Proceedings of the 2017 IEEE International Conference on Information and Automation (ICIA) Macau SAR, China, July 2017 Research on the Maximum Current Automatic CurrentSharing Control Based on DSP Qingqing Li and Shulin Liu Huisan Xu and Xiao Wang Academy of Electrical and Control Engineering Xi’an University of Science and Technology Xi’an, Shaanxi Province, China [email protected] [email protected] Academy of Electrical and Control Engineering Xi’an University of Science and Technology Xi’an, Shaanxi Province, China [email protected] [email protected] Abstract - A maximum current automatic current-sharing control method is implemented for high performance control of paralleling Buck converters based on DSP controller. The system samples the output current and sends it to a high precision instrumentation amplifier INA128 to amplify. Then, the amplified signal is sent to each parallel module by ADC module through CAN bus, and transmitted back to DSP after decoded. The DSP compares the received sampling current data and selects a main module with the maximum output current. The main module calculates the average current and sends it back to the DSP. Meanwhile, the main module voltage is set as the output voltage of current-sharing bus and compared with the reference voltage to steady output voltage. Through comparing the current-sharing value from each module and output current, DSP calculates and sends the current error increment to advanced PI regulator and changes the data in CMPR of ePWM module to control the duty ratio of MOSFET. Then, a steady current-sharing output of parallel Buck power supplies system could achieved. The parallel Buck modules are used for digital current-sharing experiment and relevant parameters are tested, which shows the prototype can achieve bus current-sharing excellently. sharing algorithm and DSC loop adjustment program module. The function of CAN communication module is to exchange data of every DC power modules in the distributed power supply system and send modules current data to currentsharing algorithm module. And the algorithm selects a main module with the maximum output current from divided modules. DSC loop processing module sets the maximum current as the reference current and regulates the module current loop based on the difference between the reference current and the module output. Through adjusting constantly, the system would be able to achieve current-sharing . Keywords-digital control; maximum current current-sharing method; advanced PI algorithm II. TECHNOLOGY SOLUTION The current-sharing of multiple paralleling switch power system supplies is shown in Fig. 1. automatic I. INTRODUCTION Nowadays, power electronic technology is applied in wide fields such as communication, spatial science and computer science. And there has been a growing demand of power source with higher power, higher power density and reliability. In single power system, there would be difficulties in terms of choosing power devices with achieving higher switching frequency and higher power density. Furthermore, the whole system would collapse if the power fails in single power system . Therefore, paralleling power is an important direction of high power supply technology. Similar to single power supply, paralleling power supplies should be able to run reliably with constant output voltage while input voltage and output load change . Thus, measures need to be taken to prevent electrical and heat stress unbalance caused by the load difference of each paralleling power supply . Current-Sharing (CS) is a technique that could provide equal current distribution of the load current among the parallel-connected power supplies. In this research, digital current-sharing is realized with the method of maximum current automatic current-sharing. Digital CS design comprises CAN communication module, current- 978-1-5386-3154-6/17/$31.00 ©2017 IEEE Fig. 1 Parallel Current-sharing of multiple switching power supply. A. The Principle of Automatic Maximum Current - Sharing Method The hardware structure block diagram of paralleling power system based on the maximum current-sharing technology is shown in Fig. 2. The experiment platform consists of main power conversion circuit, driving circuit of MOSFET, current sensing circuit, voltage sampling circuit, auxiliary power supply circuit, over-current and over-voltage protection circuit and control circuit based on TMS320F2803. B. The Principle of Automatic Maximum Current - Sharing Method Maximum current CS method is also known as automatic main-minor setting method or democratic CS method. In paralleling power supply system, the output of every separated parallel module is mounted on the current- sharing bus through an external diode. The current-sharing connection method is shown in Fig. 3 . 1044 CAN总总 • • CAN总总 均均均总线均 • • • • • • • • Fig. 2 Hardware structure block diagram achieved by automatic maximum current CS method. The block diagram of main circuit current-sharing control is shown in Fig. 5. VˆAC − Fr + GCS VˆA Vref + Fig. 3 Method of connection between separated parallel module and currentsharing bus. − In the process of the automatic maximum current-sharing, Vb (the voltage of current-sharing bus) reflects the voltage of the module with the highest current output. In steady state, parallel power modules distribute balance current to load. If the output current of one module increased and became the maximum current, the voltage of this branch Vi would also increased, and it will be set as bus voltage. In other words, Vb is set equal to Vimax. Meanwhile, this module would be set as the main module, and others become minor modules. Then, those voltage data of minor modules are compared with the Vb of current-sharing bus, and the output increments make effect on the reference voltage Vr of voltage closed loop to regulate the reference voltage Vr' of voltage amplifier. In this way, automation maximum current CS can be achieved . The circuit control principle of automatic maximum current CS is shown in Fig. 4. Gv ( s ) 1 Um Gvd(s) Vout Fig. 5 Main circuit current sharing control block diagram. In Fig. 5, VAC is current sharing signal, VA is current sharing error signal, H is the gain of output current, Gcs is the gain of current sharing, Fr is the transfer function of module load current convert into the voltage process . III. HARDWARE DESIGN TMS320F28035 is adopted to improve the performance of experimental prototype, which has higher efficiency CPU that frequency can reach 60 MHz per second, compared to traditional DSP chip. The block diagram of experimental prototype circuit is shown in Fig. 6. Fig. 4 Circuit control principle of maximum current-sharing. C. Small Signal Control Model of Maximum Current-sharing Considering the high current sharing requirement of designed system, the built model with voltage closed loop is Fig. 6 Block diagram of experimental prototype circuit based on DSP. 1045 In this research, PCA82C250 is implemented as the CAN bus driver. It serves as the interface between CAN protocol controller and physical bus, and enables CAN bus with the capability of differential transmission and receiving . The eCAN interface circuit is shown in Fig. 7. CPU clock source TIMES0/1/2. In this paper, EPWM cycle interrupt is adopted to start A/D conversion. The A/D interrupt service routine will be triggered by the generated A/D interrupt signal after the conversion is completed. In the process of A/D interrupt service routine, CAN communication service program, current-sharing regulation and DSC loop adjustment program, which affect the performance of the whole system. And these three sub-programs are fundamental for the digital current-sharing adjustment algorithm. The interrupt service program flowchart is shown in Fig. 9. Fig. 7 eCAN interface circuit. IV. SOFTWARE RESULTS The design of system software consists of main program and interrupt response service program. The main program is responsible for system initialization and settings. And functions constant setting, variable setting, I/O, interrupt vector, A/D module and key register of HRPWM are involved in these two items. Due to the control system is mainly realized by A/D interrupt service routine, the initialized system goes into circle until receiving interrupt signal is received. Then, the CAN communication service program, the current-sharing regulation and the DSC loop adjustment program would be called by the main program in A/D interrupt procedure. Then, the controlled quantity calculation and the output updates can be accomplished. The main program flowchart is shown in Fig. 8. Fig. 9 A/D interrupt service program flowchart. B. eCAN Communication Service Program In this design, TMS320F28035 sends current-sharing data to each divided module through CAN bus. Each module calculates the fine adjustment quantity by comparing the branch and CAN current, the final output current of each module can be determined by loop regulation, and the output current of all sub-modules of the system can be realized. Finally, system could achieve current-sharing with loop adjustment. The program flowchart of eCAN communication service is shown in Fig. 10. The main purpose of CAN communication services program is to send ID from DSP controller to sub-modules. After ID is received, the outputs of each divided modules are compared and the main module with the maximum output current is selected. This main module determines the voltage increment by comparing the feedback voltage and reference voltage. Then DSP sends the voltage regulation command to Fig. 8 Software main program flowchart. A. Interrupt Service Routine of A/D A/D interrupt subroutine is the core of the whole digital control, as it performs tasks such as calculating real-time PID parameters in every cycle. A/D conversion can be triggered by S/W, EPWM, external interrupt trigger, ADCINT1 1/2 and 1046 regulate output voltage. Meanwhile, minor-module calculates the number of online modules and receives the maximum current to prepare current-sharing. The CAN communication module rechecks the order of divided modules per 500ms to update the main module and ensure the number of online modules is same as the initial situation. Through applying the above method, the reliability of the system can be enhanced, and power sub-modules can be achieved data interaction with the assist of CAN bus . sharing not only performs digital current-sharing adjustment algorithm but also ensures the system minimum flow accuracy indicators and achieves higher stability and reliability . Digital current-sharing adjustment algorithm flowchart is shown in Fig. 11. Fig. 11 Digital current-sharing adjustment algorithm flowchart. D. DSP Loop Adjustment Program The DSC loop procedure regulates the data in compareregister of HRPWM module based on the current-sharing increment. Through changing duty ratio of switching tube in DC/DC circuit, the final output current converges to expected value. The DSC loop adjustment program flowchart is shown in Fig. 12. Fig. 10 eCAN communication service program flowchart. C. Current-sharing Adjustment Algorithm DSP receive the current value from the CAN bus in the implementation of current-sharing control algorithm. DSP determines the current adjustment increment based on the maximum current. If the real module current is smaller than maximum current, the output current would increased; if the real current is same as the average current or there is only a fine difference, DSP would find the difference between real output current and maximum current and compare this difference with system accuracy requirement of currentsharing. Among this comparing, if the difference will less than the requirements, it indicates that the digital current-sharing has been completed. Otherwise, system would be fine adjusted to ensure system accuracy. The strategy of digital current- Fig. 12 DSC loop adjustment program flowchart. It can be seen from Fig. 12 that the principle of currentsharing output based on current increment in the various submodules of the parallel current-sharing system. DSC loop 1047 adjustment strategy has strong stability and robustness. The control quality is insensitive to the changing of controlled object . An advanced digital PI algorithm is adopted in increment processing to restrain integral saturation problem and improve the stability of parallel system. Integral method with extinction limiting is adopted in this research to restrain integral saturation phenomenon. The basic principle is that system weakens the integral accumulation and ignores the accumulating command while the control volume entering saturated region. In other words, the system judges whether the last u(n-1) has beyond the limit before calculating u(n). If u(n-1) is larger than the umax, system would only performs negative deviation accumulation; if u(n-1) is smaller than umin, system would only performs positive deviation accumulation . This can be represented as follows. E. The Advanced Digital PI Algorithm Digital controller is based on discrete sampling control method. It calculates increments only with the deviation data derived from sampling times. Thus, we have to discretize the analog PID algorithm . Discretization is achieved by using bilinear transformation that sampling period is 10 μs. The difference equation is given by u ( n) = u ( n − 1) + k1e( n) + k 2 [e( n) − e(n − 1)] (1) (3) u (n − 1) < umin In this formula, umax is the biggest value for the period register; umin is the minimum value for period register. In the above formula, u(n) is the PI regulator output of the Nth sampling. Because of the regulation of output voltage and output current is achieved by the duty ratio of DC/DC system. And DSP28035 changes the duty ratio through changing the CMPR data of HRPWM module. Thus, the u(n) here is the value of compare register CMPR; е(n) is the error between the voltage and reference voltage of the Nth sampling. The expression is given as follows. e( n) = V f − Vref u (n − 1) > umax u u (n) = max umin V. ANALYSIS OF CURRENT-SHARING EXPERIMENTAL RESULTS A current-sharing performance test of power supplies which in parallel in steady state is made based on the prototype which is under 24V rated input voltage. 9 sets of data was measured and listed in table I, including bus output current, output current of each modules, the relative error of currentsharing. (2) TABLE I The Result of Current-Sharing Performance Test In order to avoid the integral saturation, the improved PI algorithm is used in the program, and the control algorithm flowchart is shown in Fig. 13. 7.91 1# output current /A 0.20 2# output current /A 0.21 7.97 0.60 7.98 0.80 8.01 7.92 system output current /A relative error /% 0.41 4.8 0.61 1.21 3.3 0.82 1.62 2.5 0.99 1.01 2.00 2.0 1.19 1.21 2.40 1.7 8.06 1.38 1.40 2.78 1.4 7.95 1.59 1.61 3.20 1.3 7.92 1.78 1.80 3.58 1.1 7.96 2.00 2.02 4.02 0.9 system output voltage /V From the table I, the relative error of current-sharing can be expressed by CSerror = N ⋅ ΔI o max Io (4) The voltage adjustment rate of paralleling power supplies is calculated based on table I. The result is followed. SV = 8 − 7.91 × 100% = 1.125% ≤ 3% 8 (5) According to the analysis results, these two paralleling power modules have realized current-sharing with the proportion nearly 1:1. And the relative error of current-sharing is less than 5 percent of the expected relative error, which meets the expected requirements of technical indicators. Fig. 13 Advanced PI control algorithm flowchart. 1048  VІ. CONCLUSION A Buck converter experimental prototype based on the maximum current automatic current-sharing method is made in this paper, and the practical application of digital currentsharing technology is analyzed. An advanced digital PI algorithm is used for restrains integral saturation problem, and TMS320F28035 is implemented in this prototype to improve the stability, real-time and reduce the cost of parallel system. According to the current-sharing performance test of paralleled power converter, the prototype is able to realize currentsharing steady and reliably. The measured relative error of current sharing is less than 5% which means the prototype can achieve bus current-sharing excellently.      REFERENCES      S. L. Liu, J. Liu, Analysis and design of the switching converter, Beijing: China Machine Press, 2010. Y. F. GAO, X. J. Hu, T. Chen, Y. C. Fan, “Parallel current-sharing system in blocking switch power supply,” Chinese Journal of Power Sources, Vol. 35(2), pp.210-212, Feb. 2011. J. J. Wang, “A common sharing method for current and flux-linkage control of switched reluctance motor,” Electric Power Systems Research, vol. 131, pp. 19–30, 2016. D. Y. Qiu, B. Zhang, “Study of paralleled Buck converters with improved automatic current-sharing technique,” Transactions of China Electrotechnical Society, Vol.20, pp. 41-47, Oct. 2005.   1049 D. K. Cheng, Y. S. Lee, Y. Chen, “A current-sharing interface circuit with new current-sharing technique.” IEEE Trans. on Power Electronics, Vol. 20(l), pp. 35~43, 2005. T. F. Zhang, “Digital technique of parallel balanced current in switching power supply,” Journal of Huaihai Institute of Technology(Natural Sciences Edition), Vol. 15, pp. 29-32, Mar. 2006. C. J. Zhang, G. T. Chen, F. Zu, W. Y. Wu, “An interactive following Current-Sharing control strategy for single phase paralleled inverters in full digital,” Proceedings of the CSEE, Vol. 26, pp. 63-66, May. 2006. J. Ma, Q. Du, J. Luo, B. J. Qi, “Research of automatic current sharing technology of parallel system for switching power supply,” Chinese Journal of Power Sources, Vol. 35, pp. 969-973, Aug. 2011. C. J. Zhang, G. T. Chen, F. Zu, W. Y. Wu, “An interactive following Current-Sharing control strategy for single phase paralleled inverters in full digital,” Proceedings of the CSEE, Vol. 26, pp. 63-66, May. 2006. C. H. Cheng, P. J. Cheng, M. J. Xie, “Current sharing of paralleled DC– DC converters using GA-based PID controllers,” Expert Systems with Applications, Vol. 37, pp. 733-740, 2010. Y. Gao, S. C. Tong, Y. M. Lee, “Fuzzy adaptive output feedback DSC design for SISO nonlinear stochastic systems with unknown control directions and dead-zones,” Neurocomputing, Vol. 167, pp. 187-194, Nov. 2015. Aruchamy Sakthivel, P. Vijayakumarb, A. Senthilkumara, et al, “Experimental investigations on Ant Colony Optimized PI control algorithm for Shunt Active Power Filter to improve Power Quality,” Control Engineering Practice, Vol. 42, pp. 153–169,Sep. 2015. X. R. Jia, X. Z. Zhang, “Fuzzy PID control based on given system,” Microcomputer & Its Applications, Vol. 30, pp. 79-81, 2011.