analyzed.
Once five discrete forward rotations had been
simulated, the 5 arrays of data points were averaged and analyzed using an FFT to
resolve the harmonic contents of the average wave form.
Continuous Forward Rotation
Continuous probe rotation is a method of data acquisition used in many magnet
measurement laboratories. To accomplish continuous rotation, slip rings must be
used to allow the measurement probe to rotate multiple times in one direction
without the need for reversing.
To simulate continuous rotation, the limit switches were removed from the
rotation stand and the encoder position zeroed approximately 45 degrees behind
the optical synchronization trigger. A trigger arm attached to the motor shaft
caused the optical sensor to fire a TTL signal, triggering the PDI to begin data
acquisition and the HP 33120A to begin a ten period burst as the motor rotated
through five revolutions. Continuous rotation provided additional zero crossings
allowing the FFT function to better resolve the waveform. The PDI collected
1,000 data points during the five rotations before completing data acquisition and
transferring the integrated voltage samples to the host computer.
49
6.3
Continuous Rotation Testing with Reference Signal
Continuous Rotation – Direct PDI Connection
Tests were conducted to quantify the repeatability of simulated continuous probe
rotation. As described previously, a single ten period cycle was used to represent
the signal induced from five continuous forward rotations of a measurement probe
in a quadrupole magnet.
Figure 6-I, shows the results of the test, where 1,000 individual samples, 200
samples per revolution for five revolutions, were collected during data
acquisition. Data sets for ‘Run 1’, ‘Run 2’ and ‘Run 3’ show the averaged
quadrupole term from ten independent measurements.
5 Continuous Revolutions
Signal Connected at the PDI Directly
10 Measurements per
Run
Run 1
Run 2
Run 3
Overall
Maximum Integrated Voltage
Amplitude (uV*Sec)
30950.42
30950.67
30950.13
30950.67
Minimum Integrated Voltage
Amplitude (uV*Sec)
30948.94
30948.70
30949.17
30948.70
Difference (uV*Sec)
1.48
1.97
0.96
1.97
Deviation from Average (%)
0.005%
0.006%
0.003%
0.006%
Table 6-I Initial system repeatability using a simulated signal - Direct
connection / continuous revolutions. Three separate runs consisting of
10 measurements were compared to determine system repeatability over
the course of a run and overall system repeatability across runs.
50
Continuous Rotation – Coil 1 Probe Location
To investigate system noise, the reference signal input was moved from a direct
connection on the PDI unit, to the coil 1 input location for the rotating coil probe.
From this location the reference signal passed through the entire data acquisition
system, a series of twisted pair cables, DIN connectors, a signal chassis box, and a
multiplexer before reaching the PDI unit.
Table 6-II shows the measurement results after the reference signal was moved to
the coil 1 location. Data sets for ‘Run 1’, ‘Run 2’ and ‘Run 3’ show the averaged
quadrupole term from ten independent measurements. The system repeatability
for a given ten run data set was better than 0.02%. However, the maximum
spread across the entire thirty measurements comprising these three runs was
0.05%. A contributor to this degradation in repeatability is associated with the
signal drift across the three runs.
The drift could be associated with
environmental factors or small synchronization errors between the function
generator burst and motor encoder, causing the PDI unit to integrate different
amounts of the reference signal for the individual runs. The stability of the signal
produced by the function generator was found to be very dependent on the
stability of the temperature of the function generator itself.
51
5 Continuous Revolutions
Signal Connected at Coil 1 Probe Location
10 Measurements per
Run
Run 1
Run 2
Run 3
Overall
Maximum Integrated Voltage
Amplitude (uV*Sec)
30957.02
30960.27
30967.96
30967.96
Minimum Integrated Voltage
Amplitude (uV*Sec)
30952.36
30958.28
30964.92
30952.36
Difference (uV*Sec)
4.66
1.99
3.03
15.60
Deviation from Average (%)
0.015%
0.006%
0.010%
0.050%
Table 6-II Initial system repeatability using a simulated signal – Coil 1
location / continuous revolutions. Three separate runs consisting of 10
measurements were compared to determine system repeatability over the
course of a run and overall system repeatability across runs at the probe
connection. The input signal was propagated through the measurement
system components before integration.
Continuous Rotation – All Coil Locations
There are two other coil input locations on the rotating coil stand in addition to
the coil 1 location. Each coil location was tested using the continuous rotation
method to verify consistency in system repeatability across coil locations. A
series of three data sets, consisting of ten separate measurements of the 1.596 Hz
reference signal, were taken at each of the other two locations. Figure 6-I shows
the deviation in the quadrupole term from the measurement average, for each of
the ten measurements taken in each of three runs, at the three coil location.
52
Coil 2
0.010%
0.010%
0.008%
0.008%
Deviations from Average (%)
Deviations from Average (%)
Coil 1
0.006%
0.004%
0.002%
0.000%
-0.002%
-0.004%
-0.006%
-0.008%
0.006%
0.004%
0.002%
0.000%
-0.002%
-0.004%
-0.006%
-0.008%
-0.010%
-0.010%
1
2
3
4
5
6
7
8
9
1
10
2
3
5
6
7
8
9
10
8
9
10
Average
Coil 4
0.010%
0.010%
0.008%
0.008%
Deviations from Average (%)
Deviations from Average (%)
4
Measurement Number
Measurement Number
0.006%
0.004%
0.002%
0.000%
-0.002%
-0.004%
-0.006%
-0.008%
0.006%
0.004%
0.002%
0.000%
-0.002%
-0.004%
-0.006%
-0.008%
-0.010%
-0.010%
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
Measurement Number
Measurement Number
Figure 6-I Main harmonic amplitude reproducibility for each coil using 5
sequential cycles. The amplitude deviations for each coil meet the desired
0.01% reproducibility specification.
6.3.1 Five Cycle Averaged Rotation Testing with Reference Signal
Discrete Forward Rotations Averaged – Coil 1 Probe Location
The reference signal was connected at the coil 1 probe location and measurements
were made so as to simulate five individual forward probe rotations. The results
53
of these five rotations were averaged. This process was repeated ten times for
‘Run 1’, ‘Run 2’ and ‘Run 3’ respectively.
Table 6-III shows the results from averaging the five forward rotations. The
performance of the system is slightly degraded in terms of system repeatability for
each run when compared to the continuous rotation data. The worst case set of
ten measurements, ‘Run 3’, shows a system repeatability of 0.014%. There was,
however, less drift in the absolute value of the quadrupole term during the
measurements of these three runs when compared to the continuous rotation runs.
The maximum spread across the entire set of thirty measurements constituting
these runs was 0.019%, a factor of 2.5 better than the system repeatability of the
thirty measurements used for the continuous rotation tests.
Five Averaged Revolutions (Forward Only)
Signal Connected at Coil 1 Probe Location
10 Measurements per
Run
Run 1
Run 2
Run 3
Overall
Maximum Integrated Voltage
Amplitude (uV*Sec)
31009.98
31007.99
31010.12
31010.12
Minimum Integrated Voltage
Amplitude (uV*Sec)
Difference (uV*Sec)
31006.30
3.68
31004.28
3.72
31005.77
4.35
31004.28
5.84
Deviation from Average (%)
0.012%
0.012%
0.014%
0.019%
Table 6-III Initial system repeatability using a simulated signal – Coil
1 location / five averaged cycles. Three separate runs consisting of 10
measurements were compared to determine system repeatability over the
course of a run and overall system repeatability across runs at the probe
connection. The input signal was propagated through the measurement
system components before integration.
54
5 Discrete Forward Rotations Averaged – All Coil Locations
Tests were repeated at the other two coil probe locations. Figure 6-II shows the
deviation in the quadrupole term from the measurement average, for the ten sets
of data taken in each of three runs, at each coil location. This data is somewhat
noisier than similar data obtained for the continuous rotation tests.
55
Coil 2
0.010%
0.010%
0.008%
0.008%
0.006%
0.006%
Deviations from Average (%)
Deviations from Average (%)
Coil 1
0.004%
0.002%
0.000%
-0.002%
-0.004%
-0.006%
0.004%
0.002%
0.000%
-0.002%
-0.004%
-0.006%
-0.008%
-0.008%
-0.010%
-0.010%
1
2
3
4
5
6
7
8
9
1
10
2
3
4
Coil 3
6
7
8
9
10
7
8
9
10
Average
0.010%
0.010%
0.008%
0.008%
0.006%
0.006%
Deviations from Average (%)
Deviations from Average (%)
5
Measurement Number
Measurement Number
0.004%
0.002%
0.000%
-0.002%
-0.004%
-0.006%
-0.008%
0.004%
0.002%
0.000%
-0.002%
-0.004%
-0.006%
-0.008%
-0.010%
-0.010%
1
2
3
4
5
6
7
8
9
10
Measurement Number
1
2
3
4
5
6
Measurement Number
Figure 6-II Main harmonic amplitude reproducibility using the average
of 5 cycles. The amplitude deviations at each coil are greater than the
deviations measured using the continuous cycle method. The desired 0.01%
reproducibility specification is exceeded at each coil location, with an overall
average deviation of 0.016%. Further optimization techniques will be applied
to improve and optimize the repeatability of the measurements.
56
6.4
Signal Analysis Algorithm Comparison
During the multipole measurement process, the PDI system integrates voltage
samples according to:
t
 V (t )dt (0)  ( ) ,
(6.1)
0
for each coil rotation. These integrated values (V·s) are then transferred to the
control computer. To understand the harmonic content of the waveform, an FFT
algorithm is used to obtain the normal and skew field components before
performing amplitude and phase calculations for the desired harmonics.
The CAMAC data acquisition software used an algorithm coded at Jefferson Lab
to calculate and normalize the voltage integrals before computing the amplitude
and phase of each harmonic. The PDI software uses a LabWindows/CVI library
function to perform an FFT on the data.
To verify that the PDI and CAMAC FFT algorithms, and subsequent amplitude
and phase calculations, are consistent with one another, the CAMAC FFT
function was incorporated into the PDI code and reworked to work with the PDI
array structures and indexing. Both algorithms use similar code to compute the
amplitude of each harmonic but the phase angle computations are slightly
different. Two data sample, one that was used to process the integrated voltage
57
samples using the CAMAC algorithm and one that was used to process the
samples using the PDI algorithm, were taken to collect information for
comparison. The magnet was cycled and set to 5 Amps prior to the first run and
was left at 5 Amps through the second run to minimize error that might result
from magnet cycling. Data was taken and the phase angles were computed using
both algorithms. Results of the analysis showed reasonably consistent phase
angles at each harmonic. Table 6-IV shows phase angles using both algorithms
for the specified harmonic.
58
CAMAC FFT Algorithm Results (degrees)
Avg Curr
n= 1
n= 2
n= 3
n= 4
n= 5
n= 6
n= 7
n= 8
5.0
-133.59
-61.07
37.68
-28.78
12.39
-27.45
-0.80
-13.46
5.0
-134.03
-61.08
37.84
-31.71
24.22
-23.46
-15.53
-13.73
5.0
-132.90
-61.07
36.92
-30.24
16.29
-24.90
3.87
-12.04
5.0
-134.18
-61.09
38.08
-33.57
24.63
-26.94
5.19
-15.53
5.0
-134.75
-61.10
38.96
-31.15
26.34
-25.53
15.56
-17.08
PDI FFT Algorithm Results (degrees)
Avg Curr
n= 1
n= 2
n= 3
n= 4
n= 5
n= 6
n= 7
n= 8
5.0
-133.88
-61.09
37.49
-34.42
20.98
-29.60
2.06
-10.11
5.0
-134.22
-61.09
38.29
-31.77
20.39
-27.50
2.48
-17.37
5.0
-134.60
-61.07
38.49
-24.58
25.95
27.58
2.84
-20.18
5.0
-133.60
-61.08
37.92
-25.76
20.19
-7.52
0.09
-13.28
5.0
-133.61
-61.08
37.69
-28.09
22.45
14.01
-5.20
-13.25
CAMAC FFT Algorithm Results (degrees) (cont.)
n= 9
n = 10
n = 11
n = 12
n = 13
n = 14
n = 15
n = 16
n = 17
-14.19
-0.17
-12.31
-1.23
-8.91
0.78
-7.02
6.35
-1.43
17.83
-3.26
-12.21
-4.43
-12.74
-8.25
8.38
3.82
-6.47
16.35
0.92
-12.57
-0.28
-6.48
-11.95
-2.47
-8.11
-0.76
17.46
12.22
-13.78
0.15
-10.99
-7.00
10.20
6.67
-4.54
18.41
-14.10
-13.93
10.42
-12.48
-4.20
-11.93
10.21
-5.94
PDI FFT Algorithm Results (degrees) (cont.)
n= 9
n = 10
n = 11
n = 12
n = 13
n = 14
n = 15
n = 16
n = 17
-16.71
5.76
14.86
4.14
-10.11
0.53
11.83
8.75
-1.66
-19.55
14.84
-13.80
5.12
-11.51
-12.77
10.83
7.34
-2.35
-16.25
12.66
-13.77
5.33
-11.94
-12.66
3.07
10.72
3.35
-13.60
10.50
3.40
1.46
-3.93
-9.31
8.05
-11.00
2.79
19.86
10.20
15.69
-0.68
-10.67
-7.61
10.05
8.55
2.04
Table 6-IV: Phase Comparison between CAMAC and PDI Algorithms
Table 6-V compares the amplitudes of the two FFT methods from the same two
runs.
59
CAMAC FFT Algorithm Results
Avg Curr
n= 1
n= 2
n= 3
n= 4
n= 5
n= 6
n= 7
n= 8
5.0
84.55
1816.61
14.96
2.04
0.73
0.41
0.23
0.47
5.0
84.28
1816.73
14.71
1.77
0.57
0.61
0.28
0.59
5.0
82.83
1816.82
14.45
1.23
0.86
0.32
0.38
0.28
5.0
84.41
1817.14
14.68
1.81
0.44
0.57
0.19
0.26
5.0
85.07
1817.43
14.60
2.10
0.68
0.74
0.07
0.75
PDI FFT Algorithm Results
Avg Curr
n= 1
n= 2
n= 3
n= 4
n= 5
n= 6
n= 7
n= 8
5.0
84.04
1817.33
14.62
1.92
1.00
0.59
0.32
0.39
5.0
84.20
1817.05
14.45
1.92
0.66
0.45
0.39
0.42
5.0
85.84
1817.36
16.11
2.16
0.99
0.60
0.09
0.66
5.0
83.69
1816.81
15.36
1.91
1.18
0.16
0.30
0.34
5.0
84.14
1817.32
15.00
1.77
1.21
0.06
0.37
0.22
CAMAC FFT Algorithm Results
n= 9
n = 10
n = 11
n = 12
n = 13
n = 14
n = 15
n = 16
n = 17
0.20
0.34
0.23
0.54
0.03
0.21
0.19
0.62
0.60
0.40
0.06
0.39
0.18
0.15
0.38
0.40
0.20
0.23
0.11
0.18
0.35
0.24
0.25
0.11
0.09
0.20
0.34
0.31
0.07
0.34
0.31
0.18
0.26
0.30
0.28
0.38
0.33
0.22
0.42
0.06
0.30
0.20
0.26
0.27
0.41
PDI FFT Function Results
n= 9
n = 10
n = 11
n = 12
n = 13
n = 14
n = 15
n = 16
n = 17
0.28
0.02
0.30
0.52
0.41
0.25
0.33
0.90
0.46
0.29
0.05
0.26
0.11
0.24
0.14
0.25
0.38
0.35
0.19
0.46
0.37
0.23
0.26
0.50
0.23
0.59
0.36
0.24
0.25
0.13
0.17
0.43
0.25
0.25
0.40
0.34
0.18
0.38
0.18
0.50
0.28
0.37
0.13
0.48
0.23
Table 6-V: Amplitude Comparison between CAMAC and PDI
Algorithms
The PDI and CAMAC algorithms used to compute the harmonics are shown in
Appendix A, Figures A-I and A-2, respectively.
60
6.5
Probe Reproducibility Testing
Data was collected on each of the five rotating coil probes currently being used in
the Magnet Measurement Facility at JLab. The collected data comprised the
average of five discrete forward rotations, similar to the method used in the
CAMAC data acquisition system. Data collection in the forward direction only
was chosen to eliminate any backlash error induced in the motor to probe linkage.
The magnet known as QB103, a six inch long laminated quadrupole with a two
inch bore, was used for each probe measurement. The EPICS control system was
used to cycle hysteresis and set the magnet current at 5 Amps at the beginning of
each measurement day. The magnet current was monitored over the course of the
day to ensure that it remained constant. The current was not cycled between
measurements, but was only cycled at the beginning of each morning. Table 6-VI
shows the reproducibility of each probe as a percentage of the average amplitude.
Probe ID
P1A
P2A
P1A
P2A
P1C (100 turns)
P2B (90 turns)
Coil 1 Short Probe - 50 turns Outside Coil
N=2
Run 1
Run 2
Run 3
Run 4
Run 5
0.041%
0.029%
0.070%
0.312% 0.418% 0.215%
Coil 2 Short Probe - 100 turns inside Coil
0.120%
0.115%
% Dev from
average
% Dev from
average
% Dev from
average
% Dev from
average
0.029%
0.052%
0.056%
0.046%
0.108%
0.189% 0.238%
Coil 4 Long Probe
0.095%
0.201%
0.100%
% Dev from
average
% Dev from
average
0.041%
0.046%
0.063%
0.061%
0.094%
0.041%
0.042%
0.147%
0.138%
0.221%
0.217%
0.385%
Table 6-VI Initial probe reproducibility at 5 A as a percentage of
signal strength. This table summarizes the reproducibility of each probe
as a percentage of the average amplitude. Because the reproducibility of
each probe does not meet the desired specification of 0.01%, additional
system optimization was needed.
61
6.6
Measurement Uniformity Across Gain Settings
To test the gain deviations that might result from gain offsets, measurements of
an input signal received from the function generator were taken with a series of
different gain settings. The function generator signal was configured to mimic a
quadrupole as discussed in the previous sections. The signal was then connected
directly to the PDI and a series of measurements were made. Two signal levels
were picked for the test. The first setting corresponded to an amplitude of 200
mV, to simulate an operating condition common to a variety of quadrupole
measurements. The second setting required that a 50dB attenuator be connected
in series with the function generator and that the PDI and the signal amplitude be
set to 1V. This reduced the output signal to a level that would allow the gain to
be set at 1000 without causing an overflow. The first series of measurements
resulted in a strength term of ~19967 uV*s and a repeatability error of ~2 uV*s
(0.01%) over the 20 measurement sample. The effects from varying gain settings
are shown in Figure 6-III.
62
Difference from First Gain = 20 Measurement
(uV*s )
3
2.5
2
Gain 20
1.5
Gain 10
1
Gain 5
Gain 2
0.5
Gain 1
0
-0.5
-1
0
2
4
6
8
10
12
Measurement Number
Figure 6-III Measurement repeatability across gain settings 1 – 20.
Each curve represents 10 measurements at a different gain setting. For all
gain settings, measurement noise fell below 2.5 uV*s.
The reduced signal level (the second setting) was used to mimic a bucked
configuration. The resultant magnitude of the quadrupole term was ~323 uV*s
and was used to evaluate the overall repeatability of the higher gain settings as
shown in Figure 6-IV. At the lower setting, the overall repeatability error of the
system remained at ~1.5 to 2 uV*s, even though the magnitude of the quadrupole
term was much smaller. The effect of varying the gain settings was smaller than
the standard repeatability error of the system.
63
3
Difference from First Gain = 1000
Measurement (uV*s)
2.5
2
Gain 1000
1.5
Gain 500
1
Gain 200
Gain 100
0.5
Gain 50
0
-0.5
-1
0
2
4
6
8
10
12
Measurement Number
Figure 6-IV Measurement repeatability across gain settings 50 -1000.
Each curve represents 10 measurements at a different gain setting. Higher
gain settings are used for weaker signals therefore the deviations are not
expressed as a percentage and can be compared with the data from lower
gain setting. For the higher gain settings, measurement noise fell below 2
uV*s.
It is important to note that the gain must be set correctly to minimize the drift
associated with the system over time. The maximum expected drift per month is
(200uV / gain). In the instance of a quadrupole measurement that produces a
maximum induced voltage of ~1.61 mV, at a gain setting of 1, the drift over one
month could be ~12.4%. For the same signal, the gain can be set to 50 resulting
in a 4uV maximum drift per month or ~0.25% drift related deviation.
64
6.7
Individual Position Uniformity
For each 360 degree probe cycle, data is integrated at 200 discrete points (every
1.8 degrees). A series of 20 measurements were taken and the average integral at
each discrete angular position was calculated.
Each measurement was then
subtracted from the average and plotted to compare integral inconsistencies at
each angular position.
The first measurements were made using the normal motor velocity of 40,000
steps/sec and acceleration of 80,000 steps/sec2. The resultant differential error at
these speed settings is ~.0002 uV*s over the measurement sample, as shown in
the top most graph of Figure 6-V.
A calculation was made to verify the probe was spinning at a constant velocity
during the measurement. This was done to create better system stability noting
that integration is independent of rotation velocity.
A new acceleration and
velocity was calculated using
 2  02

2(  0 )
where

(6.1)
is the acceleration required to reach the desired velocity,
desired angular velocity and
0
if the initial angular velocity, and
 is
 - 0
the
is the
change in radial angle. This was done to try to improve the probe stability by
attaining the constant velocity from rest under the constraint of the fixed angle
between the initial probe location and the measurement trigger.
65
Data was taken with the updated velocity and acceleration. Again the differential
data was compared at each position and plotted. The data appeared noisier in the
beginning of the run as shown in the middle graph of Figure 6-V.
To further stabilize the system and minimize noise associated with the motor
speed, the velocity and acceleration were reduced to 20,000 steps/sec and 20,000
steps/sec^2, respectively, and the measurement was re-taken. This configuration
led to the best measurement resolution as shown in bottom most graph in Figure
6-V.
66
Figure 6-V uV*s deviations from average of all runs every 1.8 degrees.
The first graph shows the results of the measurement using a motor
velocity of 40,000 steps/s and acceleration of 80,000 steps/s2. The second
graph shows the results of the measurement using a motor velocity of
30,000 steps/s and acceleration of 90,000 steps/s2 which would allow the
motor to reach constant velocity before the measurement trigger. The
final graph shows the results of the measurement using a slow motor
velocity of 20,000 steps/s and acceleration of 20,000 steps/s2 resulting in
greater stability.
67
6.8
Signal Loss – Probe Input to PDI Input
Each probe location was evaluated to verify that an accurate induced voltage
would be transferred to the integrator during the measurement process. To do this
three DC voltages were applied to the each probe location to simulate the induced
voltage ranges encountered when measuring multipole magnets. For each voltage
value and each input location the voltage was read by a 6 digit resolution DVM at
the probe input and multiplexer output of the signal chassis box. Table 6-VII
shows the results of the signal loss measurements.
Probe
Input
Voltage (V)
0.010285
0.100554
1.004034
Multiplexer Output
Voltage (V)
0.010276
0.100553
1.004035
Differences
(x 10-3 V)
0.0095
0.0009
-0.0009
Coil 2
0.010299
0.100554
1.004049
0.010297
0.100561
1.004045
0.0015
-0.0073
0.0031
Coil 4
0.010331
0.100586
1.004160
0.010327
0.100591
1.004157
0.0037
-0.0049
0.0028
Coil 1
Table 6-VII Voltage signal loss from probe input to PDI input.
Voltage values at the probe input and multiplexer chassis output were read
by a 6 digit digital voltmeter. The differences are expressed relative to the
desired absolute strength specification of 1 part per 1000.
To verify the stability of the signal through the system a signal analyzer was used.
A function generator was set up to supply the 1.596 Hz, 300mV amplitude test
signal used throughout the commissioning. The signal was applied to the Coil 4
68
probe input location and supplied to the spectrum analyzer through the chassis
box multiplexor output. The electrical signal was found to be very stable and
noise free.
6.9
Short and Long Term Repeatability
Measurements were taken to determine the stability of the system over extended
time periods. A function generator was again used to create a reference signal
using the five continuous cycle method.
One hundred measurements were
collected over an eight hour period to determine long term system repeatability
and to quantify any system drift that may be occur during a daily measurement
process. Initial tests showed large fluctuations in the data over the measurement
period. It was observed that the fluctuations were correlated with the opening and
closing of the large bay door adjacent to the magnet measurement facility as
shown in Figure 6-VI.
69
0.040%
Garage Door
Opened - Function
Generator in Main
Lab
Deviation from Average (%)
0.030%
0.020%
0.010%
Garage Door
Opened- Function
Generator in Office
0.000%
-0.010%
-0.020%
-0.030%
0
1
2
3
4
5
6
7
8
Time (Hours)
Figure 6-VI Long term repeatability using generated signal
The function generator signal was plugged directly into the PDI unit for
the test. A direct correlation was noted between the large bay door of the
test lab opening, which results in a temperature change, and signal
amplitude. Additional testing showed that the measurement error was tied
directly to the function generator and not the integrating unit. Horizontal
lines represent the desired error bar for the measurement.
The function generator was moved into a semi-controlled temperature
environment and the signal reattached to the integrator input.
Further tests
verified that the majority of the measurement errors were tied to temperature
fluctuations affecting the function generator.
Cabling restraints make it
unreasonable to move the integrator into a controlled room to evaluate any error
resulting from temperature changes to it.
Figure 6-VII shows the reference signal’s deviation from average over the
sampling period. Over the entire measurement period, the system appears quite
70
stable. System drift fluctuations may be tied to slight temperature, pressure, and
humidity changes in the test lab and semi-controlled room as the overall building
environment changes.
Short term repeatability can be determined from the maximum error of any four
hour period during the long term study. The largest changes occur during the first
4 hours of the measurement and show that the error for both short and long term
repeatability is equivalent.
0.040%
Deviation from Average (%)
0.030%
0.020%
0.010%
0.000%
-0.010%
-0.020%
-0.030%
-0.040%
0
1
2
3
4
5
6
7
Time (Hours)
Figure 6-VII Long term system repeatability using temperature
controlled generated signal (8 hours). The function generator was
moved into an enclosed room before any data was collected. The signal
was plugged directly into the PDI unit using an extended BNC connection
cable. Horizontal lines represent the desired measurement error bar and
show greater measurement stability.
71
8
6.10
Optimized Repeatability
Additional tests were conducted to compare system reproducibility levels after
system optimization. Optimization factors found to effect long term measurement
repeatability include probe speed, number of cycles, forward only cycling, and
gain setting.
Prior to system optimization, a series of measurements were taken to establish
system repeatability using probe P2B on QX075. For this test, hysteresis was run
prior to the measurement sequence and the current was set to 5 Amps. Figure 6VIII shows the repeatability of 20 measurement points for seven separate runs.
72
0.10%
Deviations from individual average (%)
0.08%
0.06%
0.04%
0.02%
0.00%
-0.02%
-0.04%
-0.06%
-0.08%
-0.10%
0
2
4
6
8
10
12
14
16
18
20
9
10
Measurement Number
0.10%
Deviations from 5 Run Average (%)
0.08%
0.06%
0.04%
0.02%
0.00%
-0.02%
-0.04%
-0.06%
-0.08%
-0.10%
1
2
3
4
5
6
7
8
Measurement Number
Figure 6-VIII Initial versus Optimized Repeatability for Probe P2B.
Probe P2B was used in the optimization study because it showed the worst
reproducibility of all probes that were tested. The top graph shows a short
term measurement using un-optimized techniques. The bottom graph
shows the results of the same measurement using a velocity of 20,000
steps/s and acceleration of 20,000 steps/s2, 7 averaged forward cycles, and
a higher gain setting.
73
After optimization, Probe P2B was used to measure another available QX magnet,
known as JLab QX061, at 5 Amps as was done in previous tests. The magnet was
allowed to stabilize at 5 Amps for 20 minutes prior to testing. The magnet current
was not cycled during the measurement process. This time, the PDI gain was set
at 50 and the motor acceleration and velocity set to 20,000 steps/s. There were
seven forward cycles averaged for each measurement.
0.04%
Deviation from Average (%)
0.03%
0.02%
0.01%
0.00%
-0.01%
-0.02%
-0.03%
-0.04%
0
1
2
3
4
Time (Hours)
Figure 6-IX Short term optimized repeatability of system using
rotating coil (P2B).
This graph shows the four hour measurement
repeatability using a velocity of 20,000 steps/s and acceleration of 20,000
steps/s2, 7 averaged forward cycle, and a higher gain setting. Using the
optimization techniques, the measurement error approaches the required
specification.
The optimized repeatability shown in Figure 6-IX for probe P2B is clearly more
stable than the repeatability shown in the top graph in Figure 6-VIII which was
measured without optimized methods. Additional testing is recommended to
74
establish the optimized repeatability specifications for each probe using the
optimized measurement protocol.
75
6.11
Absolute Strength
Additional equipment will be needed to attain the desired absolute strength
specification.
It is necessary to accurately calibrate the measurement probe
regularly before measurements if this specification is to be met. Integrator drift,
integrator noise with respect to gain, and environmental effects on the probe all
contribute to the absolute strength calibration fluctuations.
A calibration
technique must be developed to overcome these issues.
Two probes, P1A and P1C, have the repeatability needed to reach the desired
absolute strength specification of 0.1% if they were properly calibrated. To reach
this level of accuracy each probe would have to be calibrated to the same level of
accuracy. Determination of the quadrupole probe constant is complicated by the
difficulties of measuring the quadrupole field with a Hall probe. One way to
correctly determine the quadrupole probe constant is to calculate the dipole
constant of the probe and then use the dipole constant to determine the quadrupole
constant for the rotating probe[12].
The theoretical value of the dipole constant Kd can be calculated by
(6.2)
Kd 
76
1
.
N (r1  r2 )
Where N is the number of turns and r1 and r2 are the coil radii. Kd can be
determined through field measurements by comparing the field integral
determined through NMR and Hall probe measurements with a rotating coil probe
integral along the same axis. The error in the integrated emf and integrated field
integral are used to determine the dipole constant uncertainty.
Once Kd is
determined, the quadrupole probe constant Kq can be calculated using
Kq 
2K d
.
(r1  r2 )
(6.3)
The uncertainty in Kq is determined by the uncertainties in both Kd and the coil
radii r1 and r2. The total uncertainty in Kq can be calculated by the quadrature of
the uncertainties by
K q
Kq
 (
K d 2
r 2
)  2(
)
Kd
r1  r2
where r is the uncertainty in the coil radii r1 and r2 and
(6.4)
K d
Kd
is the dipole
constant error. For well constructed rotating probes, r is usually on the order of
0.2 mm. Therefore, to minimize the total uncertainty in both the dipole and
quadrupole probe constant calculations, a probe with a very large r1 to r2 ratio
could be constructed. Selecting a large radii ratio will minimize the error when
calculating the theoretical dipole constant. This probe could then be used to
77
determine the absolute strength of a large aperture quadrupole calibration magnet
which would also need to be constructed. Once the absolute strength of the
calibration magnet is accurately determined, a smaller radius rotating probe could
be used to measure the same magnet. This measurement could then be used to
accurately determine Kq for the smaller probe.
Currently, there is no means by which the absolute strength specification desired
for the FEL upgrade can be reached.
Efforts to have the FEL quadrupoles
measured at off-site locations to 0.1% using Hall probes have been unsuccessful.
The current measurement stand can not physically accommodate the large
aperture magnets and the resultant signal would be too large for the PDI integrator
to process. In theory, the strength specification can be met but will require
specialized magnets and a new measurement system to be built to meet that goal.
78
7
Conclusions
System repeatability of the PDI data acquisition unit itself is at worst 0.01% over
periods of eight hours or less using a reference signal that mimics a quadrupole
magnet when the reference signal generator is placed in a controlled temperature
environment.
Though the repeatability of the measurements done using five discrete rotations
was slightly noisier than the repeatability of the measurements done using five
continuous rotations at all three coil input locations, in general, the input location
of the reference signal, direct connection to the PDI unit or any coil location at the
probe junction, did not significantly affect the system repeatability.
Measurements using a spectrum analyzer and 6 digit digital voltmeter show that
the signal through the system is very stable and virtually noise free.
Simulations using System View and MatLab suggest better FFT results are
obtained using measurement data from five continuous rotations instead of five
discrete, averaged rotations.
Probe rotation speed plays a significant role in the reproducibility of the
measurement system. Slower speeds result in better measurement repeatability.
79
FFT routines used by the PDI stand are equivalent to routines used in the existing
CAMAC stand routine.
The overall system repeatability prior to optimization for the four measurements
probes used in the magnet measurement facility was measured on a QB magnet at
5 Amps and is currently specified as:
1. P1A – 0.1%
2. P1C – 0.1%
3. P2A – 0.4%
4. P2B – 0.4%
Additional testing after measurement system optimization result in probe P2B
having a four hour repeatability of 0.06% over multiple short term tests.
Comprehensive testing of all probes should be conducted to quantify the
resolution of each in the optimized measurement system configuration.
80
Appendix A MetroLab PDI 5025
A MetroLab PDI 5025 digital integrator is used to collect sample data from the
measurement coils. The main functions of the PDI 5025 are:

Integration of the pick-up coil voltage relative to the internal time base, or
mechanical preset increments (angular or linear) generated by the motion of the
pick-up coil.

Driving a DC motor, for rotational or linear motion of the coil, and
resolving encoder input for position monitoring.

Transmission of the results to the external computer [6].
The PDI 5025 is a digital integrating magnetic measurement device which uses an
external software program written in LabWindows CVI to control and coordinate
additional hardware providing a complete magnetic measurement data acquisition
system. The rotating coils are connected through a multiplexer to the integrator
by a 4 pin lemo connector and cable. While the PDI 5025 has an internal clock
and can integrate the magnetic field with respect to the time derivative of the
measured signal, the JLab system uses an external motor with an optical encoder.
The integrator circuit for the MetroLab PDI 5025 is shown in Figure 2. The
encoder signal is fed to the integrator while the motor itself is controlled
independently. A synchronizing TTL pulse signal is then used to indicate the
beginning of the measurement. The PDI 5025 then collects data and stores it in
an output buffer block to be read by the software. The signal is fed to the PDI
81
unit and is conditioned by a preamplifier. The signal gain must be set so that the
voltage never exceeds the +/- 5V limits. The proper gain level is set, either
manually or through software, to an appropriate level for the measurement prior
to beginning the measurement sequence.
Next, the measurement commences
and the input voltage is shifted by +5 volts to ensure the resulting voltage will be
positive. The resultant signal is then transmitted to the voltage to frequency
converter, which operates between 0 – 10 V. The pulse trains created by the V/F
are sent to active counters to be stored. When triggered, the contents of the
counters are moved to the PDI microprocessor. During the transfer period, a
separate set of counters resume counting. The PDI microprocessor then does
calculations on the sampled data based on position and derives a result expressed
in V*s (volt seconds or webers). At the end of the measurement block, the
microprocessor transmits the results to the host computer. Figure A-I shows a
schematic of the MetroLab digital integrator.
82
Figure A-I Schematic of MetroLab’s PDI 5025 Integrator Circuit
83
Appendix B FFT Algorithm Comparisons
/*
/*
*
*
*
*
*
*
*/
**************************************************************
*/
vtcoil_jlab_fft
This function is used to compute the amplitude and phase angle of the
voltage samples in the same manner as the JLab CAMAC measurement system.
Ken Baggett
9/22/2004
void vtcoil_jlab_fft(double vt[], int num_har, int currentIndex, int coilIndex)
{
int i,j;
double dx;
double* xsum;
double* ysum;
double* amplitudes;
double* phases;
double pi = 3.1415927;
double theta = 0.;
double width = 0.;
// 1 count every 3.6 degrees for 100 counts
dx = (360. / vtcoil_param.num_samp_per_rev) * (pi / 180.);
xsum = (double*) malloc((num_har+1) * sizeof(double));
ysum = (double*) malloc((num_har+1) * sizeof(double));
amplitudes = (double*) malloc((num_har+1) * sizeof(double));
phases
= (double*) malloc((num_har+1) * sizeof(double));
for(i=0; i < num_har; i++)
{
xsum[i] = 0.;
ysum[i] = 0.;
}
//printf("\n=====VT Coil Readings in uV*S=======\n");
// convert from V-S to uV-S
for(i=0; i <= vtcoil_param.num_samp_per_rev; i++)
{
vt[i] = vt[i]* 1000000.0;
}
// do the integrals
for(i=0; i <= vtcoil_param.num_samp_per_rev; i++)
{
if(i == 0 || i == vtcoil_param.num_samp_per_rev)
{
width = dx / 2.;
}
else
{
width = dx;
}
theta = i * dx;
for(j=0; j < num_har; j++)
{
xsum[j] += (vt[i] * cos((j * theta)) * width);
ysum[j] += (vt[i] * sin((j * theta)) * width);
}
}
// Normalize the integrals and calc Amplitude and Phase
xsum[0] /= (2. * pi);
ysum[0] /= (2. * pi);
amplitudes[0] = xsum[0];
phases[0] = -90.;
for(i = 1; i <= num_har; i++)
{
xsum[i] /= pi;
ysum[i] /= pi;
amplitudes[i] = sqrt (pow(xsum[i],2.) + pow (ysum[i],2.));
if(ysum[i] == 0.)
{
printf("Zero Intergal");
return;
}
phases[i] = -(atan2 (xsum[i], ysum[i]) / i);
phases[i] *= (180. / pi);
}
// now store the values for future writeout
for(i=0; i < num_har; i++)
{
fftAmp[i][currentIndex][coilIndex] = amplitudes[i];
fftPhase[i][currentIndex][coilIndex] = phases[i];
}
}
Figure B-I CAMAC FFT Algorithm
84
/*
**************************************************************
*/
/*
*
vtcoil_altFFT
*
This function is used to compute the amplitude and phase angle of the
* voltage samples.
*
Amplitude is given by the magnitude of the normal and skew components
*
Phase of each harmonic is given by the angle of the harmonic normalized to
*
the period.
*
*
Ken Baggett
*
10/25/2005
*/
void
vtcoil_altFFT(double
vthar_re_ave[],
currentIndex, int coilIndex, int num_str_har)
{
double
vthar_im_ave[],
int
num_har,
int
int i;
double amplitudes[50];
double phases[50];
double temp;
amplitudes[0] = 0.0;
phases[0] = -90.;
fftAmp[0][currentIndex][coilIndex] = amplitudes[0];
fftPhase[0][currentIndex][coilIndex] = phases[0];
// now store the values for future writeout
for(i=1; i <= num_har; i++)
{
// Calculate the harmonic strength
amplitudes[i] = sqrt( pow(vthar_re_ave[i], 2) + pow(vthar_im_ave[i], 2) );
amplitudes[i] *=1000000.0;
// Calculate the south pole angle
phases[i] = - (atan2(vthar_im_ave[i], vthar_re_ave[i]) + 3.1415927 / 2) / i;
if (phases[i] > 3.1415927 / i)
phases[i] = phases[i] - 2. * 3.1415927 / i;
if (phases[i] < -3.1415927 / i)
phases[i] = phases[i] + 2. * 3.1415927 / i;
// Convert the south pole angle to degrees
phases[i] = phases[i] * 180. / 3.1415927;
fftPhase[i][currentIndex][coilIndex] = phases[i];
fftAmp[i][currentIndex][coilIndex] = amplitudes[i];
}
}
Figure B-II PDI FFT Algorithm
85
Appendix C Quadrupole Amplitude Data – Coils 1,2,4
Coil 1 (uV*Sec)
P2A
P2A
P2A
P2A
P2A
5 Cycles
5 Cycles
5 Cycles
5 Cycles
5 Cycles
Forward Only
Forward Only
Forward Only
Forward Only
Forward Only
Measurement #
PRP2A011.fft
PRP2A012.fft
PRP2A013.fft
PRP2A014.fft
PRP2A015.fft
1
2
3
4
5
6
7
8
9
10
14457.520
14427.982
14464.659
14440.012
14422.583
14425.431
14439.798
14444.892
14437.822
14467.625
8217.441
8214.719
8205.696
8214.880
8209.382
8209.819
8216.783
8197.769
8232.071
8205.618
8240.319
8222.648
8223.235
8231.669
8236.182
8236.914
8236.992
8235.400
8224.809
8230.126
8233.880
8235.853
8233.454
8236.777
8228.210
8234.438
8227.038
8231.942
8226.901
8230.144
8234.836
8231.632
8231.822
8235.318
8227.317
8232.204
8233.455
8236.806
8231.007
8234.465
Average
Min
Max
Difference
14442.8324
14422.5830
14467.6250
45.04200
8212.4178
8197.7690
8232.0710
34.30200
8231.8294
8222.6480
8240.3190
17.67100
8231.8637
8226.9010
8236.7770
9.87600
8232.8862
8227.3170
8236.8060
9.48900
% Deviation
0.312%
0.418%
0.215%
0.120%
0.115%
Table C-I Measurement amplitude repeatability for probe P2A-Coil 1
86
Coil 2 (uV*Sec)
P2A
P2A
P2A
P2A
P2A
5 Cycles
5 Cycles
5 Cycles
5 Cycles
5 Cycles
Forward Only
Forward Only
Forward Only
Forward Only
Forward Only
PRP2A016.fft
PRP2A017.fft
PRP2A018.fft
PRP2A022.fft
PRP2A023.fft
8171.222
8171.141
8155.819
8157.134
8158.787
8158.499
8160.852
8156.466
8161.070
8162.886
8168.985
8165.879
8165.739
8166.342
8166.219
8161.559
8166.035
8157.513
8176.934
8172.542
8172.499
8169.573
8171.299
8169.731
8168.465
8170.338
8168.911
8170.029
8166.266
8174.057
8164.141
8165.550
8166.825
8173.038
8170.707
8166.661
8163.027
8177.142
8160.686
8167.355
8174.368
8173.341
8172.813
8173.417
8167.962
8168.707
8168.187
8169.501
8168.171
8166.161
Average
Min
Max
Difference
8161.3876
8155.8190
8171.2220
15.40300
8166.7747
8157.5130
8176.9340
19.42100
8170.1168
8166.2660
8174.0570
7.79100
8167.5132
8160.6860
8177.1420
16.45600
8170.2628
8166.1610
8174.3680
8.20700
% Deviation
0.189%
0.238%
0.095%
0.201%
0.100%
Measurement
#
1
2
3
4
5
6
7
8
9
10
Table C-II Measurement amplitude repeatability for probe P2A-Coil2
87
Coil 1 (uV*Sec)
P1A
P1A
P1A
P1A
P1A
5 Cycles
5 Cycles
5 Cycles
5 Cycles
5 Cycles
Forward Only
Forward Only
Forward Only
Forward Only
Forward Only
PRP1A001.fft
PRP1A002.fft
PRP1A003.fft
PRP1A004.fft
PRP1A005.fft
1816.150
1816.374
1816.633
1816.672
1816.644
1816.342
1816.450
1816.191
1816.540
1816.328
1816.463
1816.353
1816.667
1816.001
1816.279
1816.048
1816.027
1816.197
1815.902
1815.838
1816.206
1815.885
1816.024
1815.899
1816.315
1816.621
1816.120
1816.200
1816.285
1815.926
1817.141
1816.717
1816.782
1816.914
1816.726
1816.619
1816.666
1816.663
1817.144
1816.711
1816.357
1816.101
1816.236
1816.634
1816.301
1815.653
1815.631
1815.369
1816.493
1816.296
Average
Min
Max
Difference
1816.4324
1816.1500
1816.6720
0.52200
1816.1775
1815.8380
1816.6670
0.82900
1816.1481
1815.8850
1816.6210
0.73600
1816.8083
1816.6190
1817.1440
0.52500
1816.1071
1815.3690
1816.6340
1.26500
% Deviation
0.029%
0.046%
0.041%
0.029%
0.070%
Measurement
#
1
2
3
4
5
6
7
8
9
10
Table C-III Measurement amplitude repeatability for probe P1A-Coil 1
88
Coil 2 (uV*Sec)
P1A
P1A
P1A
P1A
P1A
5 Cycles
5 Cycles
5 Cycles
5 Cycles
5 Cycles
Forward Only
Forward Only
Forward Only
Forward Only
Forward Only
Measurement
#
PRP1A006.fft
PRP1A007.fft
PRP1A008.fft
PRP1A009.fft
PRP1A010.fft
1
2
3
4
5
6
7
8
9
10
1750.485
1750.531
1750.155
1750.803
1750.392
1750.160
1750.149
1750.192
1750.082
1750.365
1750.861
1751.595
1751.467
1751.253
1751.611
1751.630
1751.777
1751.596
1751.353
1751.203
1751.639
1751.979
1751.572
1751.538
1751.346
1751.318
1751.502
1751.353
1751.598
1750.993
1751.520
1751.983
1751.938
1751.767
1751.242
1751.861
1751.925
1751.940
1752.048
1751.932
1751.667
1751.454
1750.817
1750.466
1750.679
1751.419
1750.404
1749.774
1750.906
1750.550
Average
Min
Max
Difference
1750.3314
1750.0820
1750.8030
0.72100
1751.4346
1750.8610
1751.7770
0.91600
1751.4838
1750.9930
1751.9790
0.98600
1751.8156
1751.2420
1752.0480
0.80600
1750.8136
1749.7740
1751.6670
1.89300
% Deviation
0.041%
0.052%
0.056%
0.046%
0.108%
Table C-IV Measurement amplitude repeatability for probe P1A-Coil 2
89
Coil 4 (uV*Sec)
P2B
P2B
P2B
P2B
P2B
5 Cycles
5 Cycles
5 Cycles
5 Cycles
5 Cycles
Forward Only
Forward Only
Forward Only
Forward Only
Forward Only
Measurement
#
PRP2B001.fft
PRP2B002.fft
PRP2B003.fft
PRP2B004.fft
PRP2B005.fft
1
2
3
4
5
6
7
8
9
10
18789.212
18761.696
18764.149
18781.217
18773.138
18764.773
18768.116
18770.797
18780.203
18767.085
18771.829
18770.157
18756.591
18759.289
18772.288
18766.121
18775.149
18764.202
18770.462
18782.459
18772.203
18768.459
18779.026
18770.494
18772.659
18778.973
18737.560
18760.130
18768.927
18752.353
18765.126
18750.299
18755.136
18765.736
18762.051
18745.686
18786.321
18761.813
18756.482
18762.011
18767.200
18769.684
18771.466
18762.874
18781.481
18789.627
18747.359
18757.169
18719.624
18717.387
Average
Min
Max
Difference
18772.0386
18761.6960
18789.2120
27.51600
18768.8547
18756.5910
18782.4590
25.86800
18766.0784
18737.5600
18779.0260
41.46600
18761.0661
18745.6860
18786.3210
40.63500
18758.3871
18717.3870
18789.6270
72.24000
%
Deviation
0.147%
0.138%
0.221%
0.217%
0.385%
Table C-V Measurement amplitude repeatability for probe P2B-Coil 4
90
Coil 4 (uV*Sec)
P1C
P1C
P1C
P1C
P1C
5 Cycles
5 Cycles
5 Cycles
5 Cycles
5 Cycles
Forward Only
Forward Only
Forward Only
Forward Only
Forward Only
Measurement
#
PRP1C001.fft
PRP1C002.fft
PRP1C003.fft
PRP1C004.fft
PRP1C005.fft
1
2
3
4
5
6
7
8
9
10
4921.630
4920.738
4921.121
4921.692
4921.000
4923.179
4923.005
4921.997
4923.822
4923.482
4921.423
4922.893
4923.679
4923.373
4924.418
4924.171
4924.223
4922.944
4923.932
4923.653
4921.237
4920.726
4923.079
4924.365
4924.255
4923.671
4924.124
4924.594
4924.606
4925.336
4923.656
4924.291
4924.048
4925.659
4924.703
4924.551
4925.271
4924.744
4923.847
4923.952
4923.824
4924.298
4924.220
4924.031
4924.183
4923.314
4924.210
4923.995
4922.220
4922.471
Average
Min
Max
Difference
4922.1666
4920.7380
4923.8220
3.08400
4923.4709
4921.4230
4924.4180
2.99500
4923.5993
4920.7260
4925.3360
4.61000
4924.4722
4923.6560
4925.6590
2.00300
4923.6766
4922.2200
4924.2980
2.07800
%
Deviation
0.063%
0.061%
0.094%
0.041%
0.042%
Table C-VI Measurement amplitude repeatability for probe P1C-Coil 4
91
Appendix D Expert Control Panels for the PDI Software
PDI Expert Panel
The PDI expert control panel is used to test and troubleshoot the MetroLab digital
integrator. For JLab magnet measurement, encoders are used to integrate the
magnet flux based on position. To view the positional information, the PDI
encoder counts can be displayed by pressing the “Aquire Counts” button. The
“Zero Counter” button is used to reset the PDI encoder count index. High level
commands can be entered into the “Send Command” edit box and sent to the PDI
to test communications.
Additionally, the “TestRun” button can be used to test a single probe rotation
based on time or position. Different types of run testing can be configured by
setting specific parameters at the beginning of the “TestRun” function. The PDI
Expert Panel is shown in Figure D-I.
92
Figure D-I PDI Expert Control Panel
Three Piece Ramp Expert Panel
The Three Piece Ramp expert panel is used to control the FP-AIO analog output
control using a defined ramp protocol developed during the PDI commissioning.
The analog output from the FP-AIO controller is connected to a Danfysik power
supply. The Danfysik reads the input voltage from the FP-AIO controller and sets
its output current using 10 V (input) = 20 A (output).
From this panel, the user can change ramp speeds, hysteresis limits, and magnet
currents.
93
Figure D-II Three Piece Ramp Expert Panel
Motion Control Expert Panel
The motion control expert panel is used to reset or troubleshoot the measurement
stand motor. The panel can be used to send the motor to various positions and
cycle the motor between motor increments. When sending the motor to it’s home
position, the motor will rotate towards zero until the home limit switch is
activated. At this point, the motor and encoder positions are rest to zero.
A panel button provides an easy means to launch National Instruments
Measurement and Automation Explorer software for advanced troubleshooting
and configuration control.
94
Figure D-III Motion Control Expert Panel
Epics Control Expert Panel
The Epics control panel can be used to control the current supplied to the magnet
by the trim rack.
The controller issues simple calls to EPICS which
independently sets the magnet current or runs hysteresis. A transducer is used in
the magnet test lab to verify the current to the magnet is correct.
95
Figure D-IV Epics Control
HP 34970A DAQ Expert Panel
The HP34970A control panel is used to read back the 22 channels on the
corresponding instrument. Currently, each channel can be configured to readback
temperatures using J type thermocouples, 2 wire RTDs, and voltages. The user
can select the desired channel configuration from a dropdown menu.
The
software will then write the configurations command to the instrument when
collecting data and save the configuration to an initialization file.
The
initialization file is read during program start up to load saved channel
configurations. The “Start Channel Update” button is used to read all channels
sequentially.
96
Figure D-V HP-34970A DAQ Switch Unit
97
Appendix E Measurement System Assembly Details
PDI Interface Module Micro Switches
The PDI unit can communicate with a host computer using either an IEEE 488
(GPIB) cable or a RS 232 serial connection. The decision was made to use the
IEEE 488 connection based on available hardware, speed of data transfer, and
GPIB code examples for the PDI module.
To configure the PDI for
communication a series of micro-switches must be set. Figure E-I shows the
configuration currently used in the JLab measurement system.
1
2
3
4
5
6
7
8
9
10
0
1
1
1
0
0
0
0
1
0
Figure E-I PDI 5025 Microswitch Configuration for GPIB
PDI Wiring
Encoder Wire
The encoder wire uses a 5 pin connection detailed in the PDI Manual (section
5.3). The current motor timing is A and B low when I is high. Also, the encoder
wire index must be left open.
98
Motor Wire
Currently not used. Motor is run externally by a National Instruments stepper
amplifier.
Coil Input Wire
The coil input wire uses a 4 pin connection detailed in the MetroLab PDI manual.
Currently the factory preset is being used with jumper J3 in place (balanced with
input impedance 2M Ohm).
Jumpers
Trigger Module
o J1 moved to 2-3 for Encoder ‘A’ channel negative polarity.
o J2 moved to 2-3 for Index negative polarity. (after testing: index not used,
use external sync)
Motor Amplifier
The National Instruments MID-7602 is a multi-axis power amplifier interface
used to power both the rotational and linear motors of the integrating system.
Axis 1 is used for the rotating coil stand. Axis 2 will be used for the translating
wire stand.
99
The MID-7602e is used along with the FlexMotion FW-7344 controller. The
Motor wiring configurations are:
Motor
1
Clear
2
Green
3
4
Red
5
Black
Encoder
Axis 1
Brown
Yellow
Red
Black
Orange
Blue
White
Green/Ground
1
2
3
4
5
6
7
8
PDI Encoder Tie in to Encoder Connector
1
2
3
4
5
6
7
8
Brown
Green
Grey
White
Note: Yellow Index wire must be disconnected when using the external trigger
Limits
Axis 1


Red = Forward
White = Reverse
100

Black = Home
1
2
3
4
5
6
Red
Black
White
Ground
Green
FW-7344
The FW-7344 motion controller is connected to the host computer using a
firewire connection. From the FW-7344, a NI stepper wire is used to deliver its
signals to the motor amplifier. Because the stepper connection wire does not
provide a path to any of the four DAC’s included on the FW-7344, an external
connection was added to the unit. This connection provides access to DAC
channel 3 (+/- 10V). A simple controller can be accessed from the startup screen
of the Jlab integrating software program (the “Control Analog Output” button).
Signal Chassis Box
A chassis box was constructed to allow the measurement system’s input signals to
be routed to a single output BNC connection. Signals can be connected in a
bucked configuration, as seen in Figure 6-I, inside the chassis box before being
routed to the multiplexer. From the BNC output, the signal is connected to the
PDI. The chassis box works in conjunction with the Keithley multiplexer and will
be able to handle inputs from up to four different measurement stands. Figure E-II
shows the signal breakout typical of each multiplexer bank.
101
Figure E-II Signal Chassis Box Wiring Breakout Configuration
102
Motion Control Setup for Motor
To configure the motion controller, import settings from the file menu.
A
working configuration is stored on the network drive in “m:\magtest\software
code\Flex Motion\ Saved Settings For Rotating Coil motor.xml”.
The FW-7344 motion controller is used in conjunction with the motor amplifier.
It is connected to the host computer using a Firewire connection. From the FW7344, a specialized NI motor communication cable is used to deliver its signals to
the motor amp. Phoenix connectors are used to connect the motor and encoder
wires to the amplifier. This allows signals to be sent and received between the
motor amplifier and the motor.
To reset the controller using Measurement and Automation Explorer setup the
Axis and Encoder configurations and then initialize the device as shown in
Figures E-III - E-IV.
103
Figure E-III IP MAX axis configuration for motor controller and encoder
104
Figure E-IVV MAX instrument initialization
Motor Setup
In a typical measurement rotation, the motor pointer will be positioned
approximately at the start position shown in the following diagram. The motor
will rotate clockwise (PDI Fwd) for the ‘+’ measurement and counter clockwise
for a reverse measurement. An example PDI command to set up a measurement
for 100 readings would be:

FWD: TRI,+,0/100,40
105
o
Meaning: Trigger the Reading Intervals, while the motor is moving PDI
forward, start at the 0 sync signal, take 100 reading, spaced 40 counts apart.

REV: TRI,-,4000/100,40
o
Meaning: Trigger the Reading Intervals, while the motor is moving PDI
Reverse, start at 4000 counts forward from the 0 sync signal, take 100 reading,
spaced 40 counts apart.
When using external sync the first sync received after the "IND,S" command will
be considered zero. Therefore, absolute position (in number of pulses) must be
set for the reverse measurement. In this case, we want to take 100 sample spaced
40 counts apart (4000 counts total). Because the forward measurement is taken
first, the reverse measurement is started at a 4000-count offset from the forward
measurement. Also, note that the motor will need to reach a constant velocity
prior to the sync signal in both the positive and reverse sense.
Jumper J2 must be put on pins 2-3 in the PDI rack 5140 Trigger slot and
"TRS,E,S" and "SYN,1" must have been sent.
106
Index
Start
 PDI Fwd
NOTE: The PDI
direction sense is
opposite of the
MAX rotation
direction.
PDI Rev 
Figure E-V IP Motor trigger and index location
Component Channels
Table E-I shows the current GPIB addresses used in the multipole measurement
system for communication.
Component
HP34970A
K7011
PDI5025
Boss Pwr Supply
GPIB Board
GPIB Channel
9
16
14
6
0
Table E-I GPIB instrumentation channels
107
GPIB ENET/100
The GPIB-ENET/100 is used to communicate GPIB commands through an
Ethernet connection. This functionality decreases cabling restrictions and adds
the ability to communicate with your GPIB instrument from any location in the
world that has an Ethernet connection. Each GPIB ENET/100 in the Magnet
Measurement group will have it’s own unique IP address and host name. These
attributes are used to set up the GPIB ENET/100. The GPIB ENET/100 requires
NI-488.2 (or higher) to be functional. When installing the software the user will
be required to remove older, incompatible GPIB drivers before installing NI488.2.
Once the software is installed you must hook up the GPIB ENET/100
power and Ethernet connections and configure the device.
The current GPIB ENET/100 device configuration is shown in Figures E-VI and
E-VII.
108
Figure E-V IP Configuration for GPIB ENET/100
Figure E-VII GPIB ENET/100 Setup
109
References
[1]
Wang, J.G. “Acceptance Tests of SNS Transfer Line Quadrupoles” ,
Spalatial Neutron Source, Technical Note NOTE MAG-112, 2003
[2]
Henrichsen, K.N. “Magnetic Field Measurements in Beam Guiding
Magnets”, Technical Note lhc-98-008, CERN accelerator laboratory,
Geneva Switzerland, 1998
[3]
Jain, A. Harmonics Coils , Spalatial Neutron Source academic training,
2003 Online: http://www.bnl.gov/magnets/staff/gupta/scmag-course/,
[4]
Harris, C.A. Cobb, J.K “Establishing the magnetic field of a solid pole
magnet to within +- 0.01%”, Stanford Linear Accelerator Center, Internal
Technical Document.
[5]
P. Schlabach, “Magnetic Measurements during Production of VLHC Low
Field Combined Function Magnets” , Fermilab, Technical Document TD01-007, 2001.
[6]
MetroLab PDI 5025 Operation Manual, MetroLab Instruments SA, 1993
[7]
N. Sclater, Wire & Cable for Electronics, McGraw Hill, New York, New
York, 1999
[8]
Tanabe, J. Conventional Magnet Design, US Particle Accelerator School,
Santa Barbara CA, 2003
[9]
L. Harwood, “Production Magnet Testing”, Jefferson Lab Technical Note
TN-0187, 1989
110
[10]
Hiatt, T.
“Some FEL UV Quadrupole Measurements” Jefferson Lab
Technical Note JLAB-TN-05-063 , 2005
[11]
Grenoble, C. “Configuring a PC as an EPICS Client” Jefferson Lab Tech
Note, JLAB-TN-02-029, 2005
[12]
Karn, J. Magnetic Measurements for the FET Recirculation Experiment,
Jefferson Lab Technical Note TN92-004, 1992.
111