
Software-related EMI Behavior of Embedded Microcontroller Shi-Yi Yuan Wei-Yen Chung Cheng-Chang Chen Chiu-Kuo Chen Department of Department of Bureau of Standards, Bureau of Standards, Communication Communication Metrology and Inspection, Metrology and Inspection, Engineering, Engineering, M.O.E.A, M.O.E.A, Feng Chia University, Feng Chia University, Taipei, Taiwan, R.O.C., Taipei, Taiwan, R.O.C., Taichung, Taiwan, R.O.C. Taichung, Taiwan, R.O.C. [email protected] [email protected] [email protected] [email protected] Abstract—This paper studies software-related EMI. microcontroller’s conducted electromagnetic interference (SW-cEMI) behaviors when microcontrollers are executing From EMI measurement standards [8], software is basically different programs of the same functionality. The SW-cEMI an infinite loop. These setups are for EMI standard behaviors of different program types, clock sources, and measurement and configuration but not for practical programmable PLL settings are measured and observed. Two applications. In this paper, EMI measurement based on major trends of microcontrollers are used as devices under test. different software implementations with the same functionality The result shows that 10-12 dB cEMI differences even on a are observed. simple LED blinking function implemented by different program types. To the best of our knowledge, this work is the first Based on this work, some software design approaches observation on EMI phenomena of software implementations on overcoming or at least partially compensating some parasitic microcontroller’s programs. Some advises are given to μC emission effects are principally observed. To the best of our programmers for cEMI reduction.. knowledge, this work is the first observation on EMI phenomena of software implementation behavior on μC’s Keywords—Software-related EMI, embedded PLL cEMI conducted EMI (cEMI). control, microcontroller EMI programming This paper is organized as follows: section 2 describes the I. INTRODUCTION program design for the target μCs. Section 3 details the target μCs’ characteristics, measurement setups, and measurement As technology advances, more and more electronic procedures. Experimental results are described in section 4. products are applied to daily lives. Electronic product more Section 5 gives our conclusions. complicated than ever and demands more functionalities in one single device. The increasing transistor count and higher II. PROGRAM TYPES AND PLL CONFIGURATIONS system clock rates result in serious mutual interference. In this paper, we use two program implementation types Pacemaker, electrocardiogram, electric wheelchairs, or other among different μCs: Infinite loop-type and interrupt-type medical-related equipment may be suffered from the chance of program implementations. malfunction by these technology advances. Therefore, component-level electromagnetic interference (EMI) becomes Fig. 1 shows the infinite loop type program implementation. one of most important issues for these devices. After initial PLL and clock setups, program goes into an infinite loop of one specially designed function. All through To solve the EMI problems, there are many techniques this paper, the designed function is a simple LED blinking proposed. Most of them are based on hardware level EMI function set to 2 Hz of any type program implantations. reduction such as PLL designs [1][2][3], or PCB layout designs [4][5][6]. The optimization of EMI problems by hardware level techniques should be done as much as possible. To the best of our knowledge, little research addresses its attention on the relationship between the software functionalities of microcontroller (μC) and EMI behaviors [7], especially on the configurations of the embedded phase lock loop (PLL) inside a μC. Generally, a modern μC is equipped with at least one programmable embedded phase lock loop (PLL). A PLL is a component that generates clock signals for μC. PLL can keep Fig. 1 Infinite loop type program implementation system clock accuracy from jitter and instability. It also controls clock sources and clock rate in a digital system. Fig. 2 shows a LED blinking program of the interrupt type Nowadays, the state of PLL of an embedded system can be program implantation (also in 2 Hz). An interrupt service easily controlled by system software. Therefore, such software routine (ISR) and initial PLL setups are firstly executed. After can control the PLL behaviors and further control the system the procedures are initially executed, the program goes into an This paper is supported by Bureau of Standard, Metrology and Inspection, Taiwan, Republic of China. 978-1-4799-5545-9/14/$31.00 ©2014 IEEE 118 infinite loop of NOP (no-operation). The LED blinking Table. 1 All clock rates for PIC18F4550 test programs function is designed inside the ISR. Test Program Name loop type interruption type Clock Rate 48 MHz, external mode pic_loop_48M_ext pic_intpt_48M_ext 8 MHz, internal mode pic_loop_8M_int pic_intpt_8M_int 31 kHz, internal mode pic_loop_31k_int pic_intpt_31k_int The naming of the test programs of LPC1114FN28 of are listed in Table. 2. The naming rule is the same as Table. 1. Table. 2 All clock rates for LPC1114FN28 test programs Test Program Name loop type interruption type Clock Rate 12 MHz, internal mode arm_loop_12M_int arm_intpt_12M_int 12 MHz, external mode arm_loop_12M_ext arm_intpt_12M_ext 24 MHz, external mode arm_loop_24M_ext arm_intpt_24M_ext 36 MHz, external mode arm_loop_36M_ext arm_intpt_36M_ext Fig. 2 Interrupt type program implementation 48 MHz, external mode arm_loop_48M_ext arm_intpt_48M_ext Two designs under test (DUTs) or μCs are chosen: one is Microchip's PIC18F4550, and the other is NXP's ARM III. TEST BOARD AND MEASUREMENT SETUPS Cortex-M0 LPC1114FN28. PIC18F4550 is an old-style, small The measurement of the IC conducted EMI (cEMI) follows and popular μC used for many embedded systems through the international IC-EMI measurement standards (IEC 61967-4) industrial history. And LPC1114FN28 is an ARM core which [11]. This standard specifies the measurement for cEMI of IC is a new-style μC popular in nowadays applications. between 150 kHz to 1GHz. The block diagram of these methods is shown in Fig. 3. In this paper, 150Ω method (Fig. 3 There are two clock source modes on the two DUTs: the upper right) is used to observe the cEMI from VDD pin. internal oscillation mode or external oscillation mode. Both modes can be controlled by PLL. The fundamental clock source is determined by external crystal and is designed to be 48 MHz according the design application note. The internal oscillators are embedded in the DUTs. The PLL configuration of PIC18F4550 is controlled by its registers: OSCCON, CONFIG1L and CONFIG1H [9]. The clock rate can be set by equation (1) and (2): 8 MHz, for internal clock mode ( 1 ) (7 - OSCCON<6:4>) 96 MHz, for external clock mode ( 2 ) CONFIG1L<4:3> - 1 The PLL configuration of LPC1114FN28 is also controlled Fig. 3 IEC-61967-4 measurement by registers [10]. The name of the registers, the clock rate The test circuit PIC18F4550 follows cEMI measurement equation, and the PLL controls are omitted here. standard [11]. It uses an external 20MHz crystal oscillator to generate 48MHz clock by PLL for external oscillator mode. The PLL setups can be changed or pre-determined The circuit layout is shown in Fig. 4 and the measurement according to the characteristics of the two μCs. For setup is shown in Fig. 5. The spectrum analyzer is Agilent PIC18F4550, the internal clock rate can be on-line changed by N1996A-CSA. program through PLL setups while the external clock rate is fixed to be a pre-determined value (in this paper, 48 MHz is set). For LPC1114FN28, the internal clock rate is set to 12 MHz and the external clock rate can be on-line changed by program. The naming of the test programs of PIC18F4550 are listed in Table. 1. The “pic_loop_48M_ext” means the PIC test program is an infinite loop type under external clock mode with the clock rate set to 48 MHz. “pic_intpt_31k_int” means the program is an interrupt type under internal clock mode set to 31 kHz clock. Other programs are named by the same rule accordingly. Fig. 4 PIC18F4550 Test Board 119 Fig. 8 shows the comparisons among interrupt type programs with different PLL setups. In Fig. 8, cEMI is almost the same under interrupt type program implementation. Combined with Fig. 7, it seems the programs have no effect on any cEMI behaviors on the PIC μC. Fig. 5 Measurement Environment The test board of LPC1114FN28 also follows the same rule as suggested by [11]. The test board is shown in Fig. 6. The measurement environment is similar to Fig. 5 and is omitted. Fig. 8 Comparisons between PIC interruption type programs under different internal clock rate (only envelopes are showed) However, Fig. 9 shows the comparisons between the external/internal clock source programs of interrupt type. From Fig. 9, the frequency response of pic_intpt_48M_ext (external oscillator mode) is higher than internal oscillator. The major frequency differences lies in the 12 MHz and its harmonics. From Fig. 8, the three internal oscillator PLL setups are generally identical; this means, in PIC μC, the internal/external oscillator setups may be one of the critical program issues on the cEMI behaviors. Fig. 6 LPC1114FN28 Test Board IV. MEASUREMENT RESULT A. Measurement Result of PIC18F4550 Fig. 7 shows the comparisons among infinite loop type programs with different PLL setups. In Fig. 7, “pic_loop” programs are almost identical; this means different program operations have no influence on the cEMI behavior of the PIC DUT. Envelops of these cEMI responses are also shown. Only the envelopes of cEMI results of the following measurements will be shown in order to hide irrelevant details. Fig. 9 Comparison of pic_intpt_48M_ext and pic_intpt_8M_int In Fig. 10, comparisons among different clock sources and program types are shown here. Generally, the frequency response of loop type programs is higher than interrupt type programs.
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages5 Page
-
File Size-