Today, battery-powered motor drive solutions can often provide hundreds of watts of power with very low operating voltages. In these applications, in order to ensure the energy efficiency and reliability of the entire system, the current of the motor-driven equipment must be properly managed. In fact, the motor current may exceed tens of amperes, leading to an increase in the internal power dissipation of the inverter. Applying higher power to the inverter components will cause the inverter’s operating temperature to rise and performance degradation. If it exceeds the maximum allowable rated power, it may even stop working suddenly. Optimizing thermal performance while reducing size is an important part of the inverter design process. If not handled properly, hidden dangers may be buried. Continuously improving prototype production with field verification methods can solve this problem. However, electro-thermal evaluation is a completely separate process, and the electro-thermal coupling effect has never been considered in the design process, because this will lead to multiple repeated designs and prolong the product. Time to market. At present, there is a more effective alternative method of electric heating evaluation, which is to use modern simulation technology to optimize the electric heating performance of the motor control system. The Cadence® Celsius™ Thermal Solver temperature simulator is an industry-leading electro-thermal co-simulation software for system analysis, which can comprehensively and accurately evaluate the design performance from both electrical and thermal perspectives in just a few minutes. As the world’s leading manufacturer of industrial motor control integrated circuits, STMicroelectronics used Celsius™ software to improve the thermal performance of the EVALSTDRIVE101 evaluation board and developed a three-phase brushless motor inverter with an output current of up to 15 Arms, designed for end applications The staff develops the inverter to provide a reference. In this article, we take this opportunity to explain how to reduce the workload of thermal optimization while making EVALSTDRIVE101 a production-level solution.
EVALSTDRIVE101 is based on a 75 V three-half-bridge gate driver STDRIVE101 and six STL110N10F7 power MOSFET switches connected into three half-bridges. STDRIVE101 adopts 4×4mm quad flat no-lead (QFN) package with integrated safety protection function, which is very suitable for battery-powered solutions. Celsius™ significantly simplifies the thermoelectric performance optimization process of EVALSTDRIVE101, enabling a compact and reliable design in a short time. The simulation results shown below are used to repeatedly adjust the position of components, improve the shape of the board and traces, adjust the thickness of the board, add or remove through holes, and finally get a production-level inverter solution. After optimization, EVALSTDRIVE101 is a four-layer PCB with a copper thickness of 2 oz. It is 11.4 cm wide and 9 cm high. It uses a 36 V battery voltage to provide up to 15 Arms of current to the load. From a thermal point of view, the most critical part of EVALSTDRIVE101 is the power stage area, which includes power MOSFET switch tubes, current-sense resistors, bypass ceramic capacitors, large-capacity electrolytic capacitors and output ports. The layout of this part has been greatly reduced to only half the size of the entire circuit board, which is 50 cm2. Here, the placement and wiring of the MOSFET have been carefully considered, because most of the power loss of the inverter is caused by these switching tubes during operation. The copper area of the drain terminals of all MOSFETs is the largest on the top layer, and the other layers should be made as large or larger as possible to improve the heat transfer efficiency of heat conduction to the bottom surface. In this way, both the front and back of the circuit board contribute to natural air convection and heat radiation. The through hole with a diameter of 0.5 mm is responsible for the electrical connection and heat transfer between the different layers, promotes air flow and improves the cooling effect. The through-hole grid is located directly below the exposed pad of the MOSFET, but the diameter of the through-hole is reduced to 0.3 mm to prevent the solder paste from reflowing in the hole.
Power consumption estimation
The thermal optimization process of EVALSTDRIVE101 starts with evaluating the power dissipation during the operation of the inverter, which is an input terminal of the temperature simulator. Inverter losses are divided into two categories: power loss caused by the Joule effect in the circuit board traces and power loss caused by Electronic components. Although Celsius™ can accurately calculate current density and circuit board loss by directly importing circuit board layout data, the loss caused by electronic components must also be considered. Although the circuit simulator can provide very accurate results, we decided to use a simplified formula to calculate a reasonable power loss and propose an approximate value. In fact, the manufacturer may not be able to obtain the electrical model of the component, and because of the lack of modeling data, it is difficult or impossible to model from scratch, and the formula we provide only requires the basic information of the product data sheet. Excluding secondary phenomena, the main reason that causes the inverter to dissipate power consumption is the power loss inside the current-sense resistor P_sh and the MOSFET. These losses include: conduction loss P_cond, switching loss P_sw and diode voltage drop loss P_dt:
The estimated power dissipation of each MOSFET is 1.303 W, and the estimated power dissipation of each current-sense resistor is 0.281 W.
Celsius™ allows designers to perform thermal simulation experiments, including electrical analysis of the system, to Display the current density and voltage drop of traces and vias. These simulation tests require that the designer must use the circuit model in the system to define the relevant current loop. Figure 1 shows the circuit model used by each half-bridge of EVALSTDRIVE101. The model consists of two constant current generators and three bypass MOSFETs and short circuits of current-sense resistors located between the output and the power supply input. These two current loops are very close to the actual average current of the entire power rail and ground plane, while the output path current is slightly higher, which is convenient for evaluating design toughness. Figures 2 and 3 show the voltage drop and current density of the EVALSTDRIVE101 with a current of 15 Arms. The voltage drop of the ground reference voltage highlights that the layout of this board has been specially optimized without bottlenecks, and the voltage drops at the output terminals of U, V and W are very balanced at 43 mV, 39 mV and 34 mV. The U output has the largest voltage drop, while the W output has the lowest voltage drop of the three because the path length from the W port to the power connector is relatively short. The current is evenly distributed in each path, and the average density is less than 15 A/mm2, which is the recommended power value for the trace size. Some areas near the MOSFETs, shunt resistors, and connectors are red, which represents a higher current density because the terminals of these components are smaller than the power traces below. However, the maximum current density is much lower than the limit of 50 A/mm2, which will not cause reliability problems in practical applications.
The simulator enables designers to install and run steady-state simulation or transient simulation tests. The steady-state simulation provides a 2D temperature map of the board layers and components, while the transient simulation provides the temperature map and heating curve at each simulation time, but the simulation time is longer. The steady-state simulation tool can be used for transient simulation, but it is also necessary to define the dissipation power function for the component. Transient simulation is suitable for defining operating conditions and evaluating the time required to reach a steady-state temperature for systems where power supplies are not working at the same time.
The simulation experiment condition of EVALSTDRIVE101 is 28 °C ambient temperature, the heat transfer coefficient is used as the boundary condition, the device analysis uses the dual resistance thermal model instead of the detailed thermal model such as Delphi, and the model can be obtained directly from the component data manual, but the simulation will be slightly sacrificed Accuracy. Figure 4 shows the steady-state simulation results of EVALSTDRIVE101, and Figure 5 shows the transient simulation results. The transient simulation uses a step power function to enable all MOSEFT and current-sense resistors in zero time. The simulation result confirms that the U half-bridge area is the hottest area on the circuit board. Q1 MOSFET (high-side) has a temperature of 94.06 °C, followed by Q4 MOSFET (low-side), R24 and R23 current-sense resistors at 93.99 °C, 85.34 °C and 85.58 °C, respectively.
Thermal characterization experimental device
EVALSTDRIVE101 thermal performance experimental characterization is done on the assembled circuit board. In order to facilitate the experiment, instead of using the motor connected to the brake table, consider using an equivalent test table, as shown in Figure 6.
EVALSTDRIVE101 is connected to the control board to generate the required drive signals and placed in the plexiglass box to obtain air convection cooling and avoid accidental air convection. A thermal imaging camera (TVS-200 of Japan Avionics Corporation) was placed on the top of the box, and all the circuit boards were put into the shooting frame through a hole in the lid of the box. The output terminal of the circuit board is connected to a three-phase load, and the drive system uses a 36 V power supply. The load is composed of three coils connected in a star-shaped structure to simulate the real working characteristics of the motor. Each coil has a saturation current of 30 A, an inductance of 300 µH, and a parasitic resistance of 25 mΩ. The low parasitic resistance greatly reduces the Joule heating effect inside the coil, which is beneficial to the lossless power transmission between the circuit board and the load. Appropriate sinusoidal voltage is applied through the control board to generate three 15 Arms sinusoidal currents inside the coil. Using this method, the working environment of the power stage is very close to the working conditions of the actual application of the motor drive, and the advantage is that no control loop is required.
Power loss measurement
The data accuracy of the dissipated power of each device in the power stage is undoubtedly a factor that affects the simulation results. The data of MOSFET and current-sense resistor are calculated using simplified formulas, so approximate values are proposed. Measure the circuit board to evaluate the quantization error of the dissipated power. The measured value of the power loss Ploss of the circuit board is the difference between the input power Pin and the output power supplied to the load by the three output terminals PoutU, PoutV, and PoutW. Use an oscilloscope (HDO6104-MS from Teledyne LeCroy) to measure and use appropriate mathematical functions in the waveform: first, calculate the product of the voltage and current at each measurement point point by point; then, calculate the number of sine cycles within an integer Average power. The following table lists the measurement data under the ambient temperature and the high temperature measurement results when the power level reaches the steady-state condition. It also gives the circuit board power dissipation estimated by the formula.
The results show that the measured value and the estimated value are very close, consistent with the proposed approximation. At room temperature, the formula overestimates the measured value by 1.5%, and at high temperature conditions, it underestimates the measured value by approximately 3.9%. This result is consistent with the variability of the MOSFET on-resistance and current-sense resistance, because the nominal value is used in the calculation. As the coil resistance and MOSFET resistance increase with temperature, the high temperature power value is higher than the room temperature power value, which is in line with expectations. The data also shows a difference in the measured power of the three outputs. This phenomenon is caused by the unbalanced three-phase load, because the L and R values of each coil are slightly different. However, the effect of this effect is negligible, because the observed difference is lower than the difference between the measurement and the estimate.
The sinusoidal current generated in the load is synchronized with the capture and photographing of the thermal imager. The thermal imaging camera is set to take a thermal image every 15 seconds, and each photo contains three temperature marks of components Q1, Q4 and R23. The system remains in working condition until steady-state conditions are reached after approximately 25 minutes. At the end of the test, it was detected that the ambient temperature in the box was about 28°C. Figure 7 shows the board temperature rise transient from the temperature marker, and Figure 8 shows the final temperature on the board. The measurement results show that Q1 MOSFET is the hottest component in the entire circuit board, with a temperature of 93.8°C, while the resistance of Q4 MOSFET and R23 reached 91.7°C and 82.6°C, respectively. According to the Celsius™ simulation results above, Q1 MOSFET is 94.06°C, Q4 MOSFET is 93.99°C, and R23 is 85.58°C, which is very close to the measurement result. By directly comparing Figure 5 and Figure 7, it is not difficult to find that the transient time constant of heat dissipation is also as high as the same.
STMicroelectronics recently released the EVALSTDRIVE101 evaluation board developed using the Cadence®Celsius™ Thermal Solver temperature simulator. The board circuit board can drive high-power low-voltage three-phase brushless motors for battery-powered equipment. This board includes a 50 cm2 compact power stage that can provide more than 15 Arms of current to the motor without the need for a radiator or additional cooling equipment. Using different simulation functions inside the temperature simulator can not only predict the temperature distribution of the circuit board and the hot spots of the power stage components, but also describe the voltage drop and current density of the power trace in detail, which is difficult or impossible to measure through experiments get. In the entire development process from the initial stage of design to finalization, the simulation results allow developers to quickly optimize the layout of the circuit board, adjust the position of components, and improve layout defects. The thermal characterization test of the infrared thermal imaging camera shows that the steady-state temperature and the transient temperature curve have good consistency between the simulated value and the measured value, which proves that the circuit board has excellent performance, and the temperature simulator can effectively help designers reduce the design Margin to speed up product launch.