Chamber Influence Estimation for Radiated Emission Testing in the Frequency Range of 1 GHz to 18 GHz

Transcription

1 Chambe Influence Estimation fo Radiated Emission Testing in the Fequency Range of 1 GHz to 18 GHz Alexande Kiz Electomagnetic Compatibility and RF-Engineeing ARC eibesdof eseach GmbH eibesdof, Austia alexande.kiz@acs.ac.at Wolfgang Müllne Electomagnetic Compatibility and RF-Engineeing ARC eibesdof eseach GmbH eibesdof, Austia wolfgang.muellne@acs.ac.at Abstact The pape pesents a compehensive analysis of the behavio of anechoic chambes in the fequency ange fom 1 GHz to 18 GHz. Numeical simulation and statistical analysis have been used to estimate the influence of the test site to adiated emission esults above 1 GHz. Fo typical EMC chambes the influence facto is aound.7 db. This facto can be impoved by using bette absobing mateial o inceasing of chambe size. Fo a wide ange of EUT s the compliance uncetainty can be estimated using the chambe influence facto. The pediction of the chambe influence fo small omnidiectional EUT s fits vey well. mall EUT s with a diective distubance souce ae oveestimated. Lage EUT s with multiple souces have not been investigated. Anechoic chambe; adiated emission testing; chambe validation; EMC above 1 GHz; compliance uncetainty I. INTROUCTION The development of a standad to measue adiated emissions above 1 GHz is an impotant topic of the EMC community in the last yeas. CIPR is woking towad this since with high pessue. The definition of a new measuement technique is quite a complex task. In case of a adiated emission test pocedue the poject can be split into seveal independent poblems. Fist of them is the development of the measuement method itself. Fo this method the calculation of the measuement uncetainty is equied. Measuement uncetainty is also called compliance uncetainty, if the measuement is aimed to test the compliance of a EUT. To estimate the compliance uncetainty all factos which influence the esult of the measuement have to be detemined. The most impotant factos ae the influence of the test site, the calibation uncetainty of the eceive antenna, the uncetainty of the cable loss and the uncetainty of the test eceive. The pocedue how to combine these factos to the compliance uncetainty has to be done accoding to the GUM [1] [] o to CIPR [3]. This pape poposes a method how to estimate the influence of the test site on the adiated emission test esult. II. ITE IMPERFECTION BELOW 1 GHZ A. Basics of Validation The basic idea of the validation technique is to substitute the EUT by a adiation souce. The adiation patten of each EUT is diffeent and unknown, so the popeties of a geneal EUT have to be assumed. These ae omnidiectional adiation chaacteistic and no nea field coupling with the test site. This geneal EUT is placed at the test site and the electical field stength is measued. The esult is compaed to a pe-calibated field stength and a maximum diffeence is allowed. One possibility to ealize such a EUT is to use boadband antennas like Bicones o Log Peiodic Antennas, which ae diven by a signal geneato o a comb geneato. This appoach is ealized by all test site validation standads like CIPR [4], ANI [5], ETI [6] and VCCI [7]. Of couse thee ae some difficulties to keep in mind. The fist one is the nea field coupling of biconical antennas with the gound plane. This is solved by intoducing a dual antenna facto [8]. Anothe poblem is that Log Peiodic Antennas ae not omnidiectional. In new standads like CENELEC [9] the poblem is avoided by using small biconical antennas up to 1 GHz. B. Nomalized ite Attenuation In a pape fom A. A. mith [1] the use of a Nomalized ite Attenuation (NA) is suggested. This technique fits vey well to the ideas pesented above. The EUT is eplaced by a calibated antenna. The taceability of the NA method is ensued by calibation of both antennas. The limit of 4 db NA deviation used by ANI and CIPR is calculated by adding the uncetainty contibution of 1 db fo each antenna, 1 db fo the eceive and 1 db fo the test site. A common mistake in the undestanding of NA testing is to think that a test site has a deviation of 4 db. The site deviation is only 1 db and the emaining 3 db is the validation uncetainty. Of couse this way of dealing with uncetainty is not state-of-the-at, but it is still applicable.

2 III. ITE IMPERFECTION ABOVE 1 GHZ ue to physics thee is a diffeent situation above 1 GHz. ome of the befoe mentioned effects also occu hee, some can be neglected, but thee ae also new effects. The effect of the absobes electomagnetic wave is attenuated by the eflectivity is nealy identical. The dependency of the eflectivity fom the impinging angle becomes moe impotant. How stong the impinging wave is scatteed by the absobe depends on the absobe quality and shape. The influence of nea field coupling between the EUT and the envionment can be neglected. The wavelength is small compaed to the distance to the absobe, so the eactive neafield egion is not distubed by the suoundings. A new effect is the stong diectivity of the eceive antenna. In this fequency ange widely common antennas ae double idged hon antennas and log peiodic antennas. Thei gain is appoximately 1 dbi fo hon antennas and 7 dbi fo log peiodic antennas. This esults in a lowe sensitivity to the suoundings, because the antenna faces the EUT and eflected waves ae not impinging the antenna in the main lobe. IV. NUMERICAL IMULATION In this pape the influence of the anechoic chambe is calculated by numeical simulation. A. Model of Anechoic Chambes Befoe building up a model some assumptions have to be made, which ae valid fo the desied fequency ange: The adiation chaacteistic of the EUT is stongly depending on the type of the equipment. Thee ae test objects with stong main lobes possible as well as omnidiectional adiatos. We assumed an omnidiectional chaacteistic, because this case will lead to the stongest influence of the anechoic chambe. The test site can be any eflection fee envionment. We choose a cubic anechoic chambe because most of the sites will be of that type. The eceive antenna is a wide band hon antenna (double idged hon antenna) with a naow beam width and a good font to back atio. The pefomance of the absobing mateial is descibed by the eflectivity. Only eflected waves ae taken unde consideation, scatteed waves ae neglected. Only single eflections ae taken into consideation, see Fig. 1. The enegy of multiple eflections is to low to change the eceived field stength significantly when typical absobes ae used. Physical effects ae not consideed like the nea-field coupling (eactive nea-field egion). Anothe neglected effect is the dependency of the eflectivity to the impinging angle. Y Z Ray 1 Ray 1 Ray Ray 3 Ray 4 Ray 5 Ray 6 Ray 6 Figue 1. Consideed eflected ays and thei designation top view font view Fo each ay a complex wave equation fo the eceived field stength is assumed [11]. ee (1) fo the Poynting vecto of the diect ay E, whee is the distance between tansmitte and eceive and λ is the wavelength. j = e π λ e The impinging angle of the Poynting vecto is detemined by a unit vecto. (1) e = e () Thee ae six eflected ays assigned as Ray1 to Ray6. The fee space attenuation fo the diffeent path length is taken unde consideation by multiplying the atio of the diect length and the length of the cuent ay W. The patten facto PF consides the adiation chaacteistics of the eceive antenna. This facto attenuates the ay additional to the absobe eflectivity R. Ray R+ PF πw j λ = 1 e eray (3) W The impinging angles of the Poynting vectos ae detemined by the unit vectos, e.g.

3 e e Ray1 Ray6 = e = e The chambe influence I in db is calculated accoding to (5). The Poynting vectos of all six eflected ays and the diect ay ae added and elated to the Poynting vecto of the diect ay. I = + (4) Ray = 1 log (5) The diectivity of a typical eceive antenna is shown in Fig.. This antenna is double idged hon antenna fo the fequency ange fom 1 GHz to 18 GHz. db -5 db -1 db -15 db - db 6 (length) by 4 m (width) by 3.5 m (height). The test distance is 3 m and the absobe eflectivity db. We scanned though a volume to estimate the chambe influence I, whee the distance between tansmit and eceive antenna is kept constant, see Fig. 3. The size of the cube is 1 m by 1 m by 1 m and the bottom of the cube has a distance of 1 m to the floo of the anechoic chambe. Regading to the and Y axis the cube is located in the middle of the chambe. The obseved fequency is 1 GHz, so the size of the plane is lage than the wavelength. 4 m 1.5 m 3.5 m 1 m Tansmit 1.5 m 1.5 m 7 m Tansmit Figue. Radiation patten of a typical eceive antenna Fom this figue the patten factos PF fo each ay can be ead. The diect ay and Ray1 ae impinging at angle, so the patten factos eads db. The eflected ay fom the wall behind the eceive antenna Ray6 eads - db. The ays eflected fom gound, ceiling and the side walls ae impinging at an angle of appox. ± 45. The patten facto in this case is aound - 9 db. ue to this behavio it becomes obvious that high gain antennas ae less sensitive to the test site than low gain antennas. TABLE I. PATTERN FACTOR PF Angle Patten Facto Ray No. [ ] PF [db] 1 ± 45-9, 3, 4, B. tatistical Evaluation of Chambe Influence All assumption made till now concen physics only and no specific popeties of an anechoic chambe. Now we assume a typical compact fully anechoic chambe, with a size of 7 m 7 m Figue 3. canning though a test volume top view, font view The chambe influence I can be visualized in a 3 diagam, see Fig. 4, whee we cut though the Y-plane at a height of 1. m. The esult is a wave patten and has about six sine peiods in the axis. This is appoximately twice the wavelength, due to a wavelength of 3 cm and a plane length of 1 m. The eason fo this is that the path diffeence is twice as high as the movement along the axis. The eflected ay has to tavel to the wall and back. We can obseve that the influence in the volume can be any value between - 1. db and +.95 db, dependent fom the chosen point. Theefoe we need to use a statistical analysis to calculate a pobable influence facto. The inteval of I is not symmetical due to the non-lineaity of the logaithm. One can calculate the pobability distibution - also called istibution Function P (I), see Fig. 5. The diagam is nomalized that the aea below the cuve coesponds to 1 %. The distibution is vey simila to the Rice distibution. This distibution occus also at mobile communication channels. In this application thee is a simila situation multiple signals aiving fom diffeent diections and with diffeent signal levels and one dominant signal (line-of-sight).

4 .8 Chambe Influence I [db] istance to wall Y [m] istance to wall [m].5 Figue 4. Chambe Influence I in Y-plane at a height of 1. m Pobability P [%] Chambe Influence I [db] Figue 5. istibution Function P(I) Pobability PF [%] db Chambe Influence I [db] Figue 6. Pobability ensity Function PF(I) With the istibution Function we know the pobability that a cetain test site influence occus. Moe elevant is the pobability that the chambe pefomance is bette than a cetain value, e.g. 1 db. Theefoe the Pobability ensity Function (PF) is calculated by using the istibution Function P: I PF( I ) = P( x) + P( x) dx (4) Using the PF (I), Fig 6 can be ead: At 95 % of all points in the test plane the test site influence is less than ±.68 db. Common standads fo calculation of the measuement uncetainty [1] [] ae using standad deviations with coveage factos of 95 %. o we define the pobable chambe influence PCI at 95 %. ( 95% ) PCI = PF (5)

5 C. Examples of chambe influence investigation To investigate the influence of the absobe eflectivity on the pobable chambe influence, we kept the chambe size (7 m by 4 m by 3.5 m) constant and vaied the eflectivity R fom 5 db to 4 db. This esult of the simulation is compaed to a vey simple estimation by Hollis [1]. R PCI = log (6) This calculation assumes a diect ay and one eflected ay in phase. The eflected ay is only attenuated by the eflectivity and not by the lage path length. Tansmit Pobable Chambe Influence PCI [db] db 18.3 db Absobe Reflectivity R [db] tatistical Evaluation imple Estimation Figue 7. Influence of eflectivity to pobable chambe influence The esults fo the statistical evaluation and the simple estimation ae shown in Fig. 7. Remakable is the good coespondence, although the calculations ae based on completely diffeent assumptions. The equied eflectivity to get a pobable chambe influence of 1 db is 18.3 db fo the simple estimation. The complex simulation leads to value of 16.8 db, so the diffeence is only 1.5 db in the absobe eflectivity. Anothe investigation is to calculate the influence of the chambe size. To do so we kept the absobe eflectivity constant (R = db) and changed the size of the chambe by a ize Facto. The dimensions of the chambe (length, width and height) ae multiplied by the ize Facto, see (5) and Fig. 8. The ize Facto is vaied fom 1 to 11. A ize facto of 11 leads to a huge chambe of 77 m by 44 m and 38.5 m. Of couse such chambes have no pactical elevance, but they can be compaed to Open Aea Test ites, whee only the gound setup influences the esult. Length' = Length ize Facto Width' = Width ize Facto Height' = Height ize Facto (5) Tansmit Figue 8. Changing the chambe size top view, font view Even at vey lage ize Factos the PCI is not deceased to zeo, see Fig. 9. The eason fo this is the absobe layout at the gound, which influence is not deceased by changing the ize Facto. Pobable Chambe Influence PCI [db] ize Facto [1] Figue 9. Pobability density Using Fig. 7 and Fig. 9 we can estimate how effective an impovement of chambe pefomance could be. If want to halve the pobable chambe influence fom.7 db to.35 db, we can incease the absobe eflectivity by 6 db to 6 db. The othe possibility to each this goal is to build a chambe which is 3.5 times lage. V. MEAUREMENT OF THE TET ITE INFLUENCE The measuement of the test site influence is called chambe validation. This chambe validation is pefomed simila to the validation technique below 1 GHz.

6 An omnidiectional antenna is used as tansmit antenna. This antenna is placed to defined locations in the test volume. The site attenuation is measued to a eceive antenna, whee the distance between the antennas ae kept constant. The eceive antenna should be the same antenna type like it is used fo emission measuements and it is facing the tansmit antenna. A calibation of both antennas is not necessay, because one has to look only to the spead of the site attenuation. This spead is called site VWR. The detemination of the NA is not equied. It is vey had to build omnidiectional antennas up to 18 GHz. Commecial available ae at the moment conical monopoles up to appox. 6 GHz and biconical antennas up to 3 GHz. ue to the lack of suitable antennas to 18 GHz, it is also possible to use a ecipocal technique. The omnidiectional tansmit antenna is eplaced by an isotopic field pobe and the electic field is tansmitted with the eceive antenna. The disadvantage of this method is the need fo powe amplifies up to 18 GHz. VI. ETIMATION OF UNCERTAINTY CONTRIBUTION The PCI simulated in the pevious chaptes descibes the influence of the anechoic chambe to the adiated emission test. The assumptions ae an omnidiectional small EUT and a typical diective eceive antenna. These assumptions lead to the stongest possible influence of the chambe to the adiated emission test esult. A EUT with diective chaacteistic illuminates the chambe less and theefoe the adiated emission esult depends less on the chambe pefomance. Theefoe the PCI can be used as uncetainty contibution in the calculation of the measuement uncetainty of omnidiectional equipment. Using the PCI fo small diective EUT s is possible as no undeestimation of the influence occus. Figue 1. Emission measuement fo small and lage EUT s Lage EUT s may contain multiple adiation souce and due to its size a height scan of the eceive antenna is equied, see Fig. 1. Futhe investigation needs to be pefomed to ensue the validity of the PCI. VII. CONCLUION This pape pesents a compehensive analysis of the behavio of anechoic chambes in the fequency ange fom 1 GHz to 18 GHz. A simulation model is built to calculate the influence facto of a chambe. An omnidiectional souce adiates a diect and six eflected ays. Absobe and eceive antenna patten data ae consideed in the calculation of the influence. By scanning though volumes and statistical analysis a pobable influence facto PCI is estimated. We investigated the dependency fom the absobe eflectivity and chambe size. The suggested PCI can be used fo the adiated emission uncetainty fo a wide ange of EUT s. REFERENCE [1] BIPM, IEC, IFCC, IO, IUPAC, IUPAP, OIML: Guide to the Expession of Uncetainty in Measuement (GUM), 1993 [] EA-/4: "Expession of the Uncetainty of Measuement in Calibation", EA Euopean co-opeation fo Acceditation, ecembe 1999 [3] CIPR Ed. 1.: pecification fo adio distubance and immunity measuing appaatus and methods - Pat 4-: Uncetainties, statistics and limit modelling - Uncetainty in EMC measuements, Novembe 3 [4] CIPR Ed. 1.: pecification fo adio distubance and immunity measuing appaatus and methods - Pat 1-4: Radio distubance and immunity measuing appaatus - Ancillay equipment - Radiated distubances, Novembe 3 [5] ANI C63.4-1: Ameican National tandad fo Methods of Measuement of Radio Noise Emissions fom Low-Voltage Electical and Electonic Equipment in the Range of 9 khz to 4 GHz, New Yok: IEEE, 1 [6] ETI ETR 73-: Electomagnetic compatibility and Radio spectum Mattes (ERM); Impovement of adiated methods of measuement (using test sites) and evaluation of coesponding measuement uncetainties; Pat: Anechoic chambes, Euopean Telecommunications tandads Institute, Febuay 1998 [7] VCCI: Regulations fo voluntay contol measues. 14th Edition, Apil [8] W. Müllne, H. Gan: Fom NA to site-efeence method fo EMC test site validation, Int. ym. on EMC, Monteal 1, page vol. [9] Technical Repot EN5147-3:1: Electomagnetic Compatibility Basic Emission tandad Pat 3: Emission measuements in Fully Anechoic Rooms, CENELEC, 1 [1] A. A. mith, R. F. Geman, J. B. Pate: Calculation of ite Attenuation fom Antenna Factos, IEEE Tansaction on Electomagnetic Compatibility, Vol. EMC-4, No. 3, August 198 [11] W. L. tutzman, G. A. Thiele: Antenna Theoy and esign, econd Edition, John Wiley & ons, Inc., 1998 [1] J.. Hollis, T. J. Lyon, L. Clyton: Micowave Antenna Measuements, cientific-atlanta Inc., 197