The role of semiconductors in the meters of the future
We polled two of the larger manufacturers of semiconductors -Jukka Eskelinen of Atmel and Dave Simpson of SAMES - to establish how they saw the role of semiconductors in metering today, and what they believe is in store in the future. These were the questions we asked; their replies are dealt with individually. Smart Energy International thanks them for their input.
- How competitive is the market?
- Has the fact that companies can buy electronic metering building blocks at relatively low cost encouraged smaller organisations to enter the market?
- Will costs continue to decline?
- How is technology changing?
- Are there new techniques as far as using semiconductor devices with sensors (e.g. shunts, current transformers, hall effect sensors) are concerned?
- What about the move from analogue to digital processes?
- A considerable amount of work is being done in standardisation groups on resistance to electro-magnetic interference. Are there any major initiatives in terms of making chips more resistant to this type of interference?
- With communications attracting more and more attention in the metering context, will there be increased levels of integration of all metering functions into fewer devices?
- From the perspective of a single tariff meter, how would you rate mechanical as opposed to electronic meters?
Q1: There are many different semiconductor solutions available, ranging from pure analogue measurement ASICs to highly integrated mixed signal products. Initially they are all measured against the electro-mechanical meters in terms of cost, functionality, performance and reliability – a tough test, especially for single phase household meters.
We believe that solid state meters will first be justified in the industrial meter market, where an increasing amount of functionality is needed. There is a range of new parameters that they can record, including those related to the quality of supply. This is where electromechanical meters fall short.
Atmel is now looking at the single phase meter market too, as we are able to meet its strict cost requirement.
Q2: Electronics has indeed lowered the barrier to entry to meter manufacturing. The initial amount of capital investment is not really high, and we will see new companies enter the market. However, much more is involved than the design and manufacture of the meter itself. The whole market is rapidly changing and companies which have supporting capabilities like networking, remote reading and data base management are likely to play a significant role. They are able to add value and can meet the demands of future metering markets.
Q3: Costs will certainly continue to decline, as a result of new manufacturing technologies, but even more as a result of the increase of chip level integration. System level integration products, both standard components and ASICs, that contain microcontroller, DSP, analogue blocks, LCD driver, logic, ROM, FLASH will drive the total system cost down. New small embedded RISC DSP cores and µC cores deliver a lot of computing power. FLASH program memory adds flexibility of customising to application specific standard products (ASSPs) which are manufactured in high volumes. All these will help bring the meter cost down.
Q4: The technology elements for the near future already exist. The question is how to pick the right solutions, the right blocks, and combine the IP so that the result maximises performance, functionality and cost. This is where semiconductor manufacturers, meter manufacturers and utilities need to work together. I believe we can easily build a chip that contains a great deal of the functionality of a meter, including communications and control, and still meet the cost target of the market.
Q5: Our devices accept any of these three front ends. The chips are configurable to fit whatever the meter manufacturer prefers. We do not believe it is feasible to integrate the front end with the rest of the chip, because then one makes compromises in terms of performance and sacrifices flexibility.
Q6: We believe that the conversion from the analogue inputs to the digital domain must be done as early as possible in the signal path. This minimises the errors and interference, and makes data processing more effective. When the data is in the binary format, a microprocessor can do what is required.
From a manufacturing perspective, the digital semiconductor processes offer a cost benefit compared to more complex analogue processes. That is why all Atmel’s metering chips are manufactured on deep sub-micron standard digital CMOS processes.
Q7: Chips need protection against ESD, which may cause them to fail; this is typically a problem during manufacturing. Again, minimising the amount of analogue signal processing improves a chip’s immunity to electro-magnetic interference. For high voltage protection a typical CMOS chip needs protection – something that the meter designer must consider.
Q8: We believe that a meter can be built around one or two chips. We also believe that these meters will be flexible, because a lot of the functionality is built on software which is stored on a reprogrammable FLASH memory. A meter can be automatically configured and calibrated at the end of a production line, as well as out on the field. More features and functionality can remotely be added to the meter later. The end users can select the level of service they want, and the utility can remotely configure individual meters.
Q9: An electro-mechanical meter is a sophisticated piece of high precision mechanics, with incredible accuracy and reliability at the price tag it carries. However, it cannot meet the additional functionality customers demand. We believe solid state meters will replace mechanical meters. A solid state meter offers a much wider range of measurement parameters and functionality, it offers easy interfacing, and it is not susceptible to fraud, just to mention some of its advantages.
Q1: There is a great deal of diversity in the performance and functional requirements of solid state meters, often dictated by the selling price constraints of the meter, which give rise to cost pressures on the bill of materials. These cost pressures, along with stringent performance requirements over a wide dynamic range and temperature and reliability criteria, mean that semiconductors will continue to play an increasing role in energy metering applications.
It is thus reasonable to assume that the number of suppliers of integrated circuits to the energy metering industry will increase. However, the market requirements should not be over-simplified. SAMES has already developed a family of over 20 different devices to address specific market needs, and the size of our product family continues to increase.
Q2: Over-simplification of the entry barriers could be a fatal error. The design and development of a laboratory functioning energy metering system is but one facet of a chain of requirements to be a successful player in this sector. One of the greatest limiting factors is the enormous cost required to take an energy metering product through to utility approval – often 30 months or more before any returns become a reality.
Q3: The meter manufacturer is concerned with the cost of the circuit implementation rather than the cost of the semiconductors alone. Reductions in die size and higher levels of integration have considerably reduced costs over the past five years.
Although the cost of energy measurement application specific standard products (ASSPs) will continue to decline, the rate of decline will be defined by two primary factors:
- Energy measurement devices are mixed signal, incorporating both analogue and digital circuitry. Although a shrink in technology may shrink digital circuitry, the performance of analogue circuitry depends on physical size and does not benefit to the same extent.
- In order to improve system reliability, as well as reduce the size and cost of the printed circuit board, higher levels of integration will be required. While semiconductor prices may increase due to higher levels of integration, overall circuit implementation costs will continue to decline.
Q4: The future of solid state metering is focused towards low cost standard metering integrated circuits. Naturally all the required standard specification requirements have to be met. In addition the meter manufacturer must gain increased economies of scale on the production line through the simple programmability of both the rated voltage and current settings, while the flexibility to add further features should not be compromised.
Q5: The current sensor constitutes one of the most costly components in the solid state meter. The shunt resistor still provides the most economical current sensing solution available globally.
However, the current transformer has wide acceptance in both developed and developing markets for both single and 3-phase meters, although alternative solutions are being developed because of increasing demands for accuracy with DC content in the line current. Alternatives must meet both the performance requirements and the cost constraints.
Q6: There is no solid state metering solution available that does not have an interface between the analogue and digital domains. Solid state metering technology will always be based on such mixed signal technology, where performance is being advanced on a continuous basis, mainly in the areas of accuracy and power consumption, as well as further functionality suitable for integration.
Q7: The signal processing techno-logy employed on a chip ensures that the range of energy measurement integrated circuits does not exhibit any sensitivity to electro-magnetic influences. System design constraints, such as printed circuit board layout sensitivity, as well as the elimination of the requirement for any critical external components, are thus eliminated to a large extent.
Q8: Without doubt, higher levels of integration will continue to be applied to metering. Over time, and with the development and maturity of national and international standards for solid state meters, there will be a migration to an optimised partitioning inside the meter. The defining factors for the partitioning will be cost, the customers’ functional requirements and manufacturing flexibility of the metering system.
Q9: In order for a market for new technology to develop, there has to be some form of advantage to the customer. For the simple single tariff meter, the primary customer considerations would be cost, performance and reliability. Current technology fulfills all these criteria in most of the world’s large electro-mechanical markets.
However, two factors will influence the growth of the simple single tariff solid state meter market:
- The cost of solid state metering systems will continue to decline, as integration levels increase. Furthermore it is reasonable to assume that the raw material costs of the electro-mechanical meter will increase over time.
- The simple meter will not remain the ‘simple meter’. As soon as additional functionality is required – which is usually difficult to incorporate into the electro-mechanical meter – solid state metering becomes the preferred option.