PFC/JA-87-27
Magnetic Probes for Small Signal Detection
in a Large Background Field
Xing Chen
July 1987
Plasma Fusion Center
Massachusetts Institute of Technology
Cambridge, Massachusetts 02139 USA
Submitted to Review of Scientific Instruments.
Supported by United States Department of Energy Contract DE-AC02-78ET51013.
Abstract
High precision Hall probes and B-dot probes are described.
These
probes are designed to measure magnetic field changes of a few Gauss on a
time scale of 2 to 3 seconds in a 5 kG background field. A digital, drift free
sample and hold circuit is used with the Hall probe to subtract the vacuum
field.
1
Introduction
Magnetic probes are standard diagnostics in many experiments. 1' 2 When
applied to plasma equilibrium study, the probes are often designed to measure small magnetic field changes in a large background vacuum field. This
condition usually requires the probes to be accurate to 10-
4
to 10- 5 of the
total field. In order to achieve such accuracy, the large vacuum field must
be accurately subtracted.
In this paper, we describe a Hall probe and a B-dot probe system used
in the Constance B mirror experiment.'
These probes are designed to
measure magnetic field changes of a few Gauss on a time scale of 2 to 3
seconds in a 5 kG background field. The performance of these probes is
primarily dependent on their electronic circuits. In the Hall probe system,
a novel digital sample and hold circuit is used to store the vacuum field
information so that it can be accurately subtracted during the shot. This
digital circuit, with an essentially infinite time constant, avoids the problem of limited RC time which is characteristic of analog sample and hold
circuits. In the B-dot probe system, the signal is processed by a low drift
electronic integrator. The drifts of the integrators are on the order of one
tenth of a microvolt equivalent input noise level, corresponding to a magnetic field change of 0.02 G/sec when used with a B-dot probe of effective
area 0.05 m 2 . Both of the probe systems are reliable and easy to operate.
2
I.
Hall Probe
Our Hall probe system contains a Hall effect sensor and the electronic
circuits for signal processing. Fig. 1 shows a functional block diagram of
the system.
A Sprague UGN-3501M Hall effect integrated circuit is used as the
sensor. This monolithic sensor contains a Hall cell, differential amplifier,
differential emitter output, and a voltage regulator. It can be installed
inside half inch diameter tubing. The probe sensitivity is dependent on the
power supply voltage and the value of an external current limiting resistor.
The output signal is linear up to approximately 3 volts. One can adjust the
sensitivity according to the magnitude of the magnetic field to be measured.
The response of each probe needs to be calibrated individually.
The heart of the signal processing electronics is a sample and hold circuit constructed with a pair of analog-to-digital and digital-to-analog convertors. It provides a stable vacuum field reference to be subtracted from
the magnetic sensor during the shot so that small magnetic field changes
due to plasma currents can be distinguished. The digital sample and hold is
superior to an analog sample and hold in this application. In order to limit
the drift rate within 0.1 G/sec, the time constant of the sample and hold
has to be at least 104 seconds for a circuit which holds a few kilogauss signal. Such a long RC time is very difficult to achieve with an analog circuit
3
but poses no difficulty to a digital one. The A/D convertor (Analog Device
AD573) and the D/A convertor (Analog Device AD561) are connected in
a bipolar mode, which gives an input and output range of ± 5 Volts. The
sample and hold is controlled by an external gate on the A/D convertor.
The A/D converts the input voltage to a 10 bit digital signal when triggered, and stores it until the next trigger pulse. This digital signal is read
by the 10 bit D/A convertor and transformed back to an analog voltage
equal to the input voltage with an accuracy of ± 10 mV. The accuracy can
be improved by using 12, 14 or 16 bit convertors. The noise level at the
D/A convertor output is less than 0.2 millivolts, and it can be reduced to
about 10 OV after filtering.
During probe operation, the magnetic field signal sensed by the Hall
probe is carried differentially to the digital sample and hold input, with
an offset voltage of about 4 volts. It is converted to a signal proportional
to the total magnetic field by a low noise buffer amplifier. The output
from this amplifier is split into three paths, one to the total field output,
another to the input of the next stage differential amplifier, the third to
the digital sample and hold circuit which will store the vacuum field signal.
The sampling is controlled by a 1pts wide positive pulse from an external
pulse generator.
The conversion starts after the falling edge and takes
approximately 20ps. The differential amplifier subtracts the vacuum field
from the subsequent Hall probe output and amplifies the difference by 100,
4
providing a plasma field output with a sensitivity of about 50 mV/G.
Fig. 2(a) shows a calibration curve for the probe system in a Helmholtz
coil. A 0.1 Gauss magnetic field change can be clearly distinguished by the
probe. The noise is mostly from the Hall effect sensor used in the system.
Typical experimental data obtained in the machine operation is shown in
Fig. 2(b).
The increased noise level is due to the ripple in the magnet
current and machine vibration.
II.
B-dot Probe
The B-dot probe system consists of a pickup coil and a low drift elec-
tronic integrator, as shown in Fig. 3. The voltage between the two leads on
the pickup coil is equal to the time derivative of the magnetic flux threading
the circuit, e = -d/dt.
By integrating the voltage in time, the magnetic
flux through the coil and the average magnetic field at the probe location
are obtained.
The pickup coils are constructed on small fiberglass supports with No. 40
magnet wire and installed in half inch diameter stainless steel tubing. They
typically have 1500 turns, with an effective area of about 0.05 m 2 , and a
resistance of nearly 100 ohms. Their frequency response is constant up to
approximately 50 kHz.
A low drift integrator is the key to an accurate B-dot probe measurement. We modified the Evans model 4130 gated electronic integrators for
5
use in our system. The Evans 4130 integrator is a general purpose integrator containing three input channels, a gate and reset triggering circuit,
a pair of voltage comparators for automatic output reset and the main integrator. Several modifications were made. We used one direct input and
disconnected the other two inputs to reduce the leakage current, increased
the feedback resistance R on the input channel from 10 kI to 100 kfl, and
added a low noise preamplifier (Analog Device AD624) with a gain of 375
to increase the input signal level. The values of the feedback resistor and
the amplifier gain axe selected to obtain the maximum signal-to-noise ratio
for the system, when the amplifier and the integrator drift noise levels are
taken into account. Grounding noise has been one of the main noise sources.
No. 16 wires are used for preamplifier and integrator grounds which are soldered together at the power supply grounding point. The integrator drift
is on the order of 0.1 microvolt equivalent input noise level.
In the experiment, the voltage on a pickup coil is typically a few microvolts. It is amplified to millivolt range by the preamplifier and subsequently
integrated by the integrator and stored into computer through a CAMAC
highway. The integrator drift level is digitized for a 0.2 second time interval
before the plasma generation. The drift is later fit to a linear function during the data processing and subtracted. Because the pickup signal from the
coils is small, shielded twisted pair cable is used to transmit the signal. The
high frequency noise, which in our application occurs mainly at 360 and
6
60 Hz, is filtered out by a two stage low pass filter with a cutoff frequency
of 50 Hz at the integrator input. Since the large signal during the vacuum
field ramp charges the capacitors in the filter, this filter is kept open at the
beginning of a shot. Fig. 4 shows the probe response to a 0.1 Gauss field
in the calibration and the probe signal obtained in the experiment.
III.
Discussion
We have described the construction and operation of a Hall probe and a
B-dot probe system in detail. These probes have a sensitivity to magnetic
field changes of less than 0.1 Gauss in a 5 kG background field. The central
issues in the probes' design are subtraction of the vacuum magnetic field
and noise reduction. The B-dot probe has the advantage of not sensing the
vacuum field when the field reaches steady state. Because its main noise is
from the accumulated integrator drift, it is better suited for a short time
measurement. In the Hall probe system, the vacuum field signal is stored
digitally and the noise from thermal drift is relatively small. Therefore it
will be superior to the B-dot probe in a long time application.
Acknowledgments
The author wishes to thank Larry Holcomb for many helpful suggestions in the circuit design. He also would like to express deep gratitude
to D.L. Smatlak, B.G. Lane and R.S. Post for numerous discussions and
7
suggestions on the probe design and the text of the manuscript. This work
was supported by the U.S. Department of Energy, Contract No. DE-AC0278ET51013.
8
References
'B.H. Quon, R.A. Dandl, A. DiVergilio, G.E. Guest, L.L. Lao, N.H. Lazar,
T.K. Samec, and R.F. Wuerker, Phys. Fluids 28,1503 (1985).
2 A.
Honig and M. Moroz, Rev. Sci. Instrum. 49, 183, (1986).
'R. S. Post, et al., in Plasma Physics and Controlled Nuclear Fusion Research 1984: Tenth Conference Proceedings, London, 12-19 September
1984(IAEA, Vienna, 1984), Vol.2.
4 D.L.
Smatlak, X. Chen, B.G. Lane, S.A. Hokin, R.S. Post, Phys. Rev.
Lett. 58, 1853, (1987).
9
0
II
0
R0
IJ+
d
c
0
(0
+
Ij
.)
I')
L-
ci
-4
0.
4O
0
01
0
to
- - 0
t-m
n0
1)
OD
if
tc) C:-
.s
0
44
(
10II
0
A)
10
0
-o|
o
-e
-
LI)
Ki
0u. E
(D
O
N
0
(ssfl09) 13
4)2
0
0
U)
E
-
I
(D
N
.
I
o
0
a
I
a
Iq
O?
(ssnoDS) 8
11
I
-~
4::
-e
a
oo-
.. )
4-O
.4-0
0~
U.4)L
bO
.
0
S-4
4
00
4
4
0~
oe.
o0
-
6-
12
0
'
I
'
I
I
j
I
I
6.u.Qa)E
.)
4-
& 0
I
I
a
I
.
I
I
O
N
(D
(7i
0
I
a
It
-D
(ssno9) g
4)-
0
0
5bO
-0.-
O
I
(0
I
N
a
I
o
I
I
a
I
It
0~J
(ssnoE)
g
13
I. O
0