Control and connectivity considerations in the design of smart meters by Jonathan Page, MSC Gleichmann - Part 2

We’ve looked at the aspect of local control, which comes from providing the end-user, the consumer, with a real-time readout of consumption in a more convenient, easy-to-read location within the home (or business premises). Closing the loop here requires the user to take action based on their energy or water usage i.e. to notice how much is being used and what for and to decide whether this is reasonable or if they can cut back or change their usage pattern e.g. by switching off unnecessary appliances or perhaps changing to a time of day when energy tariffs are lower.

It’s a similar story for remote monitoring and control by the utility companies, except that this relies on suitable methods for communicating data between the consumer premises and the supplier, which we’ll look at in the next piece. For this article however we will focus on the measurement process itself, looking at the sensor technology and the interface to the ubiquitous microcontroller that is at the heart of every modern meter design.

Part 2 – It’s all about sensing the flow

While what is being measured doesn’t change with smart meters, the means of effecting that measurement has to be adapted to the requirements of capturing that data electronically. Many early meter designs were largely, or even wholly, mechanical designs e.g. where the flow of water or gas along a pipe is used to drive a rotating element that then turns a series of geared dials to provide a local display. Even until recently, the majority of electricity meters used electromechanical induction to drive either an analog dial-type display or a pseudo-digital (but still mechanical) readout similar to the mileage register (odometer) found in cars.

Consequently the challenge faced by utility companies when they first implemented automatic meter reading, was extracting this mechanically stored data as an electrical signal for capture and/or transmission back to the utility’s billing office. Solutions to this problem have included generating electrical pulses from the turn of a dial or electronically encoding the mechanical register, allowing it to be interrogated by a data-logging device. The use of such techniques rapidly led to the development of meters with electronic LCD readouts replacing the mechanical displays.

This evolution in meter design has continued with the adoption of low-cost industrial microcontrollers, which not only simplify the collection, storage and local display of meter readings but also provide the necessary communication ports and protocols for transmitting data over wired or wireless networks. Clearly the inclusion of an MCU favours more direct electronic measurement techniques, such as using electromagnetic or ultrasonic transducers to measure the velocity of water flowing through a pipe. But other considerations (reliability, accuracy, cost and especially power consumption) may reinforce the on-going use of simple, proven solutions for capturing the output from traditional meter mechanisms.

One such solution employs a magneto-resistive sensor to pick up the changes in magnetic field from permanent magnets placed on a rotating disc. This arrangement is shown in figure 1, which illustrates the application in a water meter. Note here that strictly the impeller rotation is measuring velocity but this can be converted to volume by knowing how much water flows through the impeller chamber per rotation. The output from the magneto-resistive sensor is a pulse train whose frequency is proportional to the rotation speed.

##IMAGE_2_C##
Figure 1: Using a magneto-resistive sensor to measure rotation

MR sensors have a number of advantages over other magnetic devices like Hall sensors or reed switches, including greater sensitivity and more stable operation. This is because normally four magneto-resistive elements are connected in a Wheatstone bridge configuration in a physical arrangement that maximises sensitivity while minimising any temperature influences. So when these elements are exposed to a magnetic field, their resistance changes and causes in an imbalance in the bridge, resulting in an output voltage proportional to the magnetic field strength.

Commercial MR sensors, such as the Renesas MRUS72S from MSC Gleichmann, integrate the four MR elements on chip with all the necessary circuitry to generate the required output signals for rotation measurement applications as shown in figure 2. This device not only provides the superior sensitivity and highly stable performance that comes from using MR sensors but also does so over a wide temperature range (-40 oC to +85oC) and yet it consumes less than 1μA when used in combination with a low-power MCU.

##IMAGE_3_C##
Figure 2: Internal schematic for Renesas MRUS72S MR Sensor (R1 – R4 are the MR elements)

Power consumption is a critical design criteria for utility meters, particularly for water and gas meters that usually required to operate on batteries that have to last for the installed life of the meter, which could be up to 20 years although increasingly the replacement of meters every 10 years is now mandated in many countries. Even so, the demands of a modern smart meter, especially for radio communications, are considerable and so every possible means of conserving battery life must be exploited.

Low-power MCUs alone aren’t enough. What is needed are MCUs that offer a range of power-saving modes from fully active right through to a deep sleep state when the clock is stopped. And beyond this the key differentiator for applications like metering are MCUs where the peripheral circuits can respond to and pre-process external signals from our measurement sensors without constantly having to interrupt and wake up the core processor.

Atmel’s AVR 32 series includes just such a microcontroller, the AT32UC3L. This device, which is available from and supported by MSC Gleichmann, is targeted at smart meter applications and is one of the industry’s lowest-power 32-bit MCUs. Its picoPower technology enables the AT32UC3L to operate on less than 1.5μA with the 32kHz real time clock active, and below 100nA with all oscillators stopped. But the really innovative feature of this Atmel MCU is its Peripheral Event System, which allows peripherals to send signals (events) to other peripherals without involving the CPU. By offloading these tasks, the time spend handling interrupts is drastically reduced, either freeing up the CPU to handle other tasks or allowing it to remain in an energy-saving sleep mode for longer. And featuring in the line up of peripherals on this device are: an 8-channel analog comparator, an 8-channel 12-bit ADC, a windowed watchdog timer and PWM on all the general purpose I/O pins.

##IMAGE_4_C##
Figure 3: The Atmel AT32UC3L microcontroller

Summary

Utility metering has evolved from early mechanical designs that simply record gas, water or electrical power usage on dials for manual readout through to the modern electronic smart meter that communicate data both locally, to a monitoring unit in the consumers’ premises, and remotely, via a wired or wireless network, back to the utilities’ billing centres. Throughout this the primary purpose has remained unchanged, namely to accurately measure and record this information.

What the transition to smart meters has required is alternate measurement devices either completely replacing the mechanical sensors with wholly electronic transducers or converting mechanical movement to electrical signals at source. The adoption of microcontrollers in smart meters has posed the challenge of power consumption for water and gas meters, which don’t generally have electrical power available where they are installed. Consequently they need to run from batteries so ultra low-power MCUs that include sophisticated power-saving features are vital to ensuring meter designs can meet target life criteria.

This article has looked at a couple of solutions to these problems: the use of of magneto-resistive sensors to convert mechanical rotation to electrical pulses and MCUs that can handle peripheral signals while remaining in low-energy sleep modes. The products cited, from Renesas and Atmel, are both available from and fully supported by MSC Gleichmann.