UNIVERSAL CURRENT SENSOR SEMINAR REPORT 2010

1. INTRODUCTION

The measurement of electric current strength is not always easy, especially

when the measured signal requires further electronic conditioning. Simply
connecting an ammeter to an electrical circuit and reading out the value is no
longer enough. The current signal must be fed into a computer in which sensors
convert current into a proportional voltage with minimal influence on the measured
circuit. The basic sensor requirements are galvanic isolation and a high bandwidth,
usually from DC up to at least 100 kHz. Conventional current measurement
systems therefore tend to be physically large and technically complex.

2. EARLY SOLUTIONS
The oldest technique is to measure the voltage drop across a resistor placed in the
current path. To minimize energy losses the resistor is kept very small, so the
measured voltage must be highly amplified. The amplifier’s offset voltage must be
as small as possible and its supply voltage must be at the potential of the circuit,
often 110 V mains (230 V in Europe) with high parasitic peaks from which its
output must be isolated. This requirement increases overall system cost.

Another widespread principle is the transformer. Its construction is much

simpler, but it doesn’t allow the measurement of DC signals. Isolation between
primary and secondary sides is implicitly given. A problem is the limited
frequency range.

DEPT OF ELECTRICAL & ELECTRONICS 1 MET’S SCHOOL OF ENGINEERING

UNIVERSAL CURRENT SENSOR SEMINAR REPORT 2010

3. HALL CURRENT SENSORS

Hall sensors also measure the magnetic field surrounding the conductor but,
unlike current transformers, they also sense DC currents. A circular core of soft
magnetic material is placed around the conductor to concentrate the field. The Hall
element, which is placed in a small air gap, delivers a voltage that is roportional to
the measured current. This sensor also offers a galvanic isolation.
The very small output voltage of the Hall element must be highly amplified,
and the sensitivity is temperature dependent and requires adequate compensation.
There is an inevitable offset, i.e., a small DC voltage at zero current; the offset
amplitude and temperature coefficient are subject to significant fluctuations. The
smaller the current to be measured, the higher the offset-induced relative error.
Also of note is sensitivity to short current peaks in the circuit: according to the
hysteresis properties of the core material, these peaks can cause a static
magnetization in the core that results in a permanent remanence, and finally to an
offset alteration of the Hall element.

The two types of Hall effect current sensor are open loop and closed loop. In
the former, the amplified output signal of the Hall element is directly used as the
measurement value. The linearity depends on that of the magnetic core. Offset and
drift are determined by the Hall element and the amplifier. The price of these
sensors is low, but so is their sensitivity.

Closed-loop Hall sensors are much more precise. The Hall voltage is first
highly amplified, and the amplifier’s output current then flows through a
compensation coil on the magnetic core (see Figure 1). It generates a
magnetization whose amplitude is the same but whose direction is opposite to that

DEPT OF ELECTRICAL & ELECTRONICS 2 MET’S SCHOOL OF ENGINEERING

UNIVERSAL CURRENT SENSOR SEMINAR REPORT 2010

of the primary current conductor. The result is that the magnetic flux in the core is
compensated to zero. (The principle is similar to that of an op amp in inverter
mode, for which the input voltage is always close to zero.)

The nonlinearity and temperature dependence of the Hall element are thus
compensated but the offset remains. Closed-loop current sensors work up to
frequencies of ~150 kHz. They are not cheap, though, and for high currents they
become very bulky.

4. MAGNETO RESISTIVE FIELD SENSORS

Practical magnetic field sensors based on the magnetoresistive effect are
easily fabricated by means of thin film technologies with widths and lengths in the
micrometer range. They have been in production for years in many different
executions. To reduce the temperature dependence, they are usually configured as
a half or a full bridge. In one arm of the bridge, the barber poles are placed in
opposite directions above the two magnetoresistors, so that in the presence of a
magnetic field the value of the first resistor increases and the value of the second
decreases (see Figure 2).

DEPT OF ELECTRICAL & ELECTRONICS 3 MET’S SCHOOL OF ENGINEERING

UNIVERSAL CURRENT SENSOR SEMINAR REPORT 2010

For best performance, these sensors must have a very good linearity between
the measured quantity (magnetic field) and the output signal. Even when improved
by the barber poles, the linearity of magnetoresistive (MR) sensors is not very
high, so the compensation principle used on Hall sensors is also applied here. An
electrically isolated aluminum compensation conductor is integrated on the same
substrate above the permalloy resistors (see Figures 3 and 4). The current flowing
through this conductor generates a magnetic field that exactly compensates that of
the conductor to be measured. In this way the MR elements always work at the
same operating point; their nonlinearity therefore becomes irrelevant. The
temperature dependence is also almost completely eliminated. The current in the
compensation conductor is strictly proportional to the measured amplitude of the
field; the voltage drop across a resistor forms the electrical output signal.

DEPT OF ELECTRICAL & ELECTRONICS 4 MET’S SCHOOL OF ENGINEERING

UNIVERSAL CURRENT SENSOR SEMINAR REPORT 2010

Magnetoresistive sensors, as are Hall elements, are very well suited for the
measurement of electric currents. In such applications it is important that external
magnetic fields do not distort the measurement. This is achieved by forming a full
bridge made of four MR resistors, where the two arms of the bridge are spatially
separated.

The barber poles have the same orientation in the two arms, so that only a
field difference between the two positions is sensed. This configuration is
insensitive to external homogeneous perturbation fields. The primary current
conductor is U-shaped under the substrate, so that the magnetic fields acting on the
two arms of the bridge have the same amplitude but opposite directions. This way
the voltage signals of the two half-bridges are added.

The sensors have been in production for several years. A ceramic plate is
used as the substrate, onto the back of which the primary conductor is glued to
achieve an isolation of several kilovolts.

The sensors require neither a core nor a magnetic shielding, and can
therefore be assembled in a very compact and cheap way. They have a linearity
error of <0.1%, gain error <0.2%, offset voltage <10 mV, and temperature drift of
the gain <100 ppm/K. The output signal is calibrated to 2.5 V at nominal current
by a laser trimming process or by a digital calibration. The rise time (10%–90%) is
~1.7 ms, which corresponds to a frequency bandwidth (–1 dB) >50 kHz; this value
results from the speed of the regulation. The sensor is powered by a standard
bipolar voltage of ±15 V and the power consumption is 320 mW to 640 mW,
depending on the measured current value.

DEPT OF ELECTRICAL & ELECTRONICS 5 MET’S SCHOOL OF ENGINEERING

UNIVERSAL CURRENT SENSOR SEMINAR REPORT 2010

The size of the actual MR sensor chip with the four permalloy strips on
silicon is 1 mm by 2 mm2, the distance between the two arms of the bridge being
1.6 mm. It is mounted with the electronics on a 23 mm by 35 mm2 single in-line
hybrid circuit designed for through-hole PCB mounting with very low profile. The
NT-xx product family from Sensitec or F.W. Bell includes sensors for nominal
currents of 5 A, 15 A, 25 A, and 50 A.

Among the advantages of these sensors are:

 Significantly smaller volume and weight compared to conventional current
sensors, permitting greater flexibility in application-specific design
 No remanence in the event of overload
 Measurement of DC and AC currents without additional loss
 Wide frequency range due to low inductive design
 No auxiliary supply necessary on the level of current to be measured
 Low system costs

Initially developed for use in inverters and servo-regulators, MR current

sensors have a broad range of other applications: energy distribution; power
electronics systems such as inverters and mains receivers; battery management
(charge and discharge currents); solar technology; electrical vehicles;
uninterruptable power supplies; robotics; automotive (e.g, drive-by-wire); and
safety devices.

DEPT OF ELECTRICAL & ELECTRONICS 6 MET’S SCHOOL OF ENGINEERING

UNIVERSAL CURRENT SENSOR SEMINAR REPORT 2010

The signal conditioning electronics of the NT-xx family is a discrete circuit;

in the new NTS-xx family, it has been integrated into an ASIC operating at
frequencies up to 100 kHz. The output signal is scalable to meet the requirements
of external A/D converters. Additional functions are a comparator I/O with
adjustable trigger level for short-circuit detection, an open-drain concept that
permits three-phase monitoring, and a true rms output for power measurement and
protection.
These sensors are supplied with a unipolar voltage of +5 V, compatible with
most digital interfaces. Since no core is used to concentrate the magnetic field,
there is no remanence, which means that overcurrents produce no permanent offset
changes. The compensation method guarantees a high linearity. The profile on the
PCB board is low, and the mounting is very easy. The ASIC allows digital
calibration of the current measurement system even after installation in an
application. The calibration data are stored on the chip. The MR sensor is located

DEPT OF ELECTRICAL & ELECTRONICS 7 MET’S SCHOOL OF ENGINEERING

UNIVERSAL CURRENT SENSOR SEMINAR REPORT 2010

on a second chip. A co-integration of sensor and interface electronics on the same

silicon substrate is possible but is not yet cost effective. The principal application
areas will be in electronic systems for the automotive industry (see“Current
Sensors in Tomorrow’s Cars”), motion control, automation, or energy management
in buildings.

In the standard sensor types described thus far, the nominal current is set by
the geometry of the primary conductor that is part of the system and by the
distance between the MR chip and the conductor. Recent investigations, however,
have demonstrated that the high-current bus bar need not be interrupted or guided
through a hole, as required with Hall transducers. Instead, it can simply be shaped
in the form of a bus bar plate. A sensor module realized as a dual in-line surface
mount device component can be mounted on the power PCB board placed above
the bus bar plate so that the current flow can be directly measured.

DEPT OF ELECTRICAL & ELECTRONICS 8 MET’S SCHOOL OF ENGINEERING

UNIVERSAL CURRENT SENSOR SEMINAR REPORT 2010

A very clever and cost-effective solution for the measurement

of high currents, especially in series applications, was presented at PCIM’99 in
Nuremberg, Germany, where it drew considerable attention. In this execution, the
current is directly measured in the primary conductor (flat conductor, PCB) and the
sensor module is then simply clipped into a slot made in the conductor. The result
is a differential field measurement system that is insensitive to homogeneous
external magnetic perturbations. The sensors and the ASIC are mounted on an
appropriate substrate and encapsulated in a plastic package (see Figure 5). With
this sensor, a broad range of currents can be measured simply by adapting the
geometry of the conductor. Potentially heterogeneous perturbation fields can easily
be shielded.

The advantages are obvious: these current sensors are not only small,
compact, and light, but also cheap and easy to mount. The primary current
conductor can be part of the application and need not be mounted separately. This
opens the way to completely new construction possibilities for developers of power
electronic modules and devices such as for mains receivers and frequency
inverters.

DEPT OF ELECTRICAL & ELECTRONICS 9 MET’S SCHOOL OF ENGINEERING

UNIVERSAL CURRENT SENSOR SEMINAR REPORT 2010

5. THE MAGNETO RESISTIVE EFFECT

The anisotropic magnetoresistive effect (AMR effect) is known to be present
in a whole family of ferromagnetic alloys. Most of these alloys are composed of
iron (Fe), nickel (Ni), and chromium (Cr), and may be binary (two components) or
ternary (three components). They have in common a more or less strong anisotropy
in their magnetic properties. Whenever these materials are exposed to a magnetic
field during crystal formation, a preferred orientation in magnetization will result.
The same happens when the materials are forced into shape, i.e., a mechanical
anisotropy is imposed.

Long before chip or even thin film technologies were developed,

simple wires drawn of NiFe were known to possess an orientation of their
magnetization, a magnetic moment, along their linear axes. Interestingly, it was
found that changing the orientation of the magnetic moment in the wire caused a
current passing through it to change correspondingly. The orientation could be
changed by applying an external magnetic field, and generally an increase in
current (i.e., a decrease in resistance) was observed. This phenomenon was called
anisotropic magnetoresistive effect.

Today, the ferromagnetic materials can be deposited as thin films and

structured into small strips that are typically 40 nm thick, 10 mm wide, and 100
mm long. A magnetic field is applied during the process.

In modern device fabrication an alloy commonly called Perm alloy (81% Ni,
19% Fe) turns out to be the best compromise in terms of device sensitivity,
longevity, and reproducibility. As does the wire, the strip has an orientation of the

DEPT OF ELECTRICAL & ELECTRONICS 10 MET’S SCHOOL OF ENGINEERING

UNIVERSAL CURRENT SENSOR SEMINAR REPORT 2010

inner magnetic moment, M0, parallel to its long axis. When an external field is
applied, the total magnetization, M, of the strip is turned at an angle, ø.

In the most general case, the electrical resistance of AMR material depends
on the angle, ø, between the direction of the magnetization, M, and the direction of
the current going through it. This dependency can be described by the relative
change in resistivity, DR/R being proportional to cos2ø. This means that when the
current and magnetic moment, M, are parallel, the resistance of the strip is greatest;
when they are at a 90° angle to each other, it is smallest. Disregarding the influence
of film thickness, the maximum relative change in resistance is a material constant.
In Perm alloy it is typically 2%–3%.
The characteristics of a simple strip make it less than desirable for
measurement applications.

DEPT OF ELECTRICAL & ELECTRONICS 11 MET’S SCHOOL OF ENGINEERING

UNIVERSAL CURRENT SENSOR SEMINAR REPORT 2010

The reasons are that it exhibits a vanishing sensitivity at low fields; that the
direction of the external field cannot be determined; and except for the region
around 45°, there is nearly no signal linearity.

Adding small structures made of a highly conductive material such as

aluminum to the strip creates a barber pole configuration. In the simplest case,
barber poles are just tiny blocks of material sitting on top of the Perm alloy. Their
typical linear dimension in all directions is a couple of microns, and they are
arranged at an angle of <45° to the long axis of the strip. Serving as local shorts for
the Perm alloy, they force the current to flow through it at that angle.

Thus, an angle of 45° between current flow and magnetization, M, results

when no external field is applied. This gives the barber pole device the desired
maximum sensitivity at zero field and a corresponding linearity around the
working point. Working at roughly an average value between maximum and
minimum resistance of a simple strip, the barber pole device reacts in an
unequivocal way to changes in the external field direction.

In the majority of practical applications AMR sensor chips are composed of

barber pole devices wired as Wheatstone bridges. AMR chips have significant
advantages over other technologies such as Hall devices:

They can be used at high frequencies (some MHz are no problem) and high
temperatures such as those in automotive environments (continuous service at
150°C, peak values at 190°C), and exhibit low and stable offset values. Due to

DEPT OF ELECTRICAL & ELECTRONICS 12 MET’S SCHOOL OF ENGINEERING

UNIVERSAL CURRENT SENSOR SEMINAR REPORT 2010

their high sensitivity (~10 X that of Hall devices) they can be used to detect very
weak fields (the present limit is ~10 nT)

6. CURRENT SENSORS IN TOMORROW’S CARS

Fuel efficiency first—this is the present motto of automotive development
worldwide. Not only the combustion engine, but al l other automobile components
as well are being optimized to achieve this goal. Many components that are at
present powered by the car’s engine will in the near future operate on electricity—
air conditioning compressors, water pumps, oil pumps, and the like. Due to
continuously changing engine speed, these components have rather poor efficiency
ratings. Electric motors have the capability of optimizing their operation,
independent of engine speed.

This change, though, would overload the conventional 12 V car generators

and batteries. Rather than increasing the current and corresponding cable cross-
sections, it makes more sense to make 42 V the future standard. One immediate
consequence will be a combined starter-generator placed directly on the crankshaft.

Prototypes of this unit were presented by Mannesmann-Sachs AG at the

1999 IAA car exhibition in Frankfurt, Germany. The unit is a three-phase
synchronous engine using a permanent excitation. It is integrated into the flywheel
of the clutch and can be used as either a starter or a generator.

Efficiency is increased significantly in both modes of operation. On one

hand, the engine is started at a higher starter speed, and on the other, the new
generator will produce much more electric power without increasing power

DEPT OF ELECTRICAL & ELECTRONICS 13 MET’S SCHOOL OF ENGINEERING

UNIVERSAL CURRENT SENSOR SEMINAR REPORT 2010

demand on the combustion engine. Very effective start/stop systems can thus be
realized. The engine is switched off whenever the vehicle stops and started again
automatically when the driver wants to go on. These measures should lead to fuel
savings of around 30% compared to today’s vehicles.

A prototype of the starter-generator control unit made by Mannesmann-

Sachs, requires the use of three-phase MOSFET alternators. For control and safety
reasons, these alternators need current sensors for at least two of these phases.
Mannesmann uses a customer-specific version of Sensitec’s NTS-xx system, in
this case an NTS-xx hybrid without a conductor for the working current.

The conductor in Photo 6 is configured as a 0.8-mm-thick copper bus bar. At

the location of the NTS-xx, the bar takes a U-shaped configuration to generate the

DEPT OF ELECTRICAL & ELECTRONICS 14 MET’S SCHOOL OF ENGINEERING

UNIVERSAL CURRENT SENSOR SEMINAR REPORT 2010

necessary magnetic field gradient. For this application, the NTS-xx is mounted
directly above the bar and electronically calibrated in situ.

This arrangement has many advantages. Production is much easier because

there is no need to break or disconnect the conductor carrying the working current
or even feeding it through the NTS-xx itself. Furthermore, because this version of
the NTS-xx can be used for practically all measurement ranges (e.g., 10 A, 100 A,
or 1000 A), provided the busbar is adapted correspondingly, a great reduction in
manufacturing cost can be realized.

DEPT OF ELECTRICAL & ELECTRONICS 15 MET’S SCHOOL OF ENGINEERING