21st Century Current Sensors
By Mark England and Richard Woodham. This article was first published in Sensors Magazine in 2005Abstract
We are all familiar with the traditional electromechanical Ferraris Disc electric meter, first developed in 1884, as a device for measuring electric power usage. This has been a hugely successful technology, and is still the most common means for recording electric energy usage. Solid-state electric meters allow utility companies to collect more data than just the basic kWh reading, and to provide a powered communications platform to other utility meters, such as gas and water. The enabling technology is the availability of high-power current sensors, the "heart" of a solid state meter, with the appropriate combination of cost and performance.In this article Sentec examines some of the new techniques for current sensing and their commercial applications, with some notes for engineers who might wish to incorporate the features of such current sensors into their products.
Mark England and Richard Woodham, Sentec Ltd
Market drivers
It is in the interests of both provider and consumer that the electrical energy supplied is metered accurately - no-one wants to be overcharged and no utility wants to supply free electricity. But metering technologies have been the Cinderellas of energy infrastructure - they work hard as cash registers for the utilities, using electromechanical designs which have altered little for the past 30 years.However, the utility market is changing, and as the cost of digital microelectronic devices comes down, the opportunities offered by solid-state meters are looking increasingly attractive for more intelligent metering systems.
One key driver is the cost of reading meters. Mechanical meters are most commonly read manually, and in the USA, for example, it is common to read as often as one a month. This is facilitated by the practice of mounting meters outside the building, so that a meter reader can always gain access, day or night.
In Europe, by contrast, the convention is to have the meters mounted inside. Readings are taken much less frequently as a result, as it is expensive for the utility to make repeated visits to obtain a reading.
Another driver is demand management. Utilities have to build generating capacity to be able to supply the peak demands from their customers. Rather than investing in new power stations to cope with higher peak demands, one solution is to sign customers up to a scheme where they allow non-critical loads such as pumps and water heaters to be turned off dynamically at peak times, in return for discounts or credits. For this to be effective, the utility must know how much controlled load is available to be shed, which requires metering devices with built-in communications.
The challenge to develop a solid-state meter capable of supporting these communication devices must be met without compromising the accuracy, robustness or reliability of the meter for its expected twenty year plus lifespan.
Some physics
The energy delivered from an A.C. power source is:
where V(t) and I(t) are the time-varying voltage current. Both V and I alternate at the line frequency, but are not necessarily in phase. So simply measuring average voltage and current levels and multiplying the two does not give an accurate value for the energy delivered. For accurate billing, the product V.I has to be generated (with a bandwidth that is large compared with the line frequency - typically > 1 kHz, to measure the harmonic energy content correctly) before the summation.
Integrated analogue and digital functions
A particularly favourable approach is to digitize the current and voltage signals, and to carry out the multiplication and summation in the digital domain. A review of the performance criteria might suggest that, in addition to the digital multiplier function, expensive A-to-D converters would be required. Fortunately, however, one low cost technology, namely sigma-delta converters, have the useful property that the noise can be reduced almost arbitrarily by progressive digital filtering. Sigma-delta converters can be implemented in CMOS technology, so a complete energy metering function can be integrated into one chip.
Requirements for a fiscal billing sensor
The voltage sensing is achieved simply by dividing down the line voltage using a standard potential divider circuit. Current sensing, by contrast, is rather more taxing. The main requirements for current sensing in a billing meter are:
Ability to measure high currents (up to 200A) with low loss
Linear over a wide dynamic range
Low offset - no-one wants to pay for energy they don't use
Stable over time and temperature
Insensitive to currents in nearby conductors - no-one wants to pay for their neighbours energy
Suitable bandwidth - for accurate calculation of energy delivered in the presence of harmonics
Low cost
Ideally isolated from line voltage, for extension to multiple phase supply installations
Accuracy
Typical commodity meters have a nominal accuracy of 1 or 2%. At first sight this might not sound too difficult to achieve, but there are a number of requirements that make this a significant challenge. The meter must maintain this accuracy over its whole service life, so the initial accuracy required is often 0.1%. The accuracy has to be maintained over several decades of current, over a very wide temperate range (Alaskan snow fields to Arizona desert).
Current sensor solutions
Electrical current sensing is used in many applications, and there are two common existing techniques. In the first, a 'shunt' resistance in the current path generates a voltage proportional to the current. In the second, the magnetic field produced by the current flowing in a conductor is detected and measured. This second method provides isolation, and can take advantage of permeable materials as part of the magnetic flux path.
Shunt methods
In its simplest form, the current to be measured passes through a very low value resistor with low temperature coefficient, and the potential drop across the resistor is measured. For a shunt capable of measuring >100A, the shunt is a relatively large four-terminal resistor fabricated from relatively expensive low temperature coefficient of resistance (tempco) material. An alternative solution is to fabricate a resistive current divider, which is a network with legs comprising two high-value resistors connected to the shunt, feeding into a virtual earth differential input current amplifier. The benefit of this approach is that the shunt and the two divider resistors can have a much higher tempco, provided that they are the same, for example, by making them all from the same material. This is because the current division is determined by the resistance ratio.
A particularly elegant solution is to fabricate all three resistors from copper, for example as layers of a printed circuit board. A thick outer layer carries the current, whilst on inner layers, thin meander tracks form the divider legs. Although copper has a relatively high tempco, the layers are in good thermal contact, so that the tempco's track, whilst clever design can eliminate any undesirable inductive coupling effects.
Field methods
The magnetic field generated by a flowing current can be used to measure current in a number of ways. A Hall-effect device can measure the field directly, generally using some suitable magnetic pole pieces to direct the flux through the device. In a closely-coupled current transformer or CT, the field in a permeable core is cancelled by current flowing in a secondary winding on the core with many more turns than the primary. This much lower secondary current is measured using a small shunt (or burden) resistor.
The presence of permeable materials at high magnetic field or current levels results in non-linearities. Therefore, this has led to the adoption of a Rogowski coil arrangement for current measurement - an air-cored toroid around a straight conductor. The coil measures dI/dt, and requires an integrator to recover current. Although bulky, it has excellent linearity and overload response. To minimise sensitivity to external magnetic fields, a conventional Rogowski coil has a rigid coil former and symmetrical coil winding, as small errors in the positioning of the windings render the Rogowski coil susceptible to fields from nearby conductors. This has prevented the technique being widely used in commercial applications thus far.
Latest solutions
However, the latest generation of utility meters uses a unique solution to achieve the symmetry required in a Rogowski coil - fabricating the coil windings using a printed circuit board (PCB). In this case, the geometries are tightly controlled, and the resulting response can be predicted with great accuracy. As an added benefit, the measurement electronics can be integrated on the same PCB.
A typical geometry of coils used is shown in Figure 1. There are two mirror-image coil layers, each of which consists of two concentric sections wound in opposite directions, such that the turns-area of the inner and outer sections are equal. The current to be measured flows in bus bars formed in a plane above or below the plane of the coils, as illustrated in Figure 2. Magnetic flux from the current flowing in the bus bars is coupled into the sensor coil. Other geometries of coils and bus bar are possible, tailored to fit the constraints of the application.
Figure 1 - Typical layout of one layer of sensor coil
Performance
The sensor measures the rate of change of current, and this must be integrated to calculate the actual current. For maximum stability, this is best achieved digitally. For power measurement, this can be achieved using a chip such as the Analog Devices ADE7759, which also carries out the multiplication necessary to compute power.
Figure 3 illustrates the performance of a 100A current sensor. In this instance, the sensor used was interfaced using a low-noise pre-amp to an ADE7759 metrology chip. The measurements show the registration of power in a 60Hz 240V circuit, showing a typical error of less than 0.3% of reading over four decades of current. The performance is limited by the ancillary electronics, rather than the fundamental sensor characteristics.
Figure 2 - Current flowing in adjacent conductor couples into sensor coil
Figure 3 - Performance of current sensor measuring power in a 240V circuit
Engineering notes
At present, this sensor technology has been adopted by Sensus Metering Systems as the core of their iCon meter product family.
However, outside the metering industry, there is a wide range of domestic and industrial applications for a wide dynamic range device to monitor electric current and power. Controlling and regulating the input power of devices allows their performance to be optimised for energy efficiency, which is of increasing environmental and financial importance.
Some applications of current and power sensing include:
- control system diagnosis
- current supply fingerprinting (e.g. harmonic content)
- the incorporation of safety cut-off features and surge trips
- charge integration and condition monitoring on re-chargeable batteries
- control of complex loads such as electric motors
Such benefits may be desirable but adding a significant cost burden to the design is not. This is where the PCB based solutions described above are of most benefit - the cost penalty from using a small area of an existing circuit board for current sensing purposes is minimal, whilst the signal processing required can often be carried out using, for example, an existing microprocessor.
For more information on Sentec or this article, or to interview any of the main contributers, please contact Hermione Crease, Marketing Communications Manager, on +44 1223 303800.
