SOLID STATE SOLUTIONS FOR ELECTRICITY METROLOGY Anthony Collins Analog Devices Inc., Wilmington MA 01887, USA Email: [email protected] 1.0 Introduction • Analog meters do not offer the same high Significant changes are taking place in the electric integration (cost reduction) path that digital energy industry worldwide. Deregulation of the electronics does. utilities will produce a tougher competitive environment for the suppliers of electricity. Increased For these reasons the remainder of this discussion demand for electrical energy will require better will focus on digital solid state electricity meters. management of distribution. Suppliers will need to identify ways of supplying a better level of service at Instantaneous Power Signal p(t) 1 T a reduced cost. One way to do this is to install a I P= p(t) T solid state electricity meter. 0 Apart from tangible benefits like higher accuracy – x especially in the presence of nonsinusoidal Average Real Power waveforms [1,2], the solid state meter offers more V than just the measurement of kWhrs. Given the - Analog Multiplier - LPF (transconductance) (Low Pass Filter) accuracy, flexibility and powerful network - Hall effect multiplier management possibilities offered by solid state metrology the days of the electromechanical meter Figure 1. Analog (continuous time) Signal are numbered. Processing This paper gives an overview of electricity meter design using solid state electronics. It discusses 2.1 Digital Solid State Electricity Meters various architectures and some of the key The digital meter relies on a process called technologies that allow reliable and cost effective analog-to-digital conversion. It does this by using a solid state meters to be implemented. It also device called an ADC (Analog to Digital Converter). discusses semiconductor reliability issues. The ADC takes samples or “snap shots” of the analog signals at discrete instances of time. These 2.0 Anatomy of a Solid State Meter “snap shots” or discrete time signals are in turn Solid state electricity meters are generally based on converted to numeric values by the ADC - see figure one of two types of electronic technologies, analog 2. or digital signal processing. Signal processing refers Analog Signal Sampled Signal Digital Signal to multiplication and filtering in order to extract the (continuous time signal) (discrete time signal) (numeric signal) required information (kWhrs, VARS etc.). An analog electronic meter processes analog signals, i.e. ...10011010110101.... signals which exist at every instance of time - like ADC voltage and current signals, see figure 1. By the same definition an electromechanical meter can be Figure 2. Analog to Digital Conversion said to be an analog meter. Analog signal processing is an older solid state Once in this numeric or digital format, digital circuits, technology that struggles to achieve the same e.g., microprocessors, can easily and reliably price/performance ratio as its digital counterpart. process these signals. Figure 3 shows a graphical representation of this digital processing. Disadvantages of analog signal processing Instantaneous I • Analog signal processing is not a flexible Power Signal Average Real technology. Analog meters can not be easily ...10011010110101.... ...10011010110101.... Power/Energy reconfigured to meet specific local requirements x ...10011010110101.... or upgraded at a later date. v • Analog meters do not offer the same stability as their digital counter parts over large variations in ...10011010110101.... operating environment and with time. - Digital Multiplier - Accumulator / Integrator (MCU, DSP or other) (MCU, DSP or other) • Analog meters do not offer the same simple, cost effective and stable calibration features, Figure 3. Digital (discrete time) Signal Processing which may be implemented using digital technology. Metering and Tariffs for Energy Supply, 25-28 May 1999, Conference Publication No. 462 © IEE 1999 2.2 Selecting an Analog to Digital Converter 100kHz SNR=× 6.02 12 Bits + 1.76dB + 10log " for Energy Metering 10 ! 4kHz $# The Analog to Digital conversion can be carried out SNR= 87.98dB or 14.3 Bits using one of several schemes or architectures [3]. Each of the architectures has its own associated The effective resolution of the ADC over the 2kHz advantages and disadvantages. It is important to be bandwidth of interest has been increased to 14.3 aware of these differences when matching an ADC bits. In general the effective resolution of the ADC is to a specific application. improved by ½ LSB for every doubling of the sampling frequency. 2.2.1 ADC Requirements for Energy Metering • High Resolution, i.e., 16 Bits. 2.2.3 ADC Architectures Because of the relatively wide dynamic range The main architectures for Analog to Digital (4%Ib to 400%Ib) and the accuracy Conversion are listed below. requirements (0.5%) of the application, the resolution of the ADC needs to be high. Another • SAR types (Successive Approximation) approach which, is discussed later, is to use • Flash (Parallel) and Half Flash ADC over sampling and DSP to improve resolution. • Integrating (Dual Slope) ADC • A Sampling rate of at least 2 to 4 kSPS (kilo • Voltage to Frequency (V/F) Samples Per Second) is required for the • Σ-∆ (Sigma Delta) ADC application. A basic rule of sampling theory states that the rate (frequency) of sampling must be at least twice the highest frequency content SAR Converters of the signal. This is called the Nyquist rate. In The successive approximation ADC is a very widely energy metering current ANSI and IEC used and understood technology. The conversion specifications [6] call for accurate measurement technique is based upon continuously comparing the of frequency content up to the 20th harmonic sampled input signal to know voltages in order to (1kHz or 1.2kHz depending on the line convert it to a binary word. frequency). • Low Cost. The solution must be a low cost one Advantages because the end application is particularly cost sensitive, especially in residential metering 1. High Speed, up to 1 million conversions per applications. The choice of ADC may not only second is possible. impact cost because of the price of the ADC 2. Relatively low power consumption when itself, but also because of external components compared to other architectures. required by the ADC. 3. Relatively low cost when the required resolution • The ADC must not consume excessive power. (conversion accuracy) is low, i.e., less than 12 One of the challenging aspects of solid state Bits (4096 counts) meter design is the design of the PSU (Power Supply Unit). This must be low cost and reliable Disadvantages for the life of the meter (15-20 yr.). Typically these supplies are capacitor based, as the cost 1. Becomes costly when high resolution is required of a transformer is often prohibitive. Capacitor ( >14 Bits). based designs have a low VA rating and are 2. Output signals from transducers generally typically only capable of providing 10 to 15 mA require substantial signal conditioning. at 5V. Depending on the application this can add significant cost. 2.2.2 Oversampling Another approach which reduces the resolution Flash converters requirements for the ADC, is to use quantization A Flash converter converts a sampled signal to a theory to increase the effective resolution over the binary word in one step (or two steps in the case of a desired bandwidth. This is done by oversampling the half-flash). Hence the time to do a conversion is very analog signal, i.e., sampling at a rate which is much short. The flash and half-flash architecture is usually greater than the Nyquist rate and then digitally used in applications requiring low resolution (256 – processing the resulting data [4]. For example by 1024 counts) but high-speed conversion (20 – 50 using a 12 bit ADC and sampling a pass band signal Million samples per second). Applications typically of 2kHz at 100kSPS the SNR over the 2kHz include video and communications. bandwidth is calculated as: Advantages 1. Very high speed conversions Metering and Tariffs for Energy Supply, 25-28 May 1999, Conference Publication No. 462 © IEE 1999 Disadvantages Sigma-Delta ADCs Within the last several years, the sigma-delta 1. Low resolution architecture has become more popular for realizing 2. High power consumption high-resolution Analog to Digital Conversion. One of 3. Relatively high Cost the more important advantages is that the ADC can 4. Output signals from transducers generally be combined with DSP functions on one mixed require substantial signal conditioning. signal VLSI chip. Depending on the application this can add Basically, a Sigma-Delta converter digitizes an significant cost. analog signal with a very low resolution (1 Bit) ADC at a very high sampling rate. By using over-sampling Integrating Converters techniques in conjunction with noise shaping and Integrating ADCs are used for slow, precise digital filtering (DSP), the effective resolution is measurements such as in digital voltmeters and increased—see 2.2.2 Oversampling. many applications involving slow transducers. They can offer up to 22 bits of resolution and an implicit Advantages rejection of high frequency background noise, e.g., AC mains pickup. 1. High resolution – up to 20 Bits of Peak-to-Peak resolution possible Advantages 2. Relatively higher conversion rate when compared to Integrating ADCs and VFCs. 1. High accuracy 3. Can easily be fabricated on mixed signal CMOS 2. Relatively low power consumption to provide low cost high resolution signal 3. Relatively low cost solution acquisition and digital processing on a single chip. Disadvantages 4. Due to the high over sampling techniques used, minimal filtering of the transducer signal is 1. Limited conversion rate. Accuracy decreases as required and virtually no signal conditioning. conversion rate increases. Typical accuracy at 100 – 300 conversions per second is 12 Bits. Disadvantages Voltage-to-Frequency Converters 1. Higher order modulators are required when The VFC performs an indirect conversion by first faster conversion rates are needed.