The Designing, Realization and Testing of a Network Filter used to Reduce Electromagnetic Disturbances and to Improve the EMI for Static Switching Equipment Petre-Marian Nicolae Ileana-Diana Nicolae George Mihai UNIVERSITY OF CRAIOVA MAIN BUILDING ICMET CRAIOVA FACULTY OF ELECTROTECHNICS UNIVERSITY OF CRAIOVA - FACULTY OF ELECTROTECHNICS, ICMET CRAIOVA
I. Introduction Outline II. The Designing of a Network Filter III. The Realization of a Network Filter IV. Experimental Determinations of Filter Insertion Loss V. Electromagnetic Compatibility Problems VI. Conclusions
I. INTRODUCTION High switching speed of the IGBT => high levels of EMI. The switching action of IGBT in a power converter generates both CM and DM EMI. Practical cases require: - anti-disturbances measures concerning the static converters supplying circuit - special attention to the output circuits and cases used to shield the electronic equipment Main role of network filters -> the reduction of conducted emissions of common /differential mode.
II. THE DESIGNING OF A NETWORK FILTER Experimental determinations=> a convenient solution to meet EM compatibility standards = a filter covering [150 khz 30 MHz]. Fig. 1. Theoretic schematic filter The values for capacitances and inductances are determined with the algorithm described by (1) (10), for the frequency ranges (10 480) khz and (2 3) MHz for the static switching equipment.
III. THE REALIZATION OF A NETWORK FILTER The constructive principles are very important. Reccommendations: - shortest possible link conductors => highest possible own frequency for capacitors); - compulsory grounding; - use of magnetic and dielectric materials with losses. A. Capacitors (Capacitors + Internal resistance of the disturbances source) = voltage divider that reduces the disturbing voltage corresponding to the ratio of impedances. Passing coaxial capacitors -> used only for CM disturbances. have very large own frequencies (> 1 GHz) and are efficient only for shielded cases. Non-coaxial capacitors -> used for CM and DM (symmetrical) disturbances. Possible combinations (symmetrical + non-symmetrical) anti-parasitic measures: - XYY capacitors (supplementary armature between active armatures) ; - XY capacitors (a single divided armature). - Y capacitors -> connected between the supplying and protection conductors. They have high capacitances and good operation mechanical and electrical safeness. A common solution against large capacitances on the Y paths = serial connections of some coils with current compensation.
B. Coils Coils self inductance = f (connections lengths, assembling modality, internal construction). (Small own inductance + high high-frequency internal resistance of the disturbance source) => better efficiency The solution with current compensation = advantageous at coils used only against CM disturbances. For frequencies > 1 MHz, the filtering coils are often substituted by ferrite rings introduced on conductors or toreshaped magnetic cores on which the cables are wrapped. Fig. 2 -> used for each of the four phases to realize the network filter (380V, 4 x 63A).
Input coil (L1): - copper conductor, rectangular section (2.1 x 3.8 mm); -isolation = PE 2S; - 5 aerial loops, diameter = 45 mm. Capacitor C : - 2 passing 470 pf- capacitors (each supporting 40 A) parallel-connected => filter s rated current = 63A. Output coil (L2) : - copper conductor, rectangular section (2.1 x 3.8 mm); -isolation = PE 2S; - 2 coils with opposite winding senses, made on a round base, with ferromagnetic iron plate core (impregnated sheet plates of type I 20 with a section of 6 cm 2
(a) (b) Fig. 3 (a) One cell; (b) The assembly network filter (each coil has its own shield) Box-shaped shield, made of steel plate, sandblasted and galvanic covered with passivated Zinc with a width of 1 and 2 mm. Shield with 8 compartments, (2 coils/phase X 4 phases). The passing capacitors provide the link between the coils and compartments. To obtain a better electric contact for the shield walls, between and after the clamping of the separation walls one covered their contact-side and the contacts between shield and passing capacitors with a spray for metallization with copper particles.
IV. EXPERIMENTAL DETERMINATIONS OF FILTER INSERTION LOSS One considered the insertion loss, determined for identical values of the impedances corresponding to the source and receiver, e.g. for adapted systems with standard values of 50 Ω. The substitution method was used, all tests being performed in a system of 50 Ω, according to CISPR 17 (for each cell separately). Circuit components: 1 - Sinusoidal Voltage Generator SMY 02 2 - Coaxial switch 3 - Network filter 4 - EMI Receiver SMV 42 The isolation transformers (PEGZ 0.6/10) makes the difference to Fig. 4.
Details on the insertion loss along the phase R: Test standard = CISPR 17. Test conditions: T = 29 C, humidity h = 58% Scanning mode: Fig. 6. Insertion loss along the phase R Features of the insertion loss of the filter along the phase R: - It is higher than 19 dbµv along the range 100 khz 10 MHz - It is higher than 50 dbµv along the range 700 khz 2 MHz - It has a maximum attenuation of 62 dbµv, for 900 khz - It has a resonance point at the frequency of 4.5 MHz, for which the attenuation falls up to 22 dbµv.
Details on the insertion loss along the neutral wire: Test standard = CISPR 17. Test conditions: T = 29 C, humidity h = 58% Scanning mode: Fig. 7. Insertion loss along the neutral wire Features of the insertion loss of the filter along the neutral wire: - It is higher than 19 dbµv along the range 100 khz 10 MHz - It is higher than 50 dbµv along the range 350 khz 1.7 MHz - It has a maximum attenuation of 61 dbµv, for 900 khz. - It has 2 resonance points at 1.8 MHz, when the attenuation falls up to 38 dbµv and respectively at 3.7 MHz, when the attenuation falls up to 28 dbµv
V. ELECTROMAGNETIC COMPATIBILITY PROBLEMS Scope: EMI reduction for a static switching converter. Target: conducted and radiated emissions. A. Experimental Determinations Concerning the Conducted Disturbances 1) Experimental Determinations No Filter Case The disturbances were measured considering CISPR 11. The static converter was classified as class A equipment. o o Details on equipment used at experiments electromagnetic disturbances receiver of type ESCI 3 produced by Rohde & Schwarz artificial three-phase network of type LT 32/C, produced by AFJ Italy. Receiver s test parameters Test range: 150 khz 30 MHz (according to CISPR 22, Class B, Group 1 of equipment) Band width: 9 khz Frequency step: 4.5 khz Test time/ frequency step: 100 ms (for pre-scanning) and 2 s (for final scanning). Experiments steps Firstly - a pre-scan of the entire frequency domain. For the values closer to less than the 6 db to the imposed limits or overcoming them, the quasi-peak values and the average values of the disturbances were measured.
Fig. 8 Disturbance transmitted through conduction along the phase R, before the filter installation (curve no. 3 and no. 4) and the limits imposed by CISPR 22 (curve no. 1 and no. 2). - the curve no. 1 - limit imposed by CISPR 22 for the quasi-peak values of the disturbances transmitted through conduction, introduced by the class B equipment - the curve no. 2 - limit imposed by CISPR 22 for the average values of the disturbances transmitted through conduction, introduced by the class B equipment - the curve no. 3 - quasi-peak values of the disturbances introduced along the phase R of the network by the equipment - the curve no. 4 represents the average values of the disturbances introduced along the phase R of the network by the equipment.
The values of the disturbances introduced by the tested equipment before the filter s installation overcome the limits: - For the quasi-peak value around the frequency of 200 khz and 260 khz - For the average value, around the frequency of 260 khz Fig. 8 Disturbance transmitted through conduction along the phase R, before the filter installation (curve no. 3 and no. 4) and the limits imposed by CISPR 22 (curve no. 1 and no. 2). Both sets of measured disturbances are significantly close to the limits (to less than 10 db) for other values of the frequency: - For the quasi-peak value, around the frequencies of 150 khz, 280 300 khz, 400 khz and 6.5 MHz; - For the average value, around the frequencies of 160 khz, 200 khz, 300 khz, 400 khz and 6.5 MHz.
2) Experimental Determinations Filter present Case The receiver s test parameters: -Test domain: 30 MHz -1 GHz (according to CISPR 11). - Frequency bandwidth: 120 khz. - Frequency step: 60 khz. -Test time / frequency step: 100 ms for pre-scanning and 2 s for final scanning. The tests were made in two steps: - A pre-scanning was performed for the entire frequency domain with the above parameters. - For the values approaching with less than 6 db the imposed limits (or exceeding them) the disturbances mean and quasi-peak values were determined, with a measuring time of 2 s for each measured frequency (if a single value is found to be outside the limits, the equipment is considered to be not compliant with the standards).
Visible effect of the network filter over the conducted electromagnetic disturbances produced. They are significantly reduced and so CISPR 11 regulations are obeyed. For the radiated disturbances measuring one used a receiver for electromagnetic disturbances of type ESCI 3 produced by Rohde & Schwarz and a hybrid network antenna (logarithmic periodic with wide range) of type BTA-M, made by Frankonia. Fig. 9. Disturbance transmitted through conduction along the phase R, after the network filter installation (curve no. 3 and no. 4) and the limits imposed by CISPR 22 (curve no. 1 and no. 2).
B. Experimental Determinations Concerning the Radiated Disturbances 1) Experimental Determinations Concerning the Radiated Electromagnetic Disturbances No Filter Case Fig. 10. Radiated disturbances in horizontal polarization, No filter case Conventions used in Fig. 10 and 11: - line (1) C22B-R3 = limit imposed by CISPR 11 for the quasi-peak values of the radiated disturbances introduced by the class B equipment; - line (2) = peak values of the disturbances radiated by equipment; - symbols + = quasi-peak values of the disturbances radiated by the equipment The values of the disturbances radiated by the equipment in the No filter case disobey the limits imposed by CISPR 11 around the frequencies: 34 MHz, 35 MHz, 41 MHz, 61 MHz, 73 MHz, and in the range (120-150 MHz)
2) Experimental Determinations Concerning the Radiated Electromagnetic Disturbances Filter Present Case Fig. 11. Radiated disturbances in horizontal polarization, Filter present case The experimental determinations proved that the standard CISPR 11 cannot be obeyed without supplementary protection measures for the limitation of the radiated electromagnetic disturbances. Therefore the following actions were performed: - the network filter was introduced inside a well shielded case, in order to obtain an insertion loss for the high frequency electromagnetic radiations; - static converter was properly shielded; - a proper assembling was provided for the coupling between the network filter and the static converter, such as to get practically a proper isolation between the electromagnetic environment inside the static converter s case and the external electromagnetic environment; - ferrite rings were added on the static converter s output cables (other filter systems might be used as well). Fig. 11 => the radiated disturbances were reduced up to a level accepted by CISPR 11
VI. CONCLUSION The development of power electronic equipment generates problems relative to the compliance with electromagnetic compatibility standards, supplementary measures being often required. Some of them refer to supplementary filtering measures meant to reduce significantly the conducted emissions transmitted through the supplying cables of the respective equipment. The network filters designing is not a difficult task, unlike their realization and assembling along with the equipment to be protected. Experimental determinations revealed that by their own the filters cannot always provide the electronic equipment compliance with CISPR 11 concerning the radiated emissions. Therefore supplementary measures must be taken, as this paper emphasizes.
Higher static converters powers involve more problems related to the realization of coils with large inductances and very small parasitic capacitances. Owing to the implementation-related difficulties, for the designed filter, a simplified filter schematic proved to be useful. The experiments demonstrated the efficiency of the solution consisting in the realized filter and supplementary measures relative to the significant reduction of the disturbances caused by the static switching converter for which the filter was made.