The requirements for a 3-phase multi-function power meter are always changing, as these meters are incorporating more advanced features and more new functions each year. New 3-phase multi-function energy meters can calculate active and reactive energy in four quadrants, as well as the base wave and harmonic wave energy. These meters can also perform harmonic analysis and intelligent data reading functions. Active energy measurement requires a class 0.2S energy meter, while typically, class 0.5S is required for reactive energy. Any new 3-phase multi-function power meter must be able to support the different protocols and requirements specific to different regions of the world. In developing countries like China, a low bill of materials (BOM) cost is especially important. In addition to meeting the ever-increasing technical requirements of the product, designers must also be conscious of the overall BOM cost.
THE CURRENT 3-PHASE POWER METER SOLUTIONS
Energy Meter IC + MCU: An energy meter integrated circuit (IC) typically includes ADCs and a reference circuit, and is capable of performing all signal processing operations like active, reactive, and apparent energy measurement in 3-phase configurations. These devices will communicate both frequency and logic information to the meter control unit (MCU). Typical energy ICs are ADE7758 and ADE7753 from Analog Devices Inc. (ADI). The advantages of this configuration are a simple design and low cost. The disadvantages are fixed measurement functions. These designs are typically suitable for low and middle end 3-phase multi-function power meters. 16Bit A/D + CPU: These configurations will typically employ a 6-channel analog-to-digital converter (ADC) to sample 3 phases of voltage and 3 phases of current. All processing and protocol is handled by the central processing unit (CPU).
One common configuration is the Analog Devices AD73360 (6 channel ADC) connected to an ARM CPU. The advantages of this solution are a clean system architecture, higher precision and expandability. System performance and feature complexity, however, are typically limited by the low numeric processing capabilities of the MCU. It can thus be used in higher-end metering applications, but will be limited in its complex harmonic analysis capabilities. AD73360 + DSP + MCU: The most common numeric operations in digital energy metering are multiplication and addition, which are also the primary arithmetic building blocks of digital filters and fast-Fourier transforms (FFTs).
Digital signal processor (DSP) cores typically have the multiplyaccumulator (MAC) that is used extensively in energy metering algorithms. In this solution, energy measurement and high precision harmonic analysis will be processed by the DSP. The MCU will display all data and perform all communication protocols, so the main advantages of this architecture are the higher precision and the increased expandability and flexibility. However, this solution has a more complicated system architecture, as we are now dealing with two processors. This requires more software to be written and means a higher BOM cost. One common system configuration is the AD73360 + ADSP-2185 + MCU.
0.2S 3-PHASE MULTI-FUNCTION POWER METER SOLUTION BASED ON BLACKFIN PROCESSOR
System Block Diagram: The key for performance is the DSP in the above solutions. As DSPs add both cost and power consumption to the design, selection of an optimal DSP is essential. We recommend a high performance, low power consumption and low cost processor – the Blackfin processor for class 0.2S 3-phase multi-function energy meter solutions. The system block diagram is shown in Figure 1. This is a typical configuration for a 3-phase multi-function power meter. In this system, the ADSP-BF532/531 and AD73360 perform voltage and current sampling, energy measurement and harmonic analysis. The energy calculations from the Blackfin will be sent to the MCU, which will display, save and prepare them for communication by protocol.
Precision: This energy meter must be capable of a 0.2% resolution between 400% and 1% of the current transformer (CT) measuring area in order to satisfy class 0.2S energy meter requirements. Meeting these requirements will also allow compliance with class 0.5S energy meter standards. The system must include nonlinear compensation for the CT and compensation of the filter gain parameter.
• Active, reactive and apparent power of four quadrants
• Active, reactive and apparent energy of four quadrants
• Power factor for every phase and total power factor
• Irms and Vrms
• Frequency values
• Over 21X harmonic analysis
The 400MHz ADSP-BF532 Blackfin processor can fulfil all meter functions. Despite the fact that it is a high-performance processor, the Blackfin operates on very little current, and also supports ultra-low power ‘deep sleep’ and ‘hibernate’ modes. According to the National Standard of Energy Meter, the meter must still be active when the power is off, in order to continually perform energy metering. This means that battery operation is essential. When operating from a battery, the Blackfin processor can be placed into an ultra-low-power sleep mode to maximise the battery life, and all maintenance functions can be performed by a simple low-cost, low-power MCU.
THE HIGH-PERFORMANCE BLACKFIN PROCESSOR
The advanced architecture: The Blackfin core was developed jointly by Analog Devices and Intel, and was designed to be equally suitable as a micro-controller and as a DSP. Thus it is commonly referred to as the micro-signal architecture. The Blackfin core is based on an advanced dual-MAC Harvard architecture. It combines sophisticated signal processing capabilities and high performance RISC-like CPU functionality with the ease-of-use attributes found in general purpose micro-controllers. Thus the Blackfin processor performs equally well in both signal processing and control processing applications, removing the need for separate heterogeneous processors. This is magnetic for developers who want it for both signal processing and control processing applications. For this application, we recommend BF532 or BF531; they offer the highest performance at the lowest price in this market space.
These devices both operate at 400 MHz with different amounts of memory. They are pin-compatible and cost between $5 and $8 at 10K quantities. Both BF532 and BF531 include a number of standard peripherals, including SPORT, SPI-, UART, three multi-function timers and RTC. Direct connection to AD73360 can be achieved using SPORT. Extended Development Tools and Software Resources: Analog Devices has developed a DSP development tool, VisualDSP++, which provides full support for the Blackfin processor. These tools include Integrated Development and Debugging Environment (IDDE), C/C++ compilers, C and DSP libraries. We can use these source codes and libraries and change them for material applications. If necessary, the Blackfin processor can carry out digital filter and harmonic analysis in real time – something that is difficult for other CPUs. The comparison of 256-point complex FFT on Blackfin processor and ARM9E is as follows:
Using the Blackfin processor’s high performance, we can calculate and analyse all real-time harmonic parameters. We can also record the fault waves while detecting the fault. Finally, the Blackfin processor’s dynamic power management offers a flexible, software-controlled environment that delivers the required amount of performance to the processor via independent, dynamic variation of both voltage and frequency. In this solution, the Blackfin is running at 200Mhz and dissipates less than 150mW of power.
THE FACTORS THAT AFFECT PRECISION
Improving A/D result precision: Class 0.2S requires that the low load precision is less than 0.4% at 1% Ib (CT measurement area). In practice, it should be controlled inside 0.2%. If the measuring current area from 400% Ib to 1% Ib can achieve 0.2% precision, the A/D converter must be 18 bits or 19 bits for class 0.2S energy meter. In fact, AD73360 only has 16 bits revolution. To improve precision, we must increase the sampling frequency to 8KHz. The over-sampling will satisfy the precision requirement.
Multi-measurement methods for reactive power: There are a number of reactive energy measurement methods that can be used in this solution, such as sample delaying, trigonometry, and phase shifted 90, which are all for the base wave reactive energy. Hilbert filters to calculate the harmonic energy will be added in the next generation product. I
mproving the harmonic leakage of spectrum produced by FFT: The FFT is often used to perform the harmonic analysis in a 3-phase energy meter, but the harmonic leakage of spectrum and the barrier effect must be considered when using FFT. These will cause some errors in frequency, amplitude and phase, meaning that the correct harmonic measurement cannot be achieved. We will use windowed interpolation to compensate for these errors.
This solution (Figure 3) was jointly designed by ADI and Changsha Xinzhu of China. Its main advantages are a low BOM, high performance and strong EMC design. By adding a GPRS module, it is easy to integrate the load control device or power management terminal functionality.