REO (UK) LTD, Uni ts 2 - 4 Cal l ow Hi l l Road, Craven Arms
Busi ness Park, Craven Arms, Shropshi re SY7 8NT UK
Tel : 01 588 67341 1 Fax: 01 588 67271 8 Emai l : sal es@reo. co. uk Websi te: www. reo. co. uk REO UK LTD Good EMC Engi neeri ng Practi ces i n the Desi gn and Constructi on of I ndustri aI Cabi nets 1 2 Contents 1 Introducti on and background 1. 1 The financial need for these good EMC engineering practices Because of the compl exi ty of modern i ndustri al i nstrumentati on and control products and systems, i t i s necessary, for commerci al and fi nanci al reasons, to control el ectromagneti c i nterference (EM ). Added to thi s i s the regul atory requi rement for the suppl i ers of products and systems, and owners of premi ses and si tes i n the European Uni on, to compl y wi th the el ectromagneti c compati bi l i ty (EMC) Di recti ve [1 ], especi al l y the very speci fi c requi rements i n the new EMC Di recti ve [1 ] for the use of good EMC engi neeri ng practi ces [2]. Thi s gui de addresses the practical i ssues of desi gni ng and assembl i ng i ndustri al cabi nets to better control EM , and to hel p achi eve EMC Di recti ve compl i ance. For i nformati on on other EM and EMC i ssues, such as management, testi ng, l egal , and theoreti cal background see [2], [3], [4] and [5] . Some of the techni ques descri bed here mi ght contradi ct establ i shed or tradi ti onal practi ces, but they are al l wel l -proven and i nternati onal l y standardi sed good modern EMC engi neeri ng practi ces at the ti me of wri ti ng. EMC i s a rapi dl y devel opi ng fi el d, because of the rapi d pace of progress i n el ectroni cs, computi ng, software, power control (e. g. vari abl e speed AC motor dri ves), radi ocommuni cati ons and wi red/wi rel ess data communi cati ons. The rapi dl y i ncreasi ng use of these technol ogi es i n i ndustry means that some EMC techni ques that mi ght have been perfectl y adequate i n the 1 960s (such as si ngl e-poi nt earthi ng, and bondi ng cabl e shi el ds at onl y one end) are now very bad EMC practi ce i ndeed. Al l professi onal engi neers have a duty (professi onal , ethi cal , and l egal ) to appl y the l atest and best knowl edge and practi ces i n thei r work. Remember that safety is always paramount, and should not be compromised by any EMC techniques. A typi cal exampl e of such a compromi se i s fi tti ng EM fi l ters that cause hi gh earth- l eakage currents that i ncrease safety ri sks. However, i t i s very i mportant to understand that where errors or mal functi ons i n el ectroni c ci rcui ts coul d possi bl y have i mpl i cati ons for functional safety (i ncl udi ng duri ng faul ts, foreseeabl e mi suse, overl oad, or envi ronmental extremes) merel y meeti ng the EMC Di recti ve and i ts harmoni sed EMC standards wi l l al most certai nl y not be suffi ci ent to achi eve adequate safety ri sk l evel s. Such an approach woul d al so al most certai nl y not achi eve compl i ance wi th the basi c standard on functi onal safety [6] or the i ndustry standards deri ved from i t (such as [7] and [8]), or wi th safety regul ati ons such as the Low Vol tage Di recti ve [9] and Machi nery Di recti ve [1 0]. Al though the desi gn and assembl y techni ques descri bed here are often used to hel p achi eve ' EMC for Functi onal Safety', a l ot more i s i nvol ved that i s not covered by thi s gui de for more i nformati on on thi s, vi si t [1 1 ]. 1. 2 These techniques suit a wide range of applications Thi s gui de i s concerned wi th i ndustri al i nstrumentati on and control , but the l aws of physi cs (hence EMC) appl y to al l el ectri cal and el ectroni c assembl i es and systems, regardl ess of thei r appl i cati on, i n exactly the same way as they do to i ndustri al cabi nets. An Amp i s sti l l an
Introducti on and background Creati ng an RF Reference 2 9 Wi ri ng and cabI i ng techni ques 1 6 Bondi ng ci rcui ts and uni ts to the RF Reference 45 Usi ng shi eI ded cabi nets Preventi ng gaI vani c corrosi on 55 References and further readi ng 68 70 SoI uti ons from REO 72 The Author 73 Product ExampI es 74 3 4 Amp, a mi crovol t sti l l a mi crovol t, and a MHz i s sti l l a MHz regardl ess of functi on or appl i cati on. hope that the way have wri tten and i l l ustrated thi s gui de makes the techni ques i t descri bes easy to appl y wherever el ectri cal and el ectroni c assembl i es are bei ng desi gned and constructed. 1. 3 EMphenomena and test standards REO (UK) Ltd have publ i shed a seri es of 1 7 EMC Gui des [4], whi ch descri be how the vari ous el ectromagneti c (EM) phenomena ari se, and how they can cause probl ems for el ectri cal and el ectroni c devi ces and ci rcui ts and the appl i cati ons they are used i n. They then go on to descri be the European EMC test standards, whi ch are based on nternati onal standards devel oped by the EC, and how to test usi ng them.
Many compani es do thei r own EMC testi ng accordi ng to European or nternati onal standards. There are many easi er, qui cker and l ess costl y ways to do EMC testi ng, but they are l ess accurate and not as useful for provi ng l egal compl i ance. However, they sti l l have val ue i n assessi ng the sui tabi l i ty of suppl i er's products, desi gn and devel opment, faul t-fi ndi ng and probl em-sol vi ng, checki ng workmanshi p standards and other QA acti vi ti es. For more on l ow-cost testi ng, see [1 2] . On-si te test methods exi st for testi ng equi pment outsi de of an EMC test l aboratory, wi th i ts careful l y-control l ed EM envi ronments, and can be used to save ti me/cost or where equi pment i s too l arge to be tested i n the normal way. Exampl es of on-si te methods that can be good enough to support a cl ai m of EMC compl i ance are gi ven i n [1 3] . 1. 4 Some basic EMC theory (with almost no maths) Thi s gui de focuses on practi cal ti ps and techni ques, and does not try to expl ai n why they work. Thi s approach can l eave engi neers vul nerabl e to speci al si tuati ons where an unusual approach may be needed, but tryi ng to convey the theoreti cal understandi ng requi red to devi se speci al techni ques i s outsi de the scope of thi s gui de, and many practi ci ng engi neers woul d fi nd i t very tedi ous anyway. So suggest readi ng the references at the end of thi s arti cl e, and then readi ng their references i f you sti l l need more background. But here are a few of the reasons why these EMC techni ques are needed: All modern el ectroni cs especi al l y di gi tal , swi tch-mode, and wi rel ess empl oy a wi de band of frequenci es from audi o up to at l east 1 00 MHz, maybe even up to several GHz (thousands of MHz). For them to operate correctl y and to achi eve EMC i t may be necessary to control some or al l of thei r frequency range by usi ng EMC techni ques i n thei r cabl i ng and assembl i es, and i n the cabi nets that house them. All conductors have si gni fi cant i mpedance at frequenci es above a few MHz, caused by ski n effect (whi ch i ncreases thei r resi stance) and i nductance. nductance (L) i s typi cal l y 1 H/metre for an ordi nary wi re (e. g. a green/yel l ow i nsul ated wi re), gi vi ng a reactance of 2FIL ohms at frequency I (e. g. 63 9/m at 1 0MHz). As a resul t, wi res (even ones wi th green/yel l ow col oured i nsul ati on) cannot be used to provi de an effecti ve ci rcui t reference vol tage at frequenci es above a few MHz (usual l y much l ess), and so can't provi de any EM control . All conductors such as metal work, wi res and cabl es make good acci dental antennas', and so l eak a proporti on of the power and/or si gnal s they carry i nto thei r external envi ronment. Thi s i s especi al l y the case where the conductors are l onger than one-tenth of a wavel ength (/1 0) at the hi ghest frequency of concern. The wavel ength 300/Bwhen Bi s gi ven i n MHz. Thi s i s a common cause of EM emi ssi ons probl ems. Shi el di ng can be used to reduce thi s effect, but i t i s never 1 00% effecti ve and i f done i ncorrectl y (e. g. shi el d bonded at onl y one end) can make the probl em worse. All conductors such as metal work, wi res and cabl es make good 'acci dental antennas', and so pi ck-up a proporti on of the EM energy i n thei r external envi ronment and so add vol tage and current noi se i nto the si gnal s and power they are carryi ng. Thi s i s especi al l y the case where the conductors are l onger than /1 0 at the hi ghest frequency of concern. Thi s i s a common cause of EM suscepti bi l i ty (i mmuni ty) probl ems. Shi el di ng can be used to reduce thi s effect, but i t i s never 1 00% effecti ve and i f done i ncorrectl y (e. g. shi el d bonded at onl y one end) can make the probl em worse. All conducti ve structures typi cal l y cal l ed 'earths' or ' grounds' become i neffecti ve above some frequency rel ated to thei r di mensi ons and method of constructi on. Above thi s frequency they no l onger provi de a stabl e or effecti ve ci rcui t reference vol tage i n fact, they become acci dental antennas i nstead of 'grounds' . At such frequenci es they cannot provi de EM control and may even add to EM probl ems.
The probl ems caused by ' acci dental
antennas' are i l l ustrated i n Fi gures 1 3, whi ch show how the typi cal wi re and cabl e l engths i nsi de cabi nets (0. 5 to 3 metres) can cause the el ectri cal energi es they carry (whether as si gnal s or power) to i nterfere wi th the radi o spectrum that i s vi tal for broadcasti ng and communi cati ons. The words 'earth' and ' ground' are very much mi sused i n el ectri cal and el ectroni c engi neeri ng, l eadi ng di rectl y to a great deal of confusi on, del ay and unnecessary extra costs. strongl y recommend that these words are never used, except when referri ng to an actual earth or ground el ectrode that i s buri ed i n the soi l under or around a si te. Thi s gui de wi l l try to take i ts own advi ce and use more accurate and expl i ci t terms such as: chassi s or frame; shi el ded (screened) encl osure; protecti ve bondi ng conductor (the green/yel l ow wi re i n mai ns cabl es, used for safety purposes); protecti ve bondi ng network (for a cabi net); common bondi ng network (CBN, for a si te); and of course, Reference. As has been i mpl i ed above, correct ci rcui t operati on and good control of EM and the achi evement of EMC requi res that we understand how to desi gn and create a Reference that i s effecti ve over the ful l range of frequenci es we need to control , especi al l y radi o frequenci es (RF) frequenci es above 1 50kHz. n some other publ i cati ons the Reference i s someti mes cal l ed the RF Reference, Reference Pl ane, RF Common, or other terms such as 'EMC Earth' or 'EMC Ground'. The RF Reference i tsel f must have very l ow i mpedance over the frequency range to be control l ed, much l ower than the i mpedance of the capaci tors i n any EM fi l ters. The onl y ki nd of structure that can achi eve a l ow enough i mpedance i s a metal mesh, i deal l y a metal sheet, whi ch i s why RF References are often cal l ed Reference Pl anes. 5 6 A ci rcui t' s RF Reference must al ways be physi cal l y cl ose to the ci rcui t that rel i es upon i t for operati on or EMC much cl oser than /1 0 at the hi ghest frequency to be control l ed (i deal l y /1 00 or even l ess, e. g. < 30mm for frequenci es up to 1 00MHz). Thi s i s because of al l of the conductors, i ncl udi ng l arge pi eces of metal wi th negl i gi bl e resi stance, that mi ght be used to connect the ci rcui t to the RF Reference suffer from i nductance and ' acci dental antenna' effects at l onger di stances. A metal box of whatever si ze can be used to shi el d a ci rcui t from i ts external EM envi ronment, but i t can onl y be used as the RF Reference for that ci rcui t i f the ci rcui t i s cl ose enough much cl oser than /1 0 to one of i ts metal surfaces. 1. 5 Don't rely solely on CE- marked electronics Don' t rel y sol el y on usi ng CE-marked components to assembl e a cabi net. The 'CE + CE = CE' approach whi ch assumes that as l ong as the components and products used i n the constructi on of a cabi net are al l CE marked, the cabi net as a whol e wi l l compl y wi th al l rel evant Di recti ves has no techni cal or l egal j usti fi cati on. Experi ence al l over the worl d shows that i t i s very rare i ndeed, for a cabi net constructed from CE-marked el ectroni c i tems suppl i ed by other manufacturers to actual l y meet the rel evant harmoni sed EMC standards when tested. [1 4] goes i nto thi s i ssue i n detai l , showi ng how to spot many of the tri cks that some manufacturers use when CE marki ng thei r products, and warni ng of the pi tfal l s that can compromi se the EMC of the cabi net, or the system or i nstal l ati on i t i s used i n, even when al l the components used i n the cabi net have excel l ent EMC compl i ance i ndi vi dual l y. Fig 1 Fig 2 Fig 3 7 8 1. 6 An overall EMC procedure A procedure that wi l l manage EMC to achi eve rel i abl e performance and l egal compl i ance for i ndustri al cabi nets and si mi l ar wi l l general l y requi re: Assessi ng the i ntended operati onal envi ronment for any EM di sturbances, whether conducted or radi ated, that mi ght threaten the operati on of the new cabi net. See [1 5] for more i nformati on on doi ng thi s. Assessi ng the i ntended appl i cati on for the sensi ti vi ty of other el ectroni c equi pment that mi ght be present, to the EM emi ssi ons from the new cabi net and i ts cabl es. Understandi ng al l of the EMC regul atory requi rements, for both emi ssi ons and i mmuni ty, see [2] [4] and [5] . Deri vi ng an EMC speci fi cati on for the cabi net and i ts cabl es (usual l y based around the basi c EC EMC test standards for emi ssi ons, such as C SPR 22, and i mmuni ty, such as the EC 61 000-4 seri es, see [4] ). Choosi ng thi rd-party el ectri cal and el ectroni c uni ts that are known to have the EMC performance requi red to meet speci fi cati ons deri ved from the above steps. Thi s means checki ng thei r test reports and QA systems, as di scussed i n [1 4] . Fol l owi ng the el ectroni c uni t suppl i ers' reasonabl e EMC i nstructi ons. Appl yi ng the EMC techni ques descri bed i n thi s gui de. Checki ng that the EMC techni ques have been correctl y appl i ed i n assembl y by i nspecti on and si mpl e tests (see the l ow- cost EMC test techni ques i n [1 2]).
Appl yi ng the appropri ate compl i ance procedures under the ol d or new EMC Di recti ves [1 ] [2] . EMC testi ng techni ques are descri bed i n [4] and [1 2] . Useful i nformati on on the above procedures can be found i n [1 6] , whi ch, despi te i ts ti tl e, i s of general rel evance to systems and i nstal l ati ons of al l types and the products used i n them. 1. 7 Following good EMC practices
n the ki nd of EM envi ronments covered by the generi c i ndustri al EMC test standards EN 61 000-6-2 and -4, most EMC probl ems can be sol ved by: Taki ng care to onl y uti l i se el ectri cal /el ectroni c that have proven good EMC performance [1 4] when tested to those standards or tougher ones Obtai ni ng and ful l y appl yi ng thei r suppl i er' s EMC i nstructi ons i n desi gn and constructi on Taki ng account of the bui l d-up of emi ssi ons caused by havi ng mul ti pl e uni ts [1 7]
Even so, i t i s sti l l advi sabl e to empl oy
good EMC practi ces wherever the uni ts' suppl i ers provi de no EMC i nstructi ons, or to hel p resol ve confl i cts between di fferent uni ts' EMC i nstructi ons. However, most normal EM envi ronments are worse than the ones descri bed by any of the EC or EN EMC test standards, because they speci fi cal l y do not cover the si tuati on where portabl e radi o transmi tters are used nearby whi ch i s now commonpl ace i n al l envi ronments (i ncl udi ng i ndustri al ), and cannot be control l ed wi thout very stri ngent securi ty measures. The standards al so i gnore a number of other EM envi ronmental si tuati ons that can easi l y occur. So i n al most al l real -l i fe i ndustri al si tuati ons, and especi al l y where the EM envi ronment i s more extreme than usual , the use of good EMC practi ces can be very i mportant i ndeed for preventi ng costl y l ost producti on due to i nterference probl ems. Good EMC practi ces i n the constructi on of el ectri cal and el ectroni c assembl i es have been known for decades, and are conti nual l y evol vi ng to cope wi th the i ncreasi ng frequenci es bei ng generated by modern el ectroni c technol ogi es, especi al l y di gi tal processi ng, swi tch-mode power conversi on, and wi rel ess voi ce and data communi cati ons. Rel evant standards and publ i c-domai n documents on good EMC practi ces i ncl ude [1 8] [1 9] [20] [21 ] and [22] , and there are a number of gui des to good practi ces produced by compani es that sel l i ndustri al components, such as [23] [24] and [25].
Good EMC practi ces are often di fferent from tradi ti onal el ectri cal assembl y and i nstal l ati on practi ces, and i n some l ong- establ i shed i ndustri es l arge amounts of money and ti me are sti l l needl essl y wasted by fi xi ng EMC probl ems wi th systems and i nstal l ati ons ari si ng duri ng operati on, i nstead of by desi gn, because of an apparent rel uctance to l earn about EMC or modern techni ques. t i s often the case that operati onal probl ems aren' t recogni sed as bei ng EMC-rel ated for some ti me, and even then take a l ong ti me to fi x. Part of good EMC practi ce i s to fol l ow the EMC i nstructi ons provi ded by the manufacturers of the el ectroni c uni ts that are to be used but onl y where these are reasonabl e and don' t confl i ct wi th what i s wri tten i n thi s gui de, or wi th each other. Where manufacturers' i nstructi ons di ffer or confl i ct, EMC experti se i s needed. For exampl e, some suppl i ers of i ndustri al components and modul es speci fy that shi el ded cabl es must have thei r shi el ds bonded to ' earth' at onl y one end, and they often provi de a screw- termi nal for that purpose. Whi l e thi s may sometimes be acceptabl e these days i n some speci al cases, i t wi l l general l y prevent typi cal i ndustri al cabi nets from passi ng thei r emi ssi ons and/or i mmuni ty tests and wi l l therefore general l y l ead to i naccurate or unrel i abl e operati on as wel l as non-compl i ance wi th l egal requi rements. Such poor EMC i nstructi ons are mostl y due to a l ack of knowl edge and/or poor desi gn of the el ectroni c ci rcui ts used for the i nputs and outputs. They are usual l y wri tten by compani es who have not tested emi ssi ons and i mmuni ty, or not tested them properl y, or tested them usi ng unreal i sti c set-ups. They sl avi shl y repeat the bad i nstructi ons i n thei r manual s, bel i evi ng them to be good EMC practi ce because they were tol d so 30 years or more ago. Good EMC practi ces shoul d general l y be fol l owed for al l i ndustri al cabi nets, to hel p the purchased el ectri cal and el ectroni c i tems achi eve the EM performance they are capabl e of, and to hel p EM mi ti gati on measures l i ke fi l teri ng, shi el di ng and transi ent overvol tage suppressi on, functi on correctl y. These techni ques requi re addi ti onal effort and ski l l i n desi gn, but general l y cost l i ttl e and add very l i ttl e ti me i n assembl y. 1. 8 Communicating good EMC techniques within a company Many compani es have probl ems i n turni ng the i ntenti ons of thei r desi gners i nto the constructi ons assembl ed by thei r assembl y staff. Nowhere i s thi s more 9 1 0 evi dent than i n EMC, where apparentl y smal l vari ati ons i n cabl e l ength or route, or component pl acement, can make huge di fferences. Whereas i n seri al manufacture there i s (hopeful l y) ti me al l owed for what i s constructed to be compared wi th what was desi gned and any di fferences i terated out i n custom engi neeri ng, desi gns need to be transl ated i nto products and systems wi thout errors at the first attempt i f a company i s to maxi mi se i ts profi ts and be successful . So i t i s i mportant to save ti me and money that compani es fi nd ways to communi cate the necessary EMC assembl y techni ques to thei r assembl y personnel . Thi s general l y means that the vari ous constructi on techni ques need to be documented as Work nstructi ons under a QA system, and then referenced by the desi gners on thei r drawi ngs wherever they need to be appl i ed. A number of i ndustri al cabi net manufacturers have used the graphi cs used i n thi s gui de as part of thei r Work nstructi ons, and the author wi l l be pl eased to provi de any such company wi th these graphi cs for such purposes, on request [26] . 2 Creati ng an RF Reference 2. 1 Introduction to RF References Al l i ndustri al cabi nets that contai n two or more i tems of i nterconnected el ectroni cs (e. g. a vari abl e-speed motor dri ve and i ts separate EM fi l ter, a PLC and a 24VDC power suppl y) shoul d use an RF Reference to hel p control EMC, whi ch can usual l y be created usi ng the exi sti ng cabi net metal work. Thi s secti on descri bes techni ques for creati ng RF References wi th useful EM performance from ordi nary (unshi el ded) cabi net metal work. A l ater secti on wi l l descri be the good EMC practi ces associ ated wi th shi el ded cabi nets. At frequenci es above a few MHz, onl y a hi ghl y conducti ve area or vol ume can achi eve a rel i abl e RF Reference. But an RF Reference i s onl y of any use i f i t i s a local one. ' Local ' i n thi s context means that the cabl es, devi ces and ci rcui ts shoul d remai n cl ose to the surface of the Reference at al l ti mes, wi th a spaci ng that i s l ess than /1 0 at the hi ghest frequency of concern (e. g. cl oser than 75mm at 400MHz, a typi cal wal ki e-tal ki e transmi tti ng frequency). Much cl oser spaci ng gi ves much i mproved EM performance. Where a metal chassi s (i deal l y free from j oi nts and perforati ons), metal cabi net, or metal encl osure i s used the wal l s, rear, top, bottom, or door coul d be used as l ocal RF References. ndustri al cabi nets often mount thei r el ectri cal , el ectroni c and other uni ts on a metal backpl ate, or i n a frame or cage for pl ug-i n modul es or pri nted-ci rcui t cards. The metal support structures nearest to the el ectroni cs, such as backpl ates or card cages, shoul d al ways form part of thei r l ocal RF References and have mul ti pl e metal -to-metal bonds to adj oi ni ng metal structures. 2. 2 Ineffectiveness of wires, straps and braids As was di scussed earl i er, green/yel l ow wi res or brai d straps to a si ngl e poi nt (someti mes cal l ed a 'star poi nt' ) are i neffecti ve at provi di ng an RF Reference at frequenci es above a few hundred kHz or so, dependi ng on thei r l ength. Even where el ectroni c uni ts (such as l ow- frequency anal ogue processi ng) do not empl oy or emi t RF frequenci es, the semi conductors i n i t wi l l happi l y demodul ate and i ntermodul ate any RF noi se i n thei r ci rcui ts, causi ng i mmuni ty probl ems, unl ess thei r manufacturers have taken great care i n thei r EMC desi gn. Even DC and l ow-frequency anal ogue el ectroni cs need to empl oy good EMC practi ces. 2. 3 Highly-conductive metal plating required, with no polymer passivation t i s easi est to use metal structures as l ocal RF References i f they are fi ni shed wi th a hi ghl y conducti ve metal pl ati ng that i s sui tabl e for the physi cal envi ronment and l i fecycl e of the cabi net (see secti on 6). Onl y metal surface-to-metal surface bonds have any chance of worki ng effecti vel y at the hi ghest frequenci es i n common use today. Non-conducti ve pai nt, pl asti c coati ngs, or anodi si ng, i ncrease the i nductance of bonds and reduce thei r effecti veness at hi gh frequenci es. Where pai nted or pl asti c-coated surfaces are to be bonded to, care must be taken to achi eve a good bond at RF (usual l y i nvol vi ng l ocal removal of pai nt or pl asti c), and then to prevent the corrosi on that can occur because the protecti ve coati ng has been removed. n the case of i ndustri al cabi nets fi tted wi th backpl ates, zi nc-pl ated backpl ates have become the standard for al most al l cabi net manufacturers i n recent years, to hel p wi th RF-bondi ng and EMC. Heavy zi nc or ti n pl ati ng i s the best conducti ve fi ni sh for mi l d steel , but someti mes thi nner pl ati ng i s used wi th an addi ti onal pol ymer passi vati on l ayer. Unfortunatel y, the passi vati on i s a thi n pl asti c coati ng and can rui n RF-bondi ng properti es. t i s often cl ai med that pol ymer passi vati on i s easi l y punctured by modest pressure from metal parts such as the nuts and bol ts typi cal l y used i n el ectroni c assembl y. However, i t mi ght prove necessary to use EMC gaskets, and si nce they woul d not appl y suffi ci ent contact pressure to break through the pol ymer fi l m i t i s much better to standardi se on metal work that rel i es on good metal pl ati ng to prevent corrosi on, and avoi d pol ymer passi vati on compl etel y. Some i ndustri al cabi nets are seam- wel ded, such as the stai nl ess-steel cabi nets used i n the food, pharmaceuti cal and other i ndustri es where hygi ene i s i mportant. These make excel l ent RF References, but most i ndustri al cabi nets consi st of frames to whi ch metal cl addi ng (ski n) i s appl i ed. These may be fi tted wi th shel ves for mounti ng l arge i tems of equi pment (for exampl e the tradi ti onal 1 9- i nch rack-mounti ng system), or backpl ates for mounti ng D N rai l s and/or ' chassi s- mounted' uni ts or modul es. These cabi nets consi st of many metal parts al l screwed together, and to make them functi on useful l y as an RF Reference we must RF-bond them together al l over the cabi net. Any metal part i n a cabi net that i s not RF-bonded to form part of the RF Reference, i ncl udi ng hi nged doors, i s a potenti al 'acci dental antenna' as shown i n Fi gure 3. 1 1 1 2 2. 4 Making effective RFbonds Fi gures 4 and 5 show the pri nci pl es of RF- bondi ng between two sheets of metal , the i mportant i ssues bei ng the achi evement of suffi ci entl y l ow RF i mpedances i n the bonds themsel ves, and the maxi mum spaci ng between the bondi ng poi nts.
Al though Fi gures 4 and 5 show two metal sheets bei ng bonded to form a l arger RF Reference, the same techni ques appl y to any metal parts, and i n three di mensi ons. Where a metal mesh i s bei ng used i nstead of a metal sheet or other sol i d metal part, i t can onl y be rel i ed upon to act as a moderatel y effecti ve RF Reference up to a frequency of 1 5/ MHz ( i n metres). Cl oser mesh spaci ng gi ves the Reference a better RF performance at any frequency up to 1 5/ MHz. Fig 4 Fig 5 Fi gures 6 and 7 show detai l s of metal -to- metal bonds used to connect metal parts together to create RF Reference pl anes. Any other ki nds of bonds, i ncl udi ng short wi de brai d straps, are very i nferi or to metal -to-metal bonds al though of course i t depends upon the frequency, and brai d straps or even bond wi res may be abl e to provi de a suffi ci entl y l ow- i mpedance RF Reference, i f the hi ghest frequency that i s to be control l ed i s no more than a few hundred kHz. Fi gure 6 shows the use of aggressi ve ' spi ky' washers and screw threads for RF- bondi ng two metal i tems that have an i nsul ati ng fi ni sh, such as pai nt, anodi si ng or even a pol ymer passi vati on coati ng. When tryi ng to i mprove the EM performance of an exi sti ng cabi net, thi s mi ght be al l that can be achi eved but i t i s not an i deal method (al though i t i s much better than usi ng wi res or brai d bonds). Fig 6 1 3 1 4 As menti oned earl i er, i t i s much better to desi gn cabi nets i n the fi rst pl ace to use conducti vel y-pl ated metal parts throughout, wi th no pai nt or i nsul ati ng fi ni shes appl i ed. Then the very best and most rel i abl e RF bonds can be achi eved by usi ng the fi xi ng screws to press the conducti ng metal surfaces together. Steel (apart from stai nl ess) and al umi ni um are unsui tabl e materi al s on thei r own, they al ways devel op a hi gh resi stance surface through oxi dati on, so al ways need to be ti n pl ated or al ochromed, or some other l ow resi stance corrosi on-proof fi ni sh. The RF Reference i n ' 1 9 i nch' rack - mounti ng cabi nets can be i mproved by maki ng sure that the front panel s of al l the racked equi pment make metal -to-metal bonds to the cabi net frame at thei r mounti ng poi nts (someti mes cal l ed rack mounti ng ears). Typi cal mounti ng ears are made of anodi sed al umi ni um, and because anodi si ng i s a very tough i nsul ator they provi de no RF-bondi ng. t i s much better to use ears wi th hi ghl y conducti ve pl ati ng, screwed i nto caged nuts i n a frame that al so has hi ghl y conducti ve pl ati ng. The 1 9 i nch spaci ng between the bonds at both si des of the front panel s means that the RF Reference thus created has a l ow i mpedance up to about 20MHz to control hi gher frequenci es more effecti vel y i t woul d hel p to create RF bonds al ong the l ong edges of the front panel s, to the shel ves above and bel ow. A vari ety of speci al EMC tapes are avai l abl e from compani es such as 3M, whi ch can be used to provi de a good hi gh- conducti vi ty bondi ng surface (usual l y ti n) i nstead of rel yi ng on pl ati ng. Some types have a top l ayer of maski ng tape so that the metal parts can be pai nted, then the maski ng tape removed to reveal the metal surface where the bondi ng i s to take pl ace. Fig 7 Spot-wel di ng i s as good or better than pressi ng metal -to-metal at fi xi ng poi nts, rememberi ng to space them cl oser together to al l ow for the ones that don' t work. Seam-wel di ng (or brazi ng or seam- sol deri ng) al ong al l j oi nts i s even better, and i s used where the hi ghest RF performance i s requi red. An al ternati ve to seam wel di ng/brazi ng/sol deri ng i n the creati on of an RF Reference i s to use conducti ve gaskets, often cal l ed EMC gaskets, to provi de l ow-i mpedance bondi ng al l al ong a metal j oi nt. Such gaskets are mostl y used to create EMC shi el ded cabi nets, and the way i n whi ch they are used to hel p create RF References i s no di fferent. Usi ng gaskets i nstead of mul ti pl e screw fi xi ngs and/or wel di ng hel ps speed the assembl y, and di sassembl y of cabi nets. 2. 5 Using gaskets effectively There are many suppl i ers of such gaskets, and each one offers very many di fferent gasket materi al s i n many di fferent styl es (see Fi gures 8 and 9 for some exampl es of j ust two types of gasket materi al s) because no one type of gasket i s sui tabl e for al l appl i cati ons. Thi s gui de wi l l not di scuss gaskets and thei r use i n any detai l , except to say that when assembl ed they shoul d be compressed to an amount wi thi n thei r manufacturers recommended range and thi s can requi re consi derabl e pressure. Good EMC gasket manufacturers provi de a weal th of data and appl i cati on assi stance (for exampl e [27] ), coveri ng the correct choi ce of gasket materi al s and styl es for parti cul ar appl i cati ons, and the data requi red for correct mechani cal desi gn. Even gaskets that are easi l y squashed fl at between two fi ngers can requi re very l arge compressi on forces when used i n l ong stri ps, so the effecti ve use of gaskets requi res careful mechani cal and fi xi ng desi gn to prevent metal parts from bendi ng too much. t i s not unusual to fi t stri ps of very soft conducti ve gaskets to the door of an i ndustri al cabi net, onl y to fi nd that i t becomes al most i mpossi bl e to cl ose, and once cl osed i t bends l i ke a banana openi ng up l arge gaps that defeat the purpose of the gasketti ng. 1 5 1 6 Fig 8 Fig 9 3 Wi ri ng and cabI i ng techni ques 3. 1 Routing send and return paths close together Al l el ectri cal and el ectroni c power (AC or DC) and si gnal s (whether data, control , anal ogue, i nputs, outputs, etc. ) have a current that fl ows i n a l oop from the source to the l oad (dri ver to recei ver) and back agai n. t i s vital for EMC that the area encl osed by thi s l oop i s as smal l as possi bl e, whi ch means that we must provi de a wi re for the send current path, and a conductor for the return current path, and route them together i n the cl osest possi bl e proxi mi ty over thei r whol e route, as shown i n Fi gure 1 0. t i s best to twi st the send and return conductors together, to make a twi sted- pai r cabl e (someti mes twi sted tri pl es or quads are requi red i nstead, for i nstance for three-phase AC wi th ei ther three or four wi res). Shi el ded (screened) twi sted- pai rs can al so be used and are very good i ndeed for EMC, but coaxi al cabl es are not as good, and ri bbon cabl es and bundl es of i ndi vi dual wi res can cause bi g probl ems for EMC. Fi gures 1 1 and 1 2 show some methods for i mprovi ng the EMC of unshi el ded wi res. For more i nformati on on getti ng the best EMC performance from wi re bundl es and cabl es (unshi el ded or shi el ded), see the 2006 versi on of [28] . Fig 10 1 7 1 8 Fig 11 Fig 12 Where thousands of amps of current are i nvol ved, twi sti ng the send and return conductors can cause i nsul ati on damage due to the el ectromechani cal forces on the conductors but i nsi de i ndustri al cabi nets such currents are carri ed by sol i d busbars anyway. Busbars cannot be twi sted, and thei r separati on means that thei r currents create hi gh l evel s of magneti c and el ectri c fi el ds i n thei r vi ci ni ty, whi ch can be a probl em for nearby el ectroni cs. Busbars can make excel l ent ' acci dental antennas' for any RF vol tages or currents they carry, so fi l teri ng may be needed to reduce thei r l evel s of RF, or el se the cabi net may need to be shi el ded. The best busbars for EMC use sol i d i nsul ati on (i nstead of ai r) and very cl ose spaci ng between bars carryi ng send and return currents. n three-phase mai ns suppl i es, or three-phase motor dri ve cabl es, each of the phases i s the return for the others. 3. 2 Routing cables close to the RF-bonded metalwork As di scussed i n 1 . 4, an RF Reference i s onl y useful for assi sti ng or i mprovi ng the EM performance of a ci rcui t i f i t i s l ocal meani ng cl oser than /1 0 at the hi ghest frequency to be control l ed (e. g. cl oser 30mm to control up to 1 00MHz) much cl oser spaci ng means better EMC. Thi s secti on di scusses techni ques for achi evi ng the appropri ate spaci ng for wi re bundl es and cabl es. Secti on 4 di scusses the spaci ng and bondi ng of el ectroni c uni ts wi th respect to the RF Reference. Wi res and cabl es, and bundl es of wi res and cabl es, shoul d never fl y through the ai r. They shoul d i nstead be routed al ong thei r entire l engths as cl ose as possi bl e to metal work that has a conti nuous conducti ve path wi th a l ow RF i mpedance, al l the way back to the RF Reference pl ane, as shown i n Fi gure 1 3. Fig 13 1 9 20 Thi s metal work creates a preferenti al return path for the common-mode (CM) currents that l eak from al l wi res and cabl es (even shi el ded ones), reduci ng the acci dental antenna effects shown i n Fi gure 3, and further i mproves EMC by provi di ng the wi res and cabl es wi th what i s known as an 'i mage pl ane'. The wi der the metal , the l ower the RF i mpedance and the better the i mage pl ane, so the better the EMC. f the metal structure i s a part of the RF Reference, bonded to other parts of the Reference as descri bed i n secti on 2 above that woul d be i deal for EMC. ndustri al cabi nets that use backpl ates general l y use pl asti c trunki ng for thei r wi res and cabl es, and the tendency i s to si mpl y stuff al l the wi res and cabl es i nsi de the trunki ng and cl i p i ts l i d on. But where the send and return conductors are si ngl e wi res i t woul d be better to twi st them or at l east ti e them together i n a bundl e (see Fi gure 1 1 ) before pl aci ng them i n the trunki ng, so that they cannot l i e too far from each other. Pl asti c trunki ng hel ps to keep conductors reasonabl y cl ose to the backpl ate (whi ch i s al ways the RF Reference) but some of the wi res and cabl es mi ght l i e as much as 50 or even 75mm above i t, whi ch does not gi ve the best EM performance. t woul d be better for EMC to use 'shorter' trunki ng, that keeps conductors cl oser to the backpl ate, at l east for the more EMC- cri ti cal cabl es (Cl asses 1 and 4, see 3. 3 bel ow), or el se ti e these cabl es di rectl y to the backpl ate. Fi gures 1 4 and 1 5 show how to deal wi th the routi ng of wi res and cabl es i n corners and across j oi nts i n the RF Reference. Fig 14 Fig 15 Conductors carryi ng power or si gnal s shoul d not go too cl ose to the edge of the metal structures they are routed cl ose to. Where practi cal thei r di stance from the edge of the metal shoul d be at l east three ti mes thei r hei ght above i t, and thi s i s especi al l y i mportant for Cl asses 1 and 4 (see 3. 3). Where conductors connect to el ectroni c pri nted ci rcui t boards (PCBs), uni ts, modul es or other products, they shoul d be routed very cl ose to the RF Reference as much as i s possi bl e al ong thei r enti re l ength. deal l y, the PCBs, uni ts, etc. , wi l l al so be RF-bonded to the Reference (see secti on 4), and i f they have a metal body the conductors shoul d be routed cl ose to that unti l reachi ng thei r connectors or termi nal s. Fi gure 1 6 i l l ustrates thi s for a packaged uni t, whi l e Fi gure 1 7 shows a fl at cabl e to a PCB. Shi el ded cabl es shoul d have thei r shi el ds bonded to the RF Reference at thei r ends, or as cl ose to thei r ends as possi bl e, and thi s i s di scussed i n secti on 3. 7. 21 22 Fig 16 Fig 17 3. 3 Segregating different Classes of conductors Cabl es i nsi de a cabi net shoul d be spl i t i nto at l east 4 cl asses, based on the gui dance i n [1 8] , as fol l ows: CI ass 1 i s for conductors carryi ng very sensi ti ve power or si gnal s. Low-l evel anal ogue si gnal s such as mi l l i Vol t output transducers and radi o recei ver antennae are i n Cl ass 1 A. Hi gh-rate di gi tal communi cati ons such as Ethernet are i n Cl ass 1 B. Cl asses 1 A and 1 B shoul d not be bundl ed together, al though thei r bundl es may be run adj acent to each other. Al l Cl ass 1 cabl es shoul d use ful l y shi el ded cabl es and connectors over thei r enti re path, wi th 360 shi el di ng mai ntai ned throughout, from end-to- end (see [28] ). Unshi el ded twi sted- pai rs are commonl y used for Ethernets and si mi l ar data cabl es, but they are general l y not as good for achi evi ng the ful l data rate or EMC as shi el ded twi sted pai rs of otherwi se i denti cal speci fi cati on. CI ass 2 i s for conductors carryi ng sl i ghtl y sensi ti ve power or si gnal s, such as ordi nary anal ogue (e. g. 4-20mA, 0- 1 0V, and si gnal s under 1 MHz), l ow- rate di gi tal communi cati ons (e. g. RS422, RS485), and di gi tal i nputs or outputs (i . e. on/off si gnal s, not seri al or paral l el datacommuni cati ons; for exampl e si gnal s from l i mi t swi tches, encoders, pushbuttons, etc. ). CI ass 3 i s for sl i ghtl y i nterferi ng power or si gnal s, such as l ow vol tage AC di stri buti on (< 1 kV) or DC power (e. g. 48V tel ecommuni cati ons power), where these do not al so power noi sy apparatus. Power di stri buti on that al so feeds noi sy equi pment may be converted from Cl ass 4 to Cl ass 3 by the correct appl i cati on of fi l teri ng (not a tri vi al exerci se, see [29] ). Cl ass 3 al so embraces control ci rcui ts wi th resi sti ve and i nducti ve l oads, where the i nducti ve l oads are suppressed at the l oad (e. g. the el ectri cal coi l s of rel ays, contactors, sol enoi ds, actuators, val ves, etc. ); di rect-on-l i ne (DOL) AC motors, and so-cal l ed ' sparkl ess' or 'pancake' DC motors. CI ass 4 i s reserved for strongl y i nterferi ng power or si gnal s. Thi s i ncl udes al l the power i nputs or outputs (to or from) vari abl e-speed AC motor dri ves; frequency converters; AC-AC and AC-DC power converters and thei r DC l i nks, and DC-DC power converters. Cl ass 4 al so appl i es to the cabl es associ ated wi th el ectri cal wel ders; RF equi pment (e. g. pl asti c wel ders, wood gl uers, di athermi c apparatus, mi crowave dryers and ovens); DC motors or sl i pri ngs; and si mi l ar 'noi sy' apparatus. Cabl es to RF transmi tti ng antennae and unsuppressed i nducti ve l oads are al so Cl ass 4. Al l Cl ass 4 cabl es shoul d use shi el ded cabl es and connectors wi th 360 shi el di ng mai ntai ned throughout, from end-to-end (see [28]). The swi tch-mode power el ectroni c ci rcui ts used i n vari abl e-speed AC motor dri ves; frequency converters; AC- AC and AC-DC power converters; DC- DC power converters and the l i ke (i ncl udi ng most types of uni nterrupti bl e power suppl y) produce very hi gh l evel s of RF noi se on thei r power i nputs and outputs, whi ch i s why they shoul d be assumed to be Cl ass 4 i n the absence of any EMC test data. However, thei r i nputs and outputs can be fi l tered to 23 24 reduce thei r cabl es to Cl ass 3, or even to Cl ass 2. For exampl e, most chassi s-mounted and modul ar DC power suppl i es contai n i nput and output fi l ters and are cl ai med by thei r suppl i ers to compl y wi th the rel evant emi ssi ons standards (usual l y C SPR 22, EN 55022, EN/ EC 61 000-6-3 or EN/ EC 61 000-6-4). Fi l ters are avai l abl e from numerous suppl i ers for the mai ns i nputs of al l swi tch-mode power converters, and responsi bl e manufacturers of converters shoul d at l east recommend whi ch makes/model s shoul d be used to be sure of compl yi ng wi th rel evant conducted mai ns emi ssi ons standards i f they are not al ready i ncorporated i n thei r uni t. A few fi l ter manufacturers al so make fi l ters sui tabl e for fi tti ng at the output of swi tch-mode power converters, ei ther to remove the worst of the RF noi se, or to convert the output waveform i nto a si newave or DC (as appropri ate) wi th varyi ng degrees of puri ty.
Most vari abl e-speed AC motor dri ve manufacturers speci fy the use of a shi el ded cabl e 360 bonded at both ends, for thei r motor cabl es, but when fi tted wi th a sui tabl e ' si nusoi dal output fi l ter' thei r motor cabl es can be treated as Cl ass 3 i nstead. The cost of such fi l ters appears to di scourage many i ndustri al cabi net desi gners from usi ng them, but overal l there are usual l y si gni fi cant fi nanci al and EMC benefi ts to be had i n usi ng them to el i mi nate Cl ass 4 shi el ded cabl es and thei r segregated routes from the fi nal i nstal l ati on. These four cl asses shoul d be physi cal l y segregated wi thi n the cabi net at al l ti mes, and as Fi gure 1 8 shows, l ong paral l el runs shoul d not be any cl oser than 1 00mm, i f possi bl e, as wel l as bei ng run as cl ose as possi bl e to the RF Reference. f cabl es of di fferent cl asses must cross over each other, they shoul d onl y do so at ri ght angl es. Greater spaci ngs are requi red for paral l el routes outsi de the cabi net, as shown by the l ower part of Fi gure 1 8, but good EMC engi neeri ng techni ques for systems and i nstal l ati ons are not the subj ect of thi s gui de (see [3] i nstead). The above cl assi fi cati on i s based on the power and si gnal s that are intended to be i n the conductors, but i t mi ght be necessary to i ncrease the cl assi fi cati on of a conductor dependi ng on other factors. For exampl e, many modern el ectroni c devi ces have /O si gnal s that woul d appear to be rel ati vel y beni gn i n EMC terms (e. g. audi o outputs, i nputs from pushbuttons, thermocoupl e i nputs, i ndi cator l amp outputs). What i s often not real i sed i s that where there i s di gi tal processi ng or swi tch-mode power conversi on wi thi n a uni t, these /Os can carry hi gh l evel s of unrel ated CM noi se that can have a very si gni fi cant RF content. Harmoni cs of di gi tal cl ocks and data busses are usual l y the chi ef cul pri ts, often causi ng emi ssi ons i n the hundreds of MHz from conductors ostensi bl y carryi ng very i nnocuous si gnal s. Home-made cl ose-fi el d probes and l ow- cost portabl e spectrum anal ysers, such as those shown i n Fi gure 1 9, can be used as descri bed i n Parts 1 and 2 of [1 2], to i denti fy such probl ems earl y i n a proj ect. They hel p choose i ndustri al components that have fewer EMC probl ems, and/or hel p choose the appropri ate type of shi el ded cabl es and connectors (see [28]) and/or choose appropri ate fi l ters, such as cl i p-on ferri te suppressi on chokes (see [29]) so that cabl es can be bundl ed wi th others of the same Cl asses wi thout causi ng i nterference. Fig 18 Fig 19 25 26 Another probl em ari ses wi th cabl es exposed to the strong RF fi el ds near the antennas of radi o transmi tters such as GSM, GPRS, Wi -Fi , Bl uetooth, 3G or other radi o transmi tti ng devi ces often i ncorporated i nto i ndustri al systems these days, or where cabl es are exposed to powerful RF fi el ds from radi o or TV broadcasti ng stati ons, hand-hel d wal ki e- tal ki es, vehi cl e mobi l es, and RF producti on equi pment such as i nducti on furnaces, di el ectri c heaters, pl asti c wel ders or seal ers, mi crowave dryers, etc. Thi s i s especi al l y a probl em for cabl es that l eave or enter a cabi net, so are exposed to the external EM envi ronment. Thi s i ssue i s deal t wi th by assessi ng the external EM envi ronment as descri bed i n [1 5] , then choosi ng the appropri ate types of cabl es and connectors, and/or fi l teri ng so that cabl es exposed to such fi el ds can be bundl ed wi th others of the same cl asses wi thout causi ng i nterference. The recommended spaci ngs between cabl e cl asses, as shown i n Fi gure 1 8, are based upon a number of assumpti ons about the types of cabl es and the ci rcui ts i n the el ectri cal /el ectroni c uni ts connected to the cabl es, and real -l i fe experi ence, and so they are at best a very rough gui de and cannot be expected to be adequate i n al l cases. ncreasi ng the number of cabl e cl asses used to fi ve si x or more; i ncreasi ng the spaci ng between paral l el routes, and routi ng cl oser to the RF Reference are al l ways of i mprovi ng EM performance more than woul d be achi eved by the above gui de. Actual l y testi ng cabl es (when they are operati ng wi th thei r i ntended power/si gnal s and l oads) wi th cl ose-fi el d probes and spectrum anal ysers, as di scussed above, i s a very bi g hel p i n removi ng the guesswork from thi s whol e i ssue, and i s general l y recommended no matter what EM speci fi cati ons suppl i ers cl ai m thei r products meet. 3. 4 Reducing Class spacings Where wi res and cabl es cannot be routed cl ose enough to the RF Reference, the spaci ngs between paral l el routed cabl es of di fferent cl asses shoul d be si gni fi cantl y i ncreased beyond the recommendati ons i n Fi gure 1 8. Where practi cal consi derati ons prevent the achi evement of the i deal spaci ngs between cl asses, usi ng conductors wi th i mproved EM performance wi l l reduce the spaci ngs requi red. Thi s techni que i nvol ves. Addi ng more return conductors i n wi re bundl es, see Fi gures 1 1 , 1 2 and the 2006 versi on of [28] Repl aci ng strai ght send/return wi res i n bundl es wi th twi sted pai rs, tri pl es or quads as appropri ate Repl aci ng unshi el ded (unscreened) cabl es by shi el ded types usi ng correct shi el d termi nati ons at both ends, see [28] Repl aci ng shi el ded (screened) cabl es by types wi th a hi gher shi el di ng speci fi cati ons and/or hi gher-qual i ty termi nati ons at both ends, see [28] .
t i s possi bl e to i mprove the EM
performance of conductors by so much, that al l the cl asses can be bundl ed together. The l i mi ti ng factor i s usual l y the connectors provi ded by the manufacturers of the el ectri cal /el ectroni c uni ts used i n the cabi net. For exampl e, screw-termi nal or pl asti c-bodi ed connectors make i t di ffi cul t to termi nate cabl e shi el ds correctl y (thi s requi res 360 bonds not pi gtai l s see [28]). Even where D-type connectors wi th metal bodi es are provi ded i f the manufacturer of the uni t has not correctl y bonded the D- type to hi s uni ts i nternal RF Reference, the EM performance avai l abl e from 360 termi nati ng the cabl e shi el d wi l l be wasted. t makes good sense to consi der such aspects of equi pment before purchasi ng them. t i s al so possi bl e to i mprove the EM performance and reduce the spaci ngs between cl asses by i mprovi ng the filtering appl i ed to the el ectri cal /el ectroni c uni ts at one or both ends of the conductors. t i s i mpracti cal to open up purchased uni ts to i mprove thei r i nternal fi l teri ng, so i t i s more usual to add external fi l ters to the cabl es at the poi nt where they enter/exi t the uni ts. The easi est types of fi l ters to add to cabl es are ferri te chokes, a wi de range of whi ch are avai l abl e from many manufacturers to sui t round or fl at cabl es, many of them avai l abl e i n spl i t form wi th pl asti c cl i ps that make i t easy to cl i p them onto exi sti ng cabl es, as shown i n Fi gure 20. Cabl e-mounted ferri tes are avai l abl e i n di fferent materi al s, to suppress di fferent ranges of frequenci es, so i t obvi ousl y hel ps to choose the appropri ate type. f the probl em frequenci es are unknown, they can be found wi th a cl ose-fi el d probe and spectrum anal yser such as shown i n Fi gure 1 9, or wi th the CM cabl e-current probe shown i n Fi gure 5 of Part 1 of [1 2] . t i s al so possi bl e to connect fi l ters such as those shown i n Fi gure 21 i n seri es wi th cabl es, to i mprove thei r EM performance, reduce cabl e Cl ass spaci ngs, or even to change a cabl e from one Cl ass to a l ower. Whereas addi ng ferri te chokes such as those i n Fi gure 20 i s qui te strai ghtforward, there are a number of detai l ed i ssues surroundi ng the effecti ve use of fi l ters that contai n capaci tors, such as those i n Fi gure 21 . Some of these i ssues are di scussed i n secti ons 4. 7 and 5. 2, but for more detai l pl ease read [29] . Fig 20 27 28 Someti mes fi l teri ng and shi el di ng can be effecti vel y appl i ed together. For i nstance, where a shi el ded cabl e can onl y achi eve a poor shi el d termi nati on (e. g. a pi gtai l , see secti on 3. 7. 6), a ferri te cl amped over the cabl e at that end can i mprove matters. 3. 5 Segregating cables in industrial cabinets that use backplates The vari ous el ectroni c, el ectri cal , pneumati c, hydraul i c, etc. , uni ts shoul d be l ocated to keep sensi ti ve uni ts such as transducer ampl i fi ers or programmabl e l ogi c control l ers (PLCs) wel l away from el ectri cal l y noi sy uni ts such as rel ays and contactors or vari abl e-speed AC motor dri ves ('i nverters'), to hel p prevent them from i nterferi ng wi th each other. They shoul d al so be l ocated so as to ai d the segregati on of the cabl e cl asses. Fi gure 22 i s a sketch of a real -l i fe i ndustri al control panel , usi ng an ordi nary unshi el ded cabi net, whi ch was desi gned accordi ng to thi s gui de and successful l y tested for EMC compl i ance. Fig 21 Fig 22 Noti ce that the Cl ass 1 and 2 cabl es i n Fi gure 22 are run i n the same trunki ng thi s was a compromi se that was fel t to be acceptabl e because thi s panel was qui te smal l , and al so because there were very few Cl ass 1 si gnal s and they were not very sensi ti ve. There i s nothi ng wrong wi th compromi se, i t i s the l i fe-bl ood of engi neeri ng after al l , but i t i s very i mportant that peopl e competent i n the di sci pl i nes i nvol ved (EMC i n thi s case) determi ne such compromi ses, case-by- case. As i n Fi gure 22, i t i s best to try to have no i nternal Cl ass 4 cabl es at al l , or at l east mi ni mi se thei r i nternal l engths as much as possi bl e. Thi s means fi tti ng uni ts such as i nverter motor dri ves near to the wal l of the cabi net so that thei r motor dri ve cabl es (Cl ass 4) can exi t di rectl y, and fi l teri ng any Cl ass 4 i ncomi ng cabl es (e. g. very noi sy mai ns suppl i es) as cl ose as possi bl e to thei r poi nt of entry to make them Cl ass 3 or 2. Fi gure 23 shows more detai l of the motor dri ve area of the cabi net from Fi gure 22. The manufacturer' s EMC i nstructi ons Fig 23 29 30 Fig 24 the ones wi th the hi ghest l evel s of RF noi se (rel ays, contactors, swi tch-mode power converters such as vari abl e-speed AC motor dri ves and the l i ke, etc. ) far away from the uni ts that are the most sensi ti ve (computers, PLCs, di spl ays, transducer ampl i fi ers, tel ecommuni cati ons, etc. ). Where the rack-mounted uni ts are purchased from a vari ety of suppl i ers there wi l l probabl y be no consi stency at al l between thei r rear-panel connector l ayouts, and often no segregati on between di fferent cabl e cl asses ei ther. n such cases i t i s i mportant to determi ne whi ch cabl es bel ong to whi ch Cl ass, and segregate them as cl ose as possi bl e to the rack uni ts so that they can be bundl ed wi th thei r own Cl ass for routi ng wi thi n the cabi net. Creati ng an RF Reference for conductors to be routed agai nst i s not as easy when Fig 25 usi ng a rack cabi net, as i t i s for a backpl ate- type i ndustri al cabi net. General l y, i t requi res provi di ng hori zontal shel ves for the rack uni t to stand upon (as wel l as the uni ts bei ng bonded to the frame vi a thei r rack-mount 'ears'). These shel ves shoul d extend wel l beyond the rear panel s of the uni ts, and be RF-bonded to verti cal metal sheets at the si des of the cabi net. Cabl es enteri ng/exi ti ng the rear panel s of the racked uni ts can then be routed cl ose to thei r uni ts shel ves, where they wi l l be sorted out i nto thei r Cl ass bundl es, then those bundl es routed to the si de sheets to run verti cal l y between the shel ves. Of course, al l the shel ves and verti cal sheets shoul d be RF-bonded to each other and to the frame and other metal parts of the cabi net, as shown i n Fi gures 4 7, to hel p create a good RF Reference as descri bed i n secti on 2, and the cabl es or cabl e bundl es shoul d be routed as shown i n Fi gures 1 4 and 1 5. for thi s vari abl e-speed AC motor dri ve requi red the use of a speci fi ed ferri te toroi d wi th a speci fi ed number of turns on i t, pl us the use of shi el ded motor cabl e wi th the shi el d termi nated 360 (see [28]) to the motor dri ve' s metal chassi s, and al so at the motors metal termi nal box (a typi cal requi rement for i nverter dri ves that do not have 'si nusoi dal output' fi l ters fi tted to thei r motor cabl es very cl ose to the dri ve). Fi gure 24 shows the remai nder of the route of the bundl e of cabl es i n the top l eft-hand-corner of Fi gure 22, whi ch are l eavi ng the backpl ate area to connect to el ectroni c uni ts mounted on the door of thi s ordi nary unshi el ded i ndustri al cabi net. The cabl es are run cl ose to the l ocal RF Reference over thei r whol e l ength, whi ch means exi ti ng the backpl ate at one of the poi nts where i t i s RF-bonded to the cabi net wal l or frame, actual l y ti ed down i n such a way that i t fol l ows the route of the RF bond as near as possi bl e. t i s then routed al ong the wal l or frame unti l i t crosses to the door strapped to a short brai d that RF-bonds the door to the cabi net wal l . Hi nges cannot be rel i ed upon to el ectri cal l y bond doors to cabi nets, as they usual l y contai n grease or pl asti c i nserts. n a wel l -shi el ded cabi net the door wi l l be bonded al l around by a conducti ve gasket, so i t does not matter where the cabl es cross the hi nge area. 3. 6 Segregating cables in rack- mounted equipment Where the desi gn of the rack-mounted uni ts can be control l ed, for exampl e when they are made by the same company, they shoul d segregate thei r rear-panel connectors to faci l i tate the segregati on of cabl e cl asses wi thi n the cabi net, as shown i n outl i ne i n Fi gure 25. Noti ce al so that Fi gure 25 shows how the racked uni ts shoul d be organi sed to pl ace 31 32 Some i ndustri al cabi net manufacturers make thei r own rack chassi s uni ts, usual l y where they need functi ons not avai l abl e as a standard product, or where they hope to reduce cost. Fi gure 26 shows an exampl e provi di ng general gui dance for the pl acement of el ectroni c uni ts and routi ng of i nterconnecti ng cabl es wi thi n a uni t. t shows the sensi ti ve uni ts kept far away from the ones that create the most RF noi se (the di gi tal processi ng and the swi tch-mode power suppl y), and the cabl es routed cl ose to the metal chassi s and segregated by functi on i n a si mi l ar manner to the Cl ass structure di scussed earl i er. Thi s gui de i s not i ntended for the manufacturers of el ectri cal /el ectroni c uni ts i t i s meant for i ndustri al cabi net manufacturers who purchase such uni ts from thi rd-party suppl i ers. Compani es that wi sh to know more about the good EMC practi ces i n the desi gn and assembl y of el ectri cal /el ectroni c uni ts are recommended to read and appl y al l of the Fig 26 seri es that i ncl udes [28] [29] [30] and [31 ] , and they wi l l al so fi nd [1 2] useful . 3. 7 Bonding cable shields (screens) to the RFReference 3. 7. 1 Bondi ng cabI e shi eI ds at both ends i s good EMC engi neeri ng practi ce Cabl e shi el ds shoul d general l y be bonded to thei r l ocal RF References at both ends. Thi s i s because a shi el ded cabl e that i s onl y bonded at one end can onl y provi de good shi el di ng performance up to a frequency at whi ch i ts l ength becomes a si gni fi cant fracti on of the wavel ength. The hi gher the shi el di ng effecti veness requi red for the cabl e, the smal l er the fracti on of a wavel ength permi tted. To put some rough gui del i nes to thi s: for shi el di ng of around 20dB at a gi ven frequency wi th the shi el d onl y bonded at one end the cabl e l ength shoul d be l ess than one-twenti eth of the wavel ength at that frequency. For exampl e, at 400MHz (cl ose to a typi cal transmi tti ng frequency for wal ki e-tal ki es used i n i ndustri al premi ses) one wavel ength i n ai r i s 0. 75m, and one- twenti eth of that i s 37. 5mm. Thi s means that shi el ded cabl es whi ch have thei r shi el ds bonded at one end onl y, shoul d be no l onger than 37. 5mm to mai ntai n a shi el di ng effecti veness of at l east 20dB at up to 400MHz (to hel p prevent cl ose proxi mi ty of wal ki e-tal ki es resul ti ng i n i nterference). f cabl es need to be l onger than thi s (and most wi l l be) they wi l l need to have thei r shi el ds termi nated at both ends. Of course, nothi ng i s as si mpl e as thi s, and most types of fl exi bl e shi el ded cabl es wi l l be l osi ng thei r shi el di ng effecti veness by 400MHz. Brai ded cabl es general l y gi ve better shi el di ng performance than wrapped foi l types, and are easi er to termi nate i n 360 fi tti ngs i n connectors and gl ands. The better brai d-shi el ded cabl es general l y have good opti cal brai d coverage, and doubl e- brai d or brai d-and-foi l may be used to gi ve even better performance at hi gher frequenci es. The very best fl exi bl e shi el ded cabl es are the (expensi ve) 'superscreened' cabl e types, whi ch empl oy mul ti pl e shi el di ng brai ds as wel l as at l east one l ayer of ' MuMetal ' tape. Another probl em wi th bondi ng a cabl e's shi el d at onl y one end, i s that i t then cannot provi de any shi el di ng at al l from some ori entati ons of magneti c fi el ds. Shi el di ng from these requi res a current to fl ow i n the shi el d from one end to the other, whi ch can' t happen i f the shi el d i s onl y bonded to the Reference at one end. n some more extreme i ndustri al envi ronments there can be si gni fi cant potenti al di fferences between the l ocal RF References of i tems of equi pment l ocated i n di fferent areas of the si te. These vol tages are usual l y at the frequency of the AC power suppl y, typi cal l y ei ther 50 or 60Hz. Bondi ng both ends of the shi el ds of cabl es that i nterconnect these i tems can cause hi gh l evel s of shi el d current to fl ow. Thi s i s a probl em for the i nstal l ati on rather than the i nternal assembl y of a product, and i s deal t wi th by the use of techni ques such as the meshed common bondi ng network (MESH-CBN) and the paral l el earthi ng conductor (PEC) both of whi ch are descri bed i n more detai l i n [3] and [1 8] . However, cabi nets shoul d be desi gned to al l ow thei r i nstal l ati on to use good EMC engi neeri ng practi ces, so shoul d provi de fi xi ngs sui tabl e for the connecti ons of the external PECs (whi ch coul d be wi res, ducts, armour, trays, etc. ) see [3] and [1 8]. Wi th properl y desi gned el ectroni cs, the onl y si gni fi cant consequences of shi el d currents i s heati ng of the cabl es so- cal l ed 'hum l oop' or 'ground l oop' noi se are a consequence of poor el ectroni cs desi gn, whi ch al l ows cabl e shi el d noi se currents to fl ow di rectl y i nto el ectroni c ci rcui ts (usual l y by connecti ng the cabl e shi el d di rectl y to the ci rcui t' s 0V). Thi s i ssue i s outsi de the scope of thi s arti cl e, but i s deal t wi th i n more detai l i n secti on 2. 6. 8 of the 2006 versi on of [28]. Of course, i ndustri al cabi net manufacturers usual l y rel y on thi rd-party suppl i ers for thei r el ectroni c uni ts, whi ch i s why i t i s best for them to careful l y check the EMC i nstal l ati on i nstructi ons for any uni t, modul e or product they are consi deri ng, to fi nd out i f they require any cabl e shi el ds to be bonded at onl y one end (there are other thi ngs to di scover too, see [1 4]). Thi s i s typi cal l y an i ndi cati on of poor desi gn for EMC, al though i n some equi pment i ntended for use i n expl osi ve atmospheres, i t mi ght someti mes be necessary for safety reasons. 34 33 3. 7. 2 Capaci ti ve and hybri d shi eI d bondi ng f, for some reason, bondi ng the shi el d at both ends i s i mpracti cal , i t may prove acceptabl e to connect a short-l eaded cerami c capaci tor from one end of the of the cabl e's shi el d to i ts l ocal RF Reference (i nstead of di rectl y bondi ng i t 360 metal - to-metal ). Thi s method i s someti mes cal l ed hybri d shi el d bondi ng, because one end has a di rect bond to i ts l ocal RF Reference, whi l e the other has a capaci ti ve bond. ( f both ends use capaci tors i n seri es wi th thei r shi el d termi nati ons thi s i s known as capaci ti ve shi el d bondi ng. ) The frequenci es and frequency ranges over whi ch capaci ti ve and hybri d bondi ng are effecti ve depend upon the types of capaci tors used and thei r val ues. The l engths of the capaci tors' l eads and any wi res or conductors attached to them shoul d always be mi ni mi sed. The capaci tors shoul d be rated for the vol tages they have to wi thstand, and i n the case of cabl es external to the cabi net and l onger than about 1 0 metres they shoul d be rated to wi thstand overvol tage surges and transi ents of at l east 500V, and maybe as much as 1 0kV, dependi ng on the i nstal l ati on. These surges are typi cal l y caused by l i ghtni ng el ectromagneti c pul se (LEMP) and al so by i nduced coupl i ng from mai ns cabl es or l i ghtni ng conductors carryi ng l i ghtni ng surges that mi ght be routed nearby. There may al so be other sources of surge or transi ent overvol tages i n some types of i nstal l ati on, such as l arge AC or DC motors control l ed by el ectromechani cal contactors, capaci tor banks (e. g. for power factor correcti on), or superconducti ng magnets. Where safety i s a concern, the capaci tors used may need to be safety-rated (and i t i s recommended that they are purchased as safety-approved and thei r approval certi fi cates checked wi th thei r i ssui ng bodi es to make sure they are not forgeri es. ) Unfortunatel y, wi thout usi ng speci al (and expensi ve) annul ar capaci tors i t i s di ffi cul t to make capaci ti ve shi el d bondi ng work wel l at the hi gher frequenci es bei ng used by modern el ectroni c equi pment, or work wel l over a wi de range of frequenci es. So hybri d shi el d bondi ng i s a techni que best kept i n reserve to deal wi th speci al si tuati ons, such as where 360 bondi ng at both ends i s not possi bl e for some reason and the frequenci es for whi ch the cabl e needs to have good shi el di ng are grouped i nto a fai rl y narrow range. Where a cabi net provi des for cabl e shi el d bondi ng at both ends, on-si te repl acement of di rect bonds by capaci tors i s not too di ffi cul t, and removi ng the bonds al together (shoul d i t prove necessary) i s very easy. However, i f the cabi net was desi gned to have i ts cabl e shi el ds bonded at onl y one end onl y attempti ng to fi t capaci tors or 360 bonds at the other ends to sol ve EMC probl ems on-si te or duri ng compl i ance testi ng can be very di ffi cul t and ti me- consumi ng. 3. 7. 3 It i s best not to use a cabI e's shi eI d to carry the return current Wherever possi bl e, never use the shi el d of a cabl e as the return conductor for the el ectroni c si gnal s (di gi tal , control or anal ogue) or el ectri cal power carri ed by i ts conductors al ways use a twi sted pai r, or twi sted tri pl e, or twi st whatever number of conductors i t takes to ful l y embrace al l the send and return current paths for a gi ven power or si gnal connecti on. Coaxi al cabl es are often thought to be the best for control l i ng RF, because of thei r wi despread use i n RF and EMC test equi pment but i n such appl i cati ons the cabl es are al ways used as matched transmi ssi on l i nes, rarel y the case i n the i ndustri al worl d. When not used as matched transmi ssi on l i nes i n a control l ed-characteri sti c-i mpedance i nterconnecti on system, coaxi al cabl es are not as good for EMC as shi el ded twi sted pai rs, because they carry thei r return current i n thei r shi el d, i nstead of i n a dedi cated conductor. [28] goes i nto thi s i ssue i n more detai l . 3. 7. 4 Techni ques for bondi ng cabI e shi eI ds to the RF Reference External cabl es enteri ng a cabi net shoul d have thei r shi el ds RF-bonded to the cabi net's l ocal RF Reference as soon as they cross i ts boundary. Thi s appl i es even though they may al so be bonded i nternal l y to the same Reference at another pl ace, for i nstance at an el ectroni c uni t (see secti on 4). An obvi ous way to bond a shi el d to the RF Reference i s wi th a shi el d-bondi ng Fig 27 connector, such as the types shown i n Fi gure 27 (a D-type) and Fi gure 30 (a bayonet-l ocki ng ci rcul ar connector), wi th the chassi s-mounted mati ng connectors themsel ves bonded metal -to-metal to the RF Reference at the edge where the cabl es enter or exi t the cabi net. The D-type i n Fi gure 27 shows the cabl e shi el d bonded usi ng a saddl ecl amp, whi ch does not real l y provi de a 360 shi el d termi nati on but neverthel ess i s often an acceptabl e al ternati ve. Some D-types requi re the assembl er to make a pi gtai l from the brai d or the drai n wi re of a foi l - wrapped shi el d, and trap i t under a spri ng cl i p or screw head or sol der i t to the body of the connector, l i ke the connector shown i n Fi gure 28. These types are al l noti ceabl y i nferi or to the saddl ecl amp method shown i n Fi gure 27. D-type backshel l s are al so avai l abl e that provi de a proper 360 shi el d termi nati on, and these are general l y preferred. Many shi el ded D-type connector backshel l s do not provi de a strai n rel i ef cl amp for the cabl e j acket. n such si tuati ons, where the Fig 28 Fig 29 very best EMC performance i s not requi red, i t i s usual to fol d the shi el d back over the outer j acket and cl amp both the shi el d and j acket at the same ti me. But thi s makes the EMC performance depend greatl y on workmanshi p, so where the best EMC performance i s requi red as wel l as strai n rel i ef, a D-Type (or any other type of connector) shoul d provi de 360 bondi ng of the undisturbed shi el d pl us a strai n rel i ef cl amp for the cabl e' s overal l j acket. Some connector manufacturers offer shi el di ng backshel l systems for D-Type and other mul ti way rectangul ar connectors that combi ne both shi el d- bondi ng and strai n rel i ef functi ons i n a cri mp accessory that attaches a metal fl ange to the cabl e the fl ange bei ng cl amped by the backshel l when the connector i s fi nal l y assembl ed, as shown i n Fi gure 29. Shi el ded i ndustri al connectors are avai l abl e i n round and rectangul ar styl es that wi l l take very l arge number of pi ns, and carry si gnal s or power up to hi gh currents. Fi gure 30 shows a cross-secti on Fig 30 of a ci rcul ar connector that achi eves a very hi gh qual i ty of 360 bond between cabl e shi el d and connector body, and al so provi des a strai n rel i ef and envi ronmental seal . Many other types of connector and shi el d termi nati on exi st, but onl y those that make a 360 el ectri cal bond between the cabl e's shi el d, the connector' s backshel l , and the mati ng connector's backshel l (or the mounti ng panel of the mati ng connector) work wel l for EMC. Any connector bondi ng techni que that i nvol ves di sturbi ng the l ay of the foi l or brai d of the cabl e shi el d, or extendi ng i t wi th wi res (see the secti on on 'pi gtai l s' bel ow) wi l l compromi se the shi el di ng performance of the cabl e and/or the connector. Shi el ded cabl e gl ands can be used i nstead of connectors; to bond shi el ds to the RF Reference as a cabl e enters/exi ts an RF Reference, as shown i n Fi gure 31 . Gl ands that bond wi th uni form pressure al l around an undi sturbed cabl e shi el d (e. g. a 360 bondi ng 'i ri s' spri ng or ' kni tmesh' gasket) general l y gi ve the best RF performance, and an exampl e of thi s type 35 36 Fig 31 i s shown i n the top l eft of Fi gure 31 . Thi s type of gl and uses the same desi gn pri nci pl es as the ci rcul ar connector sketched i n Fi gure 30, and i s general l y the best type to use for good EMC performance. The type shown at the bottom-ri ght of Fi gure 31 rel i es upon the assembl er cutti ng the brai d and spreadi ng i t over a pl asti c part before assembl i ng i t to the metal part that bonds to the RF Reference. Al though thi s type of gl and has a l ower cost, the extra work requi red to assembl e i t costs more, and there i s al so the possi bi l i ty that the assembl er wi l l not spread the cut brai d evenl y, or make other mi stakes that degrade EMC performance. Some manufacturers of cabi nets or termi nal s sel l thei r own cabl e shi el d bondi ng accessori es. As l ong as they provi de 360 (ful l ci rcl e) bondi ng directly between the cabl e shi el ds and the surface of the l ocal RF Reference they wi l l gi ve good performance. But beware some of these attach the cabl es' shi el ds to metal bars that have appreci abl e i nductance, and these then usual l y connect to the l ocal RF Reference by a wi re or brai d strap addi ng even more i nductance. Fig 32 Mass shi el d termi nati on as shown i n the bottom-ri ght of Fi gure 31 i s a l ow-cost techni que rel yi ng on cl ampi ng a number of exposed shi el ds between conducti ve gaskets. t i s qui te easy to desi gn si mi l ar shi el d bondi ng methods i nto, say, the base of cabi net, usi ng si mpl e metal work and standard gasket types, as shown i n Fi gure 32. Thi s type of desi gn easi l y outperforms many of the propri etary shi el d-bondi ng accessori es offered by cabi net or termi nal manufacturers. Another method of mass-termi nati ng cabl e shi el ds i s shown i n Fi gure 33. Li ke Fi gure 32, thi s method can be easi l y adapted to sui t a vari ety of si tuati ons. Fi gure 34 shows two exampl es of termi nati ng cabl e shi el ds as they enter or exi t the RF Reference pl ane i n an i ndustri al cabi net that uses a backpl ate. The saddl ecl amp method can al so use P- cl i ps, whi ch do not provi de as good a shi el d termi nati on as a saddl ecl amp, whi ch i n turn i s not as good as a proper 360 shi el d bond. But P-cl i ps may be perfectl y acceptabl e where the EMC performance i s not requi red to be the hi ghest. Saddl ecl amps and P-cl i ps used to Fig 33 37 38 have to be obtai ned from pl umbi ng, pneumati c or hydraul i c component suppl i ers, maybe because the parts were too l ow-cost to be of i nterest to other manufacturers. However, there are now some EMC component suppl i ers who offer saddl ecl amps and P-cl i ps for cabl e shi el d bondi ng. Where shi el ded cabl es don't empl oy shi el ded connectors at thei r ends or at j uncti ons, and use unshi el ded connectors such as D N rai l termi nal s i nstead, thei r unshi el ded conductors degrade thei r EMC performance. Fi gure 35 shows how to use metal saddl e-cl amps (or P-cl i ps) to bond the cabl e shi el ds to RF Reference as cl ose as possi bl e to the unshi el ded termi nal s. The mi ni mum l ength of conductors shoul d be exposed, al l the same l ength, as short as possi bl e and routed as cl ose as possi bl e to the RF Reference. Fi gure 35 shows D N rai l mounted termi nal s, but they coul d i nstead be screw or sol der termi nal s or unshi el ded Fig 34 connectors on an el ectroni cs uni t. Where the unshi el ded connector i s mounted on an el ectroni c uni t, the best pl ace to bond the shi el d i s to the metal (or metal l i sed) body of the el ectroni cs uni t i tsel f, cl ose to the connector, but i f thi s i s not possi bl e the nearest l ocal RF Reference shoul d be used i nstead general l y the metal surface the el ectroni c uni t i s mounted on. A number of practi cal al ternati ves to saddl e-cl amps exi st, and the i nventi ve desi gner wi l l have no troubl e i n creati ng new constructi ons to ease assembl y of hi s cabi nets. The EMC performance of the unshi el ded connectors and exposed cabl e conductors wi l l not be very good, but thi s desi gn techni que ai ms to make i t as good as possi bl e wi thout changi ng to a shi el ded connector. The EM performance of a shi el d bonded wi th a P-cl i p wi l l general l y not be as good as one bonded to i ts RF Reference by a saddl ecl amp, but because the EM performance of the unshi el ded connectors Fig 35 i s so poor, usi ng P-cl i ps mi ght not make i t very much worse. Fi gure 36 shows an exampl e of an i ndustri al panel usi ng P- cl i ps to bond the shi el ds. Where shi el ded cabl es are routed to unshi el ded termi nal s or connectors that are not very cl ose to the RF Reference, the hei ght of the bracket i n Fi gure 33 can be i ncreased to ' extend' the RF Reference cl oser to the termi nal s and provi de a means for bondi ng the shi el ds nearby. Al though the metal bracket adds i nductance and so has a del eteri ous effect at hi gher frequenci es, i t i s orders of magni tude better than pi gtai l i ng the shi el ds (see bel ow). Si mi l arl y, a tal l bracket coul d be used to support a conducti ve gasket cl amp such as that i l l ustrated i n Fi gure 32, or a P-cl i p or saddl ecl amp as shown i n Fi gures 34 36, cl ose to the termi nal s or connectors. A tal l thi n metal bracket i s not very much better than a pi gtai l (see bel ow) at termi nati ng a shi el d such brackets shoul d be at least three ti mes as wi de as they are tal l , and have mul ti pl e metal -to- metal bonds to the RF Reference al l al ong thei r l ength, spaced no further apart than 1 00mm. f i t i s not practi cal to make a good RF bond at the end of a shi el ded cabl e usi ng the methods descri bed above (or si mi l ar techni ques), make a good RF bond to the RF Reference as cl ose as possi bl e to the end of the cabl e, then conti nue the shi el d after thi s bond ri ght up to the end of the cabl e i ncl udi ng the shi el ded connector backshel l where one can be fi tted. 3. 7. 5 Some addi ti onaI shi eI d bondi ng techni ques Al l cabl e shi el d bondi ng methods shoul d make a ti ght fi t al l around the peri phery of thei r cabl e's shi el d (but wi thout crushi ng the cabl e), and thi s ti ght fi t must not become l oose wi th age, wear and tear. t i s al ways best not to di sturb the l ay of a 39 40 Fig 36 cabl es' shi el d when 360 bondi ng to i t, but where l ower shi el di ng performance i s acceptabl e a l onger l ength of brai d shi el d can be 'scrunched up' to make a ti ght fi t i n a sl i ghtl y l oose saddl e-cl amp or connector backshel l shi el d cl amp. Wi th foi l shi el ded cabl es i t i s i mportant to make sure that the metal surface of the foi l makes a 360 contact wi th the connector backshel l or other shi el d bondi ng method. One si de of the foi l i s non-conducti ve pl asti c, and of course i s not sui tabl e for shi el d bondi ng. Where i t i s the i nternal surface of the foi l that i s conducti ve, the foi l wi l l need to be fol ded back, and wi th a spi ral -wrapped foi l cabl e thi s i s di ffi cul t to do neatl y and ensure a 360 bond. t i s al so i mportant for any drai n wi res i n the foi l -shi el ded cabl e to be bonded al ong wi th the metal l i sed foi l surface. Where a foi l shi el ded cabl e i s a l i ttl e l oose i n a shi el d cl amp, i t mi ght be possi bl e to wrap the drai n wi re over the exposed metal l i sed foi l surface a few ti mes to make a more rel i abl e cl amp.
t has been common practi ce for many years to use the drai n wi re as the sol e means of bondi ng foi l -shi el ded cabl es, but thi s creates a ' pi gtai l ' (see bel ow) and rui ns the EM performance of the cabl e. Because of the di ffi cul ti es associ ated wi th maki ng a 360 bond to metal l i sed-foi l shi el d materi al s, and the resul ti ng suscepti bi l i ty to vari ati ons i n workmanshi p, brai d rather than foi l -shi el ded cabl es are general l y preferred. 3. 7. 6 Pi gtai I s maki ng the best of a very poor EMC techni que t has been common practi ce for decades to bond cabl e shi el ds usi ng short l engths of twi sted brai d, or the drai n wi res i n foi l - shi el ded cabl es, or by sol deri ng a wi re to ei ther of these to reach a di stant shi el d bondi ng poi nt. These days, and for the future, thi s i s a terri bl y bad practi ce that effecti vel y rui ns the shi el di ng performance of the cabl e. The author has measured emi ssi ons from i ndustri al cabi nets that fai l ed the radi ated tests around 70MHz because a si ngl e cabl e from the vol t-free contacts of a PLC had a 25mm l ong pi gtai l to the RF Reference pl ane (the cabi net' s backpl ate i n that case). Repl aci ng that very short pi gtai l wi th a metal saddl ecl amp that pressed the shi el d agai nst the backpl ate reduced the emi ssi ons around 70MHz by over 20dB and the test was passed. Fi gure 37 shows the effect of a pi gtai l on the surface transfer i mpedance (Z T ) of a 25-way submi ni ature D-type connector. Z T
i s a measure of how wel l a cabl e or connector wi l l functi on as a shi el d l ower Z T at a gi ven frequency means hi gher shi el di ng at that frequency. A Z T of around 1 0 mi l l i ohms i s general l y adequate for average l evel s of shi el di ng, but hi gh l evel s of shi el di ng requi re 1 mi l l i ohm or l ess. Fi gure 37 shows that the shi el di ng effecti veness (SE) of a submi ni ature D- Type usi ng a pi gtai l for i ts cabl e shi el d bond i s average at frequenci es up to about 20kHz, but above that frequency progressi vel y reduces unti l by 1 MHz the SE i s unacceptabl y bad. Compari ng thi s wi th the proper 360 shi el d bondi ng shows that at 1 00MHz the 360 shi el d termi nati on i s 75dB better than the pi gtai l . t has been a common practi ce among the peopl e who wi re i ndustri al cabi nets to stri p about 300mm of shi el d from the conductors at the ends of shi el ded wi res and sol der a l ong l ength of green/yel l ow i nsul ated wi re to the brai d or drai n wi re. The (now unshi el ded) conductors are connected to the D N rai l or other termi nal , Fig 37 41 42 and the green/yel l ow wi re taken to an 'earth' termi nal that mi ght be up to 1 . 5 metres away (a substanti al copper bondi ng bar, usual l y cal l ed the 'mai n earthi ng bar' , i s a common choi ce). Al l the spare conductors are hi dden i n the pl asti c trunki ng, so that the conductors that exi t the trunki ng appear short neat and ti dy. Thi s practi ce shoul d no l onger be permi tted under any ci rcumstances, for the reasons descri bed bel ow. The pi gtai l used i n the tests summari sed i n Fi gure 37 was about 30mm l ong and that was l ong enough to compl etel y rui n the cabl e's Z T (and hence i ts SE) above 1 MHz. Longer pi gtai l s, even i f they are green/yel l ow i nsul ated or even brai d straps, wi l l have even worse SE. Al so, the bundl i ng of al l of the excess l engths of unshi el ded conductors i n the pl asti c trunki ng hel ps ensure l ots of crosstal k between the si gnal s on those wi res and other cabl es qui te possi bl y what the cabl e shi el di ng was supposed to be preventi ng i n the fi rst pl ace. Someti mes al l that i s needed i s an average l evel of SE up to about 1 00kHz, for i nstance to reduce the coupl i ng of 50/60Hz el ectri c and magneti c fi el ds from mai ns power cabl es and devi ces i nto sensi ti ve transducer si gnal s such as those from thermocoupl es, strai n gauges and the l i ke. Al so, vari abl e-speed motor dri ves and other swi tch-mode power converters rated at 1 kW or more create hi gh l evel s of el ectri c and magneti c fi el ds bel ow 1 MHz, so i n some cases shi el di ng may onl y be requi red for frequenci es bel ow 1 MHz. And where unshi el ded termi nal s such as D N rai l s are used, i t may prove di ffi cul t to achi eve good SE at frequenci es much above 1 MHz i n any case. So we need a method of usi ng pi gtai l s as effecti vel y as possi bl e, and Fi gure 38, usi ng the exampl e of D N-rai l termi nal s, shows thi s. A si mi l ar arrangement may be used at the unshi el ded termi nal s of el ectroni c uni ts. To get the best EM performance from a pi gtai l , the exposed conductors and the pi gtai l from a cabl e Fig 38 Fig 39 shoul d be as short as i s possi bl e, consi stent wi th the practi cal needs of assembl y (say, around 30 mm), and where possi bl e they shoul d be kept cl ose together by i nterl eavi ng the shi el d bondi ng termi nal s wi th the si gnal termi nal s as shown i n Fi gure 38. But remember that pi gtai l ed shi el ds are never goi ng to be much use for EMC above 1 MHz. When usi ng D N rai l termi nal s to connect pi gtai l s to the l ocal RF Reference, the metal D N rai l i tsel f shoul d be bonded metal -to-metal di rectl y to the Reference at both ends, and at other posi ti ons al ong i ts l ength, preferabl y every 1 00mm or l ess. Pl aci ng shi el d-bondi ng termi nal s (usual l y col oured green/yel l ow to i ndi cate they are bonded to the D N rai l and so are at ' earth' potenti al ) ei ther si de of the si gnal /power termi nal s al so hel ps provi de a l i ttl e shi el di ng for them, al though thi s cannot be expected to have any si gni fi cant effect above about 1 0MHz. On no account shoul d the green/yel l ow termi nal s used for bondi ng cabl e shi el ds ever be grouped together at one end of a D N rai l , for ' neatness'. The RF performance of pi gtai l s can be useful l y i mproved by usi ng two pi gtai l s for each cabl e. They shoul d be sol dered to ei ther si de of the cabl e, and connect to termi nal s ei ther si de of those used by the cabl e's conductors, as shown by Fi gure 39. 43 44 4 Bondi ng ci rcui ts and uni ts to the RF Reference As di scussed i n secti on 2, an RF Reference i s onl y useful for assi sti ng or i mprovi ng the EM performance of a ci rcui t i f i t i s l ocal meani ng cl oser than /1 0 at the hi ghest frequency to be control l ed (e. g. cl oser than 30mm, to control up to 1 00MHz) much cl oser spaci ng means better EMC. Secti on 3 descri bed how best to route and bond cabl es wi th respect to the RF Reference; thi s secti on di scusses techni ques for bondi ng el ectri cal /el ectroni c ci rcui ts, uni ts, modul es, products, etc. , to the Reference. Thi s secti on assumes the use of pl ai n unshi el ded l ow-cost metal cabi nets, and the techni ques descri bed are i ntended to get the best EMC performance from them wi thout addi ng much (i f anythi ng) to cost. Secti on 5 di scusses the good EMC engi neeri ng practi ces associ ated wi th the use of shi el ded cabi nets. 4. 1 Protective bonding (safety) conductors Fi gure 40 shows the connecti on of the i ncomi ng protecti ve conductor (often cal l ed the protecti ve or safety earth; the green or green/yel l ow conductor i n the mai ns suppl y cabl e) to the protecti ve earthi ng (PE) poi nt of a cabi net. Al though thi s bond i s pri mari l y a safety concern, i t hel ps to achi eve the best EMC performance from the cabi net i f the amount of protecti ve conductor exposed wi thi n the cabi net i s 1 50mm or l ess. Bondi ng i t to the outside of the cabi net woul d be the best for EMC, but for safety reasons i t needs to be l ocated cl ose to the mai ns termi nal s. ts bondi ng termi nal i s preferabl y wel ded to the cabi net si de or rear, al though i t coul d be screwed. For safety reasons the best type of wel ded stud i s one that penetrates the cabi net from the other si de, so that i f the wel d fai l s i t i s sti l l retai ned i n pl ace and doesn't j ust pul l free. Fig 40 For i ndustri al motor control cabi nets, the safety standard EN 60204-1 requi res no more than one protecti ve conductor per termi nal . Thi s requi rement appears to be wi del y i gnored, for exampl e wi th cabi nets bei ng wi red usi ng several ri ng tags ' starred' to a si ngl e stud, or wi th two or more green/yel l ow wi res connected i nto i ndi vi dual D N rai l 'earth' termi nal s, maki ng many i ndustri al motor control cabi nets non-compl i ant wi th safety regul ati ons. A si ngl e 'earth' protecti ve bondi ng wi re per termi nal i s good advi ce for any i ndustri al cabi net, so that i f a panel (say) i s removed and i ts protecti ve bondi ng wi re di sconnected, thi s does not remove the protecti ve bondi ng for anythi ng el se. So to bond the backpl ate to the protecti ve conductor recommend a separate wel ded or screwed stud termi nal near to the i ncomi ng ' safety earth' termi nal (pl us a suffi ci ent cross-secti onal area of metal between the two, see secti on 4. 6), as shown i n Fi gure 22. Many control panel bui l ders i nstead use a Fig 41 sol i d metal termi nal bl ock (usual l y cal l ed the ' mai n earthi ng bar') bonded rel i abl y to the cabi net wal l as thei r ' safety earth' star poi nt, as shown i n Fi gure 41 . When usi ng such bars, a stud that bonds the bar to the cabi net's metal wal l shoul d be l ocated i mmedi atel y adj acent to the pl ace where the i ncomi ng protecti ve conductor i s connected. t i s al ways best to seam-wel d al l the parts of a cabi net' s structure together, or el se spot-wel d them at mul ti pl e poi nts al ong thei r j oi ns, or fi x them together wi th mul ti pl e screws or ri vets that provi de metal -to-metal bondi ng as di scussed i n secti on 2 to hel p create the best possi bl e RF Reference for the cabi net. But where i t i s necessary to bond them usi ng wi res, straps or brai ds, Fi gure 41 shows the basi c pri nci pl es for a cabinet in which safety and RF-bonding can be safely combined (see secti on 4. 6). Fi gures 22 and 41 are exampl es of protecti ve bondi ng i n typi cal i ndustri al cabi nets fi tted wi th a backpl ate, and 45 46 assume that the cross-secti onal -area (CSA) of the cabi net wal l i s suffi ci ent to handl e the earth-faul t currents associ ated wi th the type of mai ns suppl y and i ncomi ng mai ns cabl es, accordi ng to the requi rements of the rel evant safety standard (typi cal l y ei ther EN/ EC 61 01 0-1 or EN/ EC 60204-1 , maki ng sure to use the current edi ti on), taki ng the materi al (e. g. mi l d steel ) and maxi mum cabi net temperature i nto account. Safety i s more i mportant than EMC, so must never be compromi sed for EMC reasons. Gi ven that the above CSA requi rement i s met, Fi gure 41 shows that doors, removabl e panel s (and anythi ng el se associ ated wi th the cabi net structure that requi res 'safety earthi ng' ) shoul d make thei r protecti ve bondi ng to thei r nearest cabi net wal l s usi ng thei r own studs or termi nal s wi th the shortest and wi dest practi cabl e conductors, e. g. metal straps or brai ds. n general , the shorter the l ength of the protecti ve bondi ng wi re, strap or brai d, the hi gher the frequency at whi ch a metal cabi net provi des some shi el di ng benefi ts. Al so i n general , better EMC wi l l be achi eved by havi ng more than one short protecti ve bondi ng wi re, strap or brai d to the part concerned spreadi ng them out as uni forml y as practi cal al ong the l ength, hei ght or wi dth concerned, preferabl y spaced 1 50mm or l ess apart, as shown i n Fi gures 4 and 5. Fi gure 42 shows the detai l s of connecti ng a protecti ve bondi ng conductor to the RF Reference (the backpl ate i n thi s exampl e) and shoul d be compared wi th Fi gures 6 and 7. At frequenci es for whi ch the l ongest cabi net di mensi on exceeds one-twenti eth of the wavel ength (dependi ng on the SE requi red for the cabi net) such straps or brai ds provi de few EMC benefi ts, no matter how short they are, and a shi el ded cabi net usi ng conducti ve gaskets around doors and removabl e panel s may be requi red i nstead (see secti on 5). To hel p create a l arger l ocal RF Reference and get the benefi t of whatever shi el di ng i s avai l abl e from a basi c metal cabi net, the Fig 42 backpl ate shoul d be RF-bonded metal -to- metal to the metal cabi net at every one of i ts fi xi ng poi nts. Some cabi nets are made usi ng pl asti c backpl ate mounti ng brackets, and wherever the regul ar fi xi ngs don' t provi de the requi red RF-bondi ng methods (see secti on 2) short wi de straps or brai ds, or metal brackets, shoul d be fi tted between the backpl ate and the cabi net to provi de RF bonds. Si mi l ar RF bonds shoul d connect the backpl ate to the cabi net wal l wherever cabl es enter or exi t the backpl ate. 4. 2 RF-bonding insulated electrical/electronic items to the RFReference Some i nsul ated i tems of equi pment, such as ' Doubl e- nsul ated' types (accordi ng to the safety standards), requi re no safety earth connecti ons for thei r mai ns suppl y but even so they mi ght have a ' functi onal earth' that needs to be connected to the cabi net's RF Reference. nsul ated i tems can onl y be connected to the RF Reference usi ng wi res, straps or brai ds, as shown by Fi gure 42, and thi s method usi ng the shortest practi cal wi res, straps or brai ds shoul d be used to connect any protecti ve ('earth' ) conductors or any functi onal earths to the l ocal RF Reference. Fi gure 43 i l l ustrates thi s practi ce. t shoul d be understood that j ust a few centi metres of conductor (whether round wi re, wi de metal strap or brai d) can be compl etel y i neffecti ve (or even counter- producti ve) at frequenci es above a few MHz, as shown by some of the graphs i n [20], so al though thi s method attempts to get the best RF performance from an i nsul ated i tem, i t cannot be rel i ed upon to achi eve good EMC where the i tem i tsel f has a poor EMC performance (see [1 4] ). Shi el ded cabl es enteri ng/exi ti ng such i tems shoul d have thei r shi el ds bonded to the l ocal RF Reference usi ng one of the methods descri bed i n secti on 3. 7 unl ess speci fi cal l y prohi bi ted i n the suppl i er's i nstructi ons but onl y use pi gtai l s when there i s no practi cal al ternati ve, and understand that the shi el d performance above 1 MHz wi l l then be very poor. Fig 43 47 48 Where there are prohi bi ti ons about cabl e shi el d bondi ng, al ways check whether such suppl i ers real l y understand i n EMC terms why they are maki ng them, i n case they are j ust bl i ndl y fol l owi ng the tradi ti onal and l ong-outdated practi ce of tryi ng to avoi d 'ground l oops', see secti on 3. 7. 1 (and 2. 6. 8 i n the 2006 versi on of [28] ). 4. 3 RF-bonding metal-bodied electrical/electronic items to the RFReference Thi s secti on assumes the metal -bodi ed i tems have hi ghl y conducti ve surface pl ati ngs (e. g. bri ght or dul l ti n, al ochromed al umi ni um, etc. ). Where they are pai nted or anodi sed or otherwi se i nsul ated, i t i s assumed that the i nsul ati ng coati ngs are removed, and a hi ghl y-conducti ve corrosi on-resi stant pai nt appl i ed (e. g. pai nt hi ghl y l oaded wi th si l ver, zi nc or al umi ni um speci fi cal l y i ntended for creati ng conducti ve surfaces). f none of the above appl i es or i f bondi ng to an external metal chassi s i s speci fi cal l y prohi bi ted i n the suppl i er's i nstructi ons treat the i tem as an i nsul ated i tem, see secti on 4. 2 and expect si mi l ar EMC. Where suppl i ers prohi bi t bondi ng the body of a uni t, al ways check whether they understand why i n EMC terms, i n case they are j ust bl i ndl y fol l owi ng the tradi ti onal and outdated practi ce of tryi ng to avoi d 'ground l oops' (see secti on 3. 7. 1 , and 2. 6. 8 i n the 2006 versi on of [28] ). tems that empl oy conducti vel y pl ated metal bodi es often have better EMC performance than ones wi th pai nted, anodi sed or i nsul ati ng bodi es. They al so provi de more opportuni ti es for i mprovi ng thei r EMC by bondi ng thei r metal bodi es to the RF Reference. So i f there i s a choi ce of el ectroni c uni ts for a parti cul ar functi on, and there appears to be nothi ng to choose between the EMC performances offered by thei r manufacturers, i t shoul d be best to choose the one wi th the conducti vel y pl ated metal body. Di rect metal surface-to-metal surface el ectri cal bonds gi ve the best performance at radi o frequenci es above 1 MHz or so, and shoul d be used to bond al l the metal fi xi ngs on any el ectroni c uni ts to thei r l ocal RF Reference as shown by Fi gures 44 and 45 (al so see Fi gure 7 for compari son).
Where the fi xi ngs for a metal -bodi ed uni t are further apart than 1 00mm or so, addi ng more RF bonds between i ts metal case and the RF Reference wi l l general l y i mprove EMC. deal l y, l ow-profi l e metal brackets (wi th hi ghl y conducti ve surface pl ati ng) woul d be screwed between the i tem' s metal body and the RF Reference but thi s i s not usual l y acceptabl e because i t coul d damage the i tem or i nval i date i ts warranty. Acceptabl e al ternati ves i ncl ude maki ng addi ti onal RF bonds wi th pi eces of conducti ve gasket (see Fi gure 8) or metal spri ng fi ngers (see Fi gure 9). Fi gure 46 shows an exampl e of a 55kW vari abl e-speed motor dri ve i nstal l ati on i n a cabi net. The mai ns fi l ter, DC power suppl y and vari abl e speed dri ve are each i n ti n- pl ated boxes and each i s RF-bonded to the ti n-pl ated backpl ate usi ng the method shown i n Fi gures 44 and 45. Compare thi s assembl y wi th Fi gure 23, and noti ce al so that the motor cabl e' s shi el d i s RF-bonded wi th a saddl ecl amp type of fi xi ng to a wi de bracket extendi ng from the dri ve' s chassi s (i ts RF Reference). Modul ar uni ts such as Programmabl e Logi c Control l ers (PLCs) consi st of a basi c chassi s (or some other name) i nto whi ch the modul es are pl ugged. Thi s chassi s shoul d be treated as descri bed i n secti ons 4. 2 or 4. 3, dependi ng on whether i ts body i s i nsul ated or not. Fig 44 Fig 45 Fig 46 49 50 4. 4 RF-bonding PCBs to the RF Reference deal l y, PCBs woul d be contai ned wi thi n conducti vel y-pl ated boxes and be treated as descri bed i n secti on 4. 3. But someti mes cabi nets use unencl osed PCBs, especi al l y i f they have been custom-desi gned for the cabi net manufacturer. Fi gure 47 shows the general pri nci pl es of bondi ng a PCB's own RF Reference (usual l y i ts 0V pl ane) to the cabi net's RF Reference. The bondi ng poi nts shoul d be spread over the whol e PCB area, and wi thi n i t too, i deal l y spaced l ess than /1 0 apart at the hi ghest frequency to be control l ed. Of course, thi s PCB bondi ng shoul d not be done i f speci fi cal l y forbi dden by the PCB suppl i er but you shoul d al ways check whether they real l y understand i n EMC terms why they are maki ng thi s prohi bi ti on, because they mi ght j ust be bl i ndl y fol l owi ng the tradi ti onal and l ong-outdated practi ce of tryi ng to avoi d 'ground l oops' (see secti on 3. 7. 1 , and secti on 2. 6. 8 i n the 2006 versi on of [28] ). The PCB-to-Reference bondi ng poi nts do not have to be fi xi ng screws, they coul d use the modi fi ed cl i p-on mounti ng posts shown i n Fi gure 48, or spri ng fi ngers such as those i n Fi gure 9. Some manufacturers (e. g. Ki tagawa) make spri ng fi ngers speci fi cal l y i ntended for surface mounti ng and sol deri ng on PCBs, for maki ng addi ti onal RF bonds to thei r l ocal RF Reference. 4. 5 Capacitive and hybrid RF- bonding Someti mes there are very good reasons why i t i s undesi rabl e to make di rect metal - to-metal connecti ons between an el ectri cal /el ectroni c i tem (i ncl udi ng PCBs) and i ts l ocal RF Reference. n such cases RF-bondi ng can sti l l be achi eved usi ng capaci tors i n seri es wi th each bond. Where there i s a si ngl e di rect bond, and the other bonds are capaci ti ve, thi s i s known as hybri d RF-bondi ng. Capaci ti ve and hybri d bondi ng was descri bed i n secti on 3. 7. 2, for bondi ng Fig 47 Fig 48 cabl e shi el ds to the RF Reference, and exactl y the same i ssues and capaci tor sel ecti on i ssues appl y when usi ng these techni ques for RF-bondi ng el ectri cal /el ectroni c i tems i ncl udi ng PCBs. When desi gni ng a PCB that requi res capaci ti ve or hybri d bondi ng, for an i ndustri al cabi net, the seri es capaci tors shoul d be mounted on the board i n seri es wi th the bondi ng poi nts wi th thei r traces and pads desi gned to mi ni mi se thei r i nductances, to hel p achi eve the best EMC performance. 4. 6 Combining safety bonding with RF-bonding n many i ndustri al cabi nets i t i s often easy to combi ne the safety and RF-bondi ng structures together, as shown i n secti ons 4. 1 , 4. 2 and 4. 3. Thi s saves ti me, i mproves EM performance, and al so removes bundl es of green/yel l ow wi res from the pl asti c trunki ng maki ng more space and easi ng the wi ri ng of the cabi net. 51 52 However, where ' earth-faul ts' coul d resul t i n very heavy currents to a protecti vel y bonded part, the CSA of the cabi net structure mi ght not be suffi ci ent to al l ow i ts use i n the protecti ve bondi ng system. Al so, some safety i nspectors mi ght be uncomfortabl e i f they cannot see green/yel l ow wi res, straps or brai ds to a 'mai n earthi ng bar' from most/al l of the protecti vel y bonded structural parts and i tems of equi pment. Where the RF Reference cannot be used as the protecti ve bondi ng system, both the RF Reference system and the ' tradi ti onal ' protecti ve conductor system shoul d exi st i n paral l el , creati ng a l ot of 'ground l oops'. n such cases we do not care how l ong the protecti ve bondi ng conductors are, as l ong as RF-bondi ng system uses the shortest wi res, straps and brai ds (preferabl y di rect metal -to-metal bonds where practi cal ). Of course, thi s creates a great many ' ground l oops', but as was shown i n secti on 2. 6. 8 of the 2006 versi on of [28], ground l oops are general l y a good thi ng for si gnal s, EMC and safety provi di ng the el ectroni cs i s competentl y desi gned and there are a l arge number of smal l l oops and not j ust one or two l arge ones. 4. 7 Choosing filters and bonding them to the RFReference 4. 7. 1 Choosi ng fi I ters As Fi gure 21 showed, there are a great many types of mai ns fi l ter, and there are al so a great many types of si gnal fi l ter so i t i s i mportant to choose the ri ght ones for your appl i cati ons. When choosi ng a mai ns fi l ter, i t i s safest to assume that i ts attenuati on at any frequency i s no better than the worst-case deri ved from al l of i ts matched 509/509, and i ts mi smatched 1 009/0. 1 9, and 0. 1 9/1 009 performance data for both common-mode and di fferenti al -mode (known i nstead as 'asymmetri cal ' and 'symmetri cal ' by fi l ter manufacturers). Merel y usi ng the 509/509 attenuati on curves can resul t i n amplifying an unwanted noi se frequency i nstead of attenuati ng i t. Good fi l ter manufacturers wi l l provi de al l the above data as graphs coveri ng the whol e frequency range of i nterest, i ncl udi ng both the conducted range (down to 1 50kHz or l ess) and the radi ated frequency range (e. g. up to 1 GHz or more). For more on these and other fi l ter sel ecti on i ssues refer to [29] (especi al l y i ts secti ons 3. 2. 8, 3. 2. 9 and 3. 3. 3) and al so to [32]. t i s al so worth noti ng that the best fi l ters for EMC have seaml ess metal bodi es fi tted wi th fl anges or other means of di rectl y bondi ng them metal -to-metal to a l ocal RF Reference at l east at two poi nts. 4. 7. 2 Bondi ng fi I ters to the RF Reference Some fi l ters rel y sol el y on ferri tes and have no need for any connecti on to the RF Reference. Thi s type i ncl udes the cabl e chokes shown i n Fi gure 20, and they are especi al l y useful where a good qual i ty RF Reference i s not avai l abl e at the frequency to be control l ed. However, most types of fi l ters and al l medi um or hi gh-performance fi l ters contai n capaci tors, and i t i s vi tal for thei r EMC performance that thei r hi ghl y conducti ve metal bodi es are bonded metal -to-metal to a l ocal RF Reference that has l ower i mpedance than thei r capaci tors at the frequenci es of concern. The bonds must be made at least at al l of thei r fi xi ng poi nts, as shown i n Fi gures 44 and 45 and descri bed i n secti on 4. 3. Many types of connectors are avai l abl e wi th bui l t-i n fi l ters, some of whi ch are si mpl e ferri tes needi ng no Reference connecti on, some are si mpl e capaci tors, and some (more costl y) types use Tee or Pi fi l ter pi ns. To hel p protect the el ectroni cs i n a product from external EM di sturbances, the fi l ters fi tted to external cabl es shoul d be fi tted at the poi nt where the cabl e fi rst crosses the boundary of the l ocal RF Reference. But to hel p reduce i nterference inside a cabi net caused by the emi ssi ons from a noi sy el ectri cal /el ectroni c uni t, fi l ters shoul d be fi tted to that uni t' s l ocal RF Reference as cl ose as possi bl e to i t. Where fi l ters must be used for both the above purposes (typi cal of the mai ns and motor dri ve cabl es associ ated wi th a vari abl e-speed AC motor dri ve), thei r l ocati on i n a cabi net can be a di ffi cul t compromi se so i t i s general l y best to l ocate the noi sy uni t cl ose to the edge of the RF Reference, and i n the appropri ate ori entati on, so that the fi l ter can be mounted very cl ose to the noi sy uni t and al so cl ose to the edge of the Reference where the fi l tered cabl e enters or exi ts. Fi l ters fi tted to the i ncomi ng mai ns suppl y shoul d be pl aced so that the l ength of external mai ns cabl e that enters a cabi net i s mi ni mi sed, preferabl y l ess than 1 50mm for i ndustri al cabi nets and kept cl ose to the RF Reference at al l ti mes. Smal l er products shoul d use a bul khead mounted mai ns fi l ter i f at al l possi bl e, such as the popul ar EC320 pl ug-i nl et styl e, so that no external mai ns wi res penetrate the cabi net at al l . Mai ns fi l ters fi tted pri or to on/off swi tches (or door i sol ators i n i ndustri al cabi nets) wi l l stay l i ve even when the power i s swi tched off at the product, so touch protecti on and appropri ate safety warni ngs for thei r termi nal s must be provi ded (be sure to meet al l the requi rements i n the rel evant safety standards). Fi l ter i nput and output wi res must never come anywhere near each other, as they are al ways at l east one cabl e Cl ass apart (see secti on 3. 3). Cascaded mai ns fi l ters can i nteract and make the overal l EMC performance worse than that of each fi l ter on i ts own, as di scussed i n the 2006 versi on of [29], so i f i t i s necessary to cascade fi l ters on a si ngl e cabl e the addi ti onal fi l ter mi ght have to have more stages, and be l arger wi th a hi gher speci fi cati on, than mi ght seem necessary. 4. 8 A single connector panel is best t i s best to provi de a si ngl e connector panel for each cabi net, so that al l external cabl es enter or exi t the RF Reference i n one pl ace at one of i ts si des or edges. Thi s i s so that (i n conj uncti on wi th the other techni ques descri bed here) the CM RF currents (someti mes cal l ed surface currents) that can fl ow i n l ong external cabl es, especi al l y i n some el ectri cal l y noi sy i ndustri al envi ronments or duri ng thunderstorms, wi l l fl ow from cabl e to cabl e vi a the connector panel or edge of the RF Reference through the shi el d termi nati ons and fi l ters mounted i n that area. The si ngl e connector panel techni que gi ves best resul ts when al l of the external cabl es are ei ther fi l tered or shi el ded at the poi nt where they pass through the connector panel , usi ng the techni ques descri bed i n secti on 5. 2 as i f the connector panel was part of a shi el ded cabi net. Thi s i s because the external cabl es' ci rcul ati ng CM currents wi l l then prefer not to fl ow through the rest of the cabi net or RF Reference, or through the i nternal el ectroni cs and thei r cabl es. Thi s can si gni fi cantl y i mprove i mmuni ty to external EM threats, and i t can al so hel p contai n CM currents generated inside the cabi net, si gni fi cantl y i mprovi ng emi ssi ons. 4. 9 RF-bonding VGA display panels to the RFReference VGA LCD panel di spl ays are often used i n modern i ndustri al cabi nets, but can be a si gni fi cant source of emi ssi ons. f they have touchscreens, they can be a probl em for i mmuni ty. The usual techni que i s to purchase types that have a conti nuous metal back cover that i s el ectri cal l y bonded al l around i ts peri phery to thei r metal 'pi cture frame' surround. Then the VGA panel 's metal surround i s RF-bonded metal -to-metal (or wi th a conducti ve gasket) to the door or wal l of the metal cabi net, whi ch shoul d be a part of the RF Reference of the cabi net. Thi s method can 53 54 be seen as a vari ant of the Cl ean Box / Di rty Box method descri bed i n secti on 5. 3. 2. f the VGA panel s that have compl ete metal backs are not sui tabl e, one shoul d be made from thi n metal (or copper tape) and RF-bonded al l around i ts peri phery to the panel 's metal 'pi cture frame' surround, usi ng conducti ve adhesi ve or conducti ve gaskets. Then the new back or the LCD panel ' s metal surround i s bonded al l around i ts peri phery as descri bed above. f the above i s not suffi ci ent, a shi el ded wi ndow wi l l be requi red see secti on 5. 3. 2. 5 Usi ng shi eI ded cabi nets 5. 1 Introduction Secti ons 2, 3 and 4 above assumed that a metal chassi s, backpl ate, racki ng system, or cabi net was avai l abl e and was used as a l ocal RF Reference, but di d not assume that the cabi net was desi gned speci fi cal l y to provi de any speci al shi el di ng performance. However, i n some appl i cati ons shi el ded cabi nets are needed because the EM envi ronment at the i ntended operati onal l ocati on coul d be too severe for the el ectroni c uni ts used i nsi de the cabi net, or because the external EM envi ronment needs to be protected from the EM emi ssi ons from the el ectri cal or el ectroni c uni ts i n the cabi net. Cabi net shi el di ng requi res metal cabi nets, or pl asti c cabi nets wi th hi ghl y conducti ve metal -coated surfaces (pl asti c wi th conducti ve fi l l ers i s very di ffi cul t to use effecti vel y). Very careful attenti on to detai l i n desi gn and assembl y i s needed i f the shi el di ng provi ded by the cabi net i s not to be rui ned. There are two i ssues: Shi el di ng and/or fi l teri ng of al l conductors enteri ng or exi ti ng the cabi net, at the poi nt of penetrati on of the cabi net wal l , see secti on 5. 2 Control of al l apertures, i ncl udi ng at doors, removabl e panel s, di spl ays and venti l ati on, see secti on 5. 3
The detai l s associ ated wi th the desi gn and
assembl y of shi el ded encl osures of al l si zes are covered i n [30] . 5. 2 Shielding and/or filtering of all conductors entering or exiting the cabinet Shi el ded cabi nets compl ete wi th shi el ded wi ndows and venti l ati on wi th excel l ent EMC performance can be purchased from a number of suppl i ers, and can easily be completely ruined by cutti ng apertures for door-mounted uni ts, dri l l i ng hol es, poor fi l ter mounti ng, poor cabl e shi el di ng or shi el d bondi ng, or l eavi ng doors open. Thi s secti on covers cabl e penetrati ons, whi l st secti on 5. 3 covers apertures and gaps. Al l conductors enteri ng or exi ti ng a cabi net must be ei ther shi el ded and/or fi l tered wi th the shi el d and/or fi l ter RF- bonded to the cabi net wal l (or fl oor, rear, etc. ) at the exact poi nt of penetrati on of the cabi net wal l . There are no excepti ons to thi s rul e for any conductors of any type, whether they are fi bre-opti c draw wi res; metal pi pes or fl exi bl e pi pes wi th metal strengtheni ng for hydraul i cs or pneumati cs; cabl e armour, etc. The general pri nci pl es of control l i ng cabl e penetrati ons are i l l ustrated i n Fi gure 49. t i s common to fi nd the SE of a cabi net compl etel y rui ned by somethi ng as tri vi al as a mouse cabl e penetrati ng one of i ts si des. The mouse si gnal s themsel ves are not general l y a cause of emi ssi ons the probl em i s that the mouse cabl e conductors are acci dental antennas j ust l i ke any other conductor (see secti on 1 . 4 and Fi gure 3). They pi ck-up EM noi ses on ei ther si de of the cabi net wal l and re- radi ate them on the other si des thereby defeati ng the expensi ve shi el di ng of the cabi net. 5. 2. 1 Shi eI ded cabI es enteri ng/exi ti ng a shi eI ded cabi net Fi gure 50 shows the good EMC engi neeri ng practi ces requi red to be used when shi el ded cabl es penetrate a shi el ded cabi net wal l . Fig 49 55 56 Fig 50 The metal bodi es of the chassi s-mounted connectors or gl ands must make mul ti pl e metal -to-metal contacts wi th the wal l of the shi el ded cabi net, at the poi nt where they pass through i t to connect to the cabl e- mounted connectors. The RF-bondi ng techni ques for the cabl e shi el ds i n the cabl e-mounted connectors were descri bed i n 3. 7. 4 and Fi gures 27-30. t i s very i mportant to ensure that the cabi net has a hi ghl y conducti ve pl ati ng that i s sui tabl e for the physi cal envi ronment and l i fecycl e of the cabi net (see secti on 6) at l east i n the areas where the connectors or gl ands are to be i nstal l ed. Ci rcul ar connectors and gl ands general l y make a good 360 el ectri cal bond al l around thei r cabi net aperture. However, rectangul ar connectors general l y onl y achi eve rel i abl e RF bonds at thei r mounti ng poi nts (two, for a D-Type) and where hi gh val ues of SE are requi red, or where frequenci es above 1 00MHz are to be control l ed, they shoul d be fi tted wi th a conducti ve gasket duri ng assembl y to achi eve a good RF-bond al l around thei r peri meters. A number of manufacturers make EMC gaskets for di fferent types of chassi s-mounted connector, such as those shown i n Fi gure 51 . Saddl e-cl amps, P-cl i ps, pi gtai l s and any other shi el d bondi ng method that cannot achi eve a 360 el ectri cal bond around the aperture requi red for the connector or gl and i n the cabi net wal l , must not be used to bond cabl e shi el ds as they enter/exi t a shi el ded cabi net. The onl y excepti ons to thi s mi ght perhaps be i n very speci al ci rcumstances where the shi el ded cabi net i s not requi red to shi el d agai nst frequenci es above, say, 1 00kHz (and even then, pi gtai l s are not recommended). 5. 2. 2 When good shi eI di ng practi ces contradi ct suppI i er's i nstructi ons t someti mes happens that two i tems of equi pment are i nstal l ed i n separate shi el ded cabi nets, and need to be Fig 51 i nterconnected by a shi el ded cabl e but one of the equi pment i s suppl i ed wi th EMC i nstructi ons that state that i ts cabl e shi el d must onl y be connected at one end (usual l y at that i tem of equi pment, and usual l y to a screw-termi nal or connector pi n). Leavi ng asi de the i ssues of whether the suppl i er had used good EMC desi gn, or was si mpl y regurgi tati ng ' tradi ti onal ' i nstructi ons that are now decades obsol ete unl ess the suppl i er can be persuaded to al ter hi s EMC i nstructi ons they shoul d be fol l owed or el se they wi l l di scl ai m al l responsi bi l i ty for i nterference.
The probl em i s that unl ess the shi el d i s RF-bonded to the wal l s of both shi el ded cabi nets, fol l owi ng the equi pment suppl i er' s i nstructi ons to bond the shi el d at onl y one end wi l l fatal l y compromi se thei r shi el di ng performance. Fi gure 52 i l l ustrates one sol uti on use a doubl e i nsul ated shi el d cabl e and RF-bond the outermost i nsul ated shi el d to both the shi el ded cabi nets i n the approved manner 57 58 (see secti on 5. 2. 1 ). The i nsul ated i nner shi el d can then be termi nated i n accordance wi th the suppl i er's EMC i nstructi ons. Where both equi pment suppl i ers i nsi st that the cabl e shi el d must onl y be bonded at one end, and they don't agree on whi ch end, the method of Fi gure 52 wi l l preserve the SE of the cabi nets, but cannot resol ve the probl em of whi ch end to bond the i nner shi el d. 5. 2. 3 Unshi eI ded cabI es enteri ng/exi ti ng a shi eI ded cabi net Every unshi el ded cabl e that enters or exi ts a shi el ded cabi net must be fi tted wi th a fi l ter that provi des a si mi l ar l evel of RF attenuati on versus frequency as the SE requi red for the cabi net. Thi s fi l ter must be mounted at the poi nt where the cabl e penetrates the metal (or metal l i sed) wal l of the cabi net, and must make mul ti pl e metal - to-metal el ectri cal bonds at that poi nt. Fig 52 'Through-bul khead' fi l ters and fi l tered connectors cause the l east degradati on of a cabi net's SE, as l ong as thei r metal bodi es make mul ti pl e metal -to-metal bonds to the cabi net wal l al l around the peri meter of thei r cut-outs, as i l l ustrated i n Fi gure 53. Many cabi nets have had thei r SE rui ned by a l ack of provi si on of metal - to-metal bondi ng of EC 320 appl i ance- i nl et fi l ters. Where modest l evel s of SE are requi red from the cabi net, i t may be enough to rel y on the bonds provi ded by the fi xi ngs of the fi l ter, but for good SE especi al l y at frequenci es above 1 00MHz a conducti ve gasket may be requi red to bond the fi l ter's metal body to the cabi net' s metal surface al l around i ts peri phery. Note that Fi gure 53 does not show the protecti ve cover that woul d be requi red for the safety of the hi gh-performance feedthrough fi l ter. An al ternati ve to i s to use a l ower cost chassi s-mounted fi l ter. These cannot be Fig 53 i nstal l ed so that they penetrate the wal l of a shi el ded cabi net, so when used on cabl es that enter or exi t a cabi net they degrade i ts SE. However, they can be used wi th what i s usual l y cal l ed the Cl ean Box / Di rty Box method, as shown i n Fi gure 53. Attenti on to detai l i s needed to achi eve hi gh val ues of SE wi th thi s techni que, especi al l y mi ni mi si ng any gaps i n the RF-bondi ng around the edges of the Di rty Box, and reduci ng the coupl i ng of hi gh frequenci es between the i nput and output cabl es i nsi de the Di rty Box by keepi ng them short and far away from each other. There wi l l sti l l be some coupl i ng between the i nput and output cabl es i nsi de the Di rty Box, especi al l y at frequenci es above 1 00MHz, so a hi gh- frequency through-bul khead fi l ter may sti l l need to be fi tted to one of the cabl es. Often, thi s coupl i ng can be reduced suffi ci entl y by addi ng one or more soft ferri te cabl e suppressers (CM chokes, see Fi gure 20) to one or both cabl es cl ose to the Di rty Box. 'Room fi l ters' are chassi s-mounti ng fi l ters speci fi cal l y desi gned for penetrati ng the wal l s of shi el ded cabi nets wi thout compromi si ng thei r SE. They i ncorporate compartmented shi el ds for thei r i nput and output termi nal s (effecti vel y two separate Di rty Boxes), and thei r fi l tered outputs enter the shi el ded room through gal vani sed condui t wi th 360 bondi ng to the wal l of the room, as shown i n Fi gure 54. Room fi l ters are avai l abl e from a number of manufacturers, sui tabl e for every type of si gnal or el ectri cal power. They are general l y desi gned to achi eve attenuati ons of at l east 80dB from 1 00kHz to at l east 1 GHz, and types are avai l abl e that go down to kHz and/or up to 40GHz and meet mi l i tary speci fi cati ons. Room fi l ters can al so be fi tted to i ndustri al cabi nets, and are general l y requi red when a shi el ded cabi net needs to have the hi ghest EMC performance. 59 60 5. 2. 4 A segregated cabi net Fi gure 55 shows a shi el ded cabi net that has been segregated i nto ' cl ean' and 'di rty' vol umes, usi ng the methods descri bed above for termi nati ng cabl e shi el ds and screw-termi nal fi l ters. nstead of di vi di ng a cabi net, some desi gners bri ng thei r cabl es i nto a smal l Di rty cabi net that i s bol ted (preferabl y seam- wel ded) to the si de of the Cl ean cabi net. 5. 2. 5 A si ngI e connector paneI i s sti I I the best The benefi ts descri bed for a si ngl e connector panel i n secti on 4. 8 al so appl y i n the case of a shi el ded cabi net, because the i nternal and external ci rcul ati ng CM currents ('surface currents' ) do not have to cross any j oi nts or gaps i n metal surfaces, and thi s hel ps keep the i nternal currents i nsi de, and the external currents outsi de j ust what we want for good emi ssi ons and i mmuni ty respecti vel y. Fig 54 Fig 55 5. 3 Controlling apertures and gaps in shielded cabinets 5. 3. 1 Introducti on Shi el ded cabi nets compl ete wi th shi el ded wi ndows and venti l ati on wi th excel l ent EMC performance can be purchased from a number of suppl i ers, and can easily be completely ruined by cutti ng apertures for door-mounted uni ts, dri l l i ng hol es, poor fi l ter mounti ng, poor cabl e shi el di ng or shi el d bondi ng, or l eavi ng doors open. Secti on 5. 2 covered the shi el di ng and/or fi l teri ng of all cabl e penetrati ons, whi l st thi s secti on covers apertures and gaps. Apertures and gaps i n a shi el ded cabi net general l y act as 'acci dental sl ot antennas' as shown by Fi gure 56, much as conductors act as acci dental antennas as shown by Fi gure 3. t does not matter how narrow a gap i s (even as thi n as a l ayer of pai nt or anodi si ng), or even i f i t i s shaped l i ke a l abyri nth and there i s no l i ne of si ght Fig 56 61 62 through i t i t sti l l l eaks RF energy and degrades the cabi net's SE. Havi ng acci dental sl ot antennas i n the wal l of a shi el ded cabi net compromi ses i ts SE. [30] descri bes good shi el di ng practi ces for cabi nets, and goes i nto detai l on apertures and other i ssues that wi l l not be repeated here.
5. 3. 2 Di spI ays and controI s Fi gure 57 shows how to use the Cl ean Box / Di rty Box method where an i tem (such as a di spl ay, meter or control ) has to penetrate the wal l of a shi el ded cabi net. t i s i mportant to note that the i tem i n the Di rty Box does not benefi t from the shi el di ng of the cabi net, so must have emi ssi ons and i mmuni ty performance that i s adequate gi ven the external EM envi ronment. Much better EMC i s achi eved by pl aci ng di spl ays behi nd shi el ded wi ndows, Fig 57 especi al l y wi th VGA-type LCD panel s, and thi s can be i mportant where a good SE i s needed or hi gher frequenci es are to be control l ed. Shi el ded wi ndows use very fi ne bl ackened metal meshes, or conducti vel y-pl ated l ayers (often i ndi um ti n oxi de, known as TO), sandwi ched between two cl ear pl asti c panel s, and the metal mesh or pl ated l ayer must be metal -to-metal bonded to the di spl ay's cut-out i n the shi el ded wal l al l around the peri meter of the di spl ay. Fi gure 58 shows a shi el ded wi ndow that has j ust been manufactured, before the excess mesh has been removed. They are al l desi gned so that the mesh i s exposed around the four si des so that i t can be 360 bonded to the shi el d wal l . Fi gure 59 shows some exampl e fi gures for the SE achi eved by some types of shi el ded wi ndow materi al s. Note that these are for the materi al s themsel ves, Fig 58 measured under i deal condi ti ons. n practi ce the domi nant factors for a cabi net's SE are usual l y the desi gn of thei r assembl y, and the qual i ty of the workmanshi p when they are assembl ed i n the cabi net. Metal mesh wi ndows general l y gi ve better shi el di ng performance than conducti ve coati ngs, for a gi ven l evel of opti cal transmi ssi on l oss and degradati on of vi si bi l i ty, but can suffer from Moi r fri ngi ng effects i f not sel ected careful l y to sui t the pi xel si ze of the di spl ay. Al l shi el ded wi ndows make the di spl ay l ook di mmer, so backl i ghts may need to be more powerful . Touchscreens behi nd shi el ded wi ndows are possi bl y a good EMC sol uti on to the probl em of addi ng human-machi ne i nterfaces to shi el ded cabi nets, al though some touchscreen technol ogi es can be di ffi cul t to use wi th a shi el ded wi ndow. ' Honeycomb metal ' can al so be used for shi el di ng di spl ays, and has excel l ent SE val ues but has an extremel y l i mi ted vi ewi ng angl e. Thi s can be very useful i n securi ty appl i cati ons, because i t makes i t al most i mpossi bl e for anyone to see a di spl ay unl ess they are ri ght i n front of i t, but thi s i s not a desi rabl e feature i n i ndustri al appl i cati ons. 5. 3. 3 Venti I ati on Venti l ati on apertures can be shi el ded by fi tti ng a wi re mesh (wel ded at each wi re crossover i n the mesh) over the aperture, wi th each wi re el ectri cal l y bonded metal - to-metal to the wal l of the shi el ded cabi net al l around the peri phery of the cut-out i n the cabi net wal l . The smal l er the mesh si ze, the better the SE. Perforati ng the cabi net wal l wi th a number of smal l sl ots or hol es can achi eve the same SE and venti l ati on as fi tti ng a wi re gri l l e over a l arge aperture, and avoi ds the need to provi de RF bondi ng around the peri phery of a wi re mesh. 63 64 A number of shi el di ng manufacturers sel l pre-assembl ed shi el di ng gri l l es that may si mpl y be fi tted wi th a conducti ve gasket al l around thei r edges and then bol ted i nto pl ace on the cabi net wal l (whi ch of course must be hi ghl y conducti ve over the requi red bondi ng area). Some of these are based on wi re mesh, and some on more exoti c technol ogi es such as wi re wool or 'honeycomb metal ' to gi ve better SEs, and some exampl es are shown i n Fi gure 60. A techni que known as 'wavegui de bel ow cutoff' can provi de very hi gh val ues of SE wi th very l i ttl e i mpedance to the ai rfl ow. Thi s techni que i s descri bed i n detai l i n [30] , and honeycomb metal venti l ati on panel s are an exampl e of i ts appl i cati on. 5. 3. 4 Doors and removabI e paneI s To achi eve useful SEs, apertures i n an cabi net must be few i n number and smal l i n si ze, so doors and removabl e panel s must have frequent el ectri cal bonds al l around thei r edges. Bondi ng them wi th Fig 59 Fig 60 wi res or straps i s no good at al l above a few MHz. Fi gure 61 shows detai l s of the bondi ng of the doors and removabl e panel s usi ng a vol ume-conducti ve el astomer gasket (such as neoprene l oaded wi th si l ver-pl ated gl ass beads). EMC gaskets are avai l abl e that provi de envi ronmental as wel l as EMC seal i ng. Gaskets and thei r contact areas must not be pai nted, and careful choi ces of materi al s and metal pl ati ng fi ni shes are needed to ensure good EMC, and prevent corrosi on, over the l i fecycl e of the cabi net (see secti on 6).
Very many di fferent types of gaskets and spri ng fi ngers are avai l abl e from a number of manufacturers. Gasket choi ce requi res bal anci ng many physi cal consi derati ons, such as compl i ance, compressi on set, and posi ti on i n the gal vani c seri es, wi th el ectri cal ones such as contact resi stance. An openi ng door may requi re a soft gasket (for ease of manual cl osi ng) that al ways spri ngs back to i ts ori gi nal shape, but Fig 61 65 66 these are often di ffi cul t to combi ne wi th l ow contact resi stance. Spri ng fi ngers ('fi nger stock') are often used around doors and removabl e panel s, as shown i n Fi gure 62, but are qui te fragi l e and i n some appl i cati ons coul d easi l y be damaged. 5. 4 The effective use of gaskets Thi s gui de does not di scuss gaskets and thei r use i n any detai l , except to say that when assembl ed they shoul d be compressed to an amount wi thi n thei r manufacturers recommended range and thi s can requi re consi derabl e pressure. As menti oned i n 2. 5, even EMC gaskets that can easi l y be squashed fl at between two fi ngers can requi re very l arge compressi on forces overal l when used i n l ong stri ps, so the effecti ve use of gaskets requi res careful mechani cal and fi xi ng desi gn to prevent metal parts from bendi ng too much i n del i veri ng those forces. t i s not unusual to fi t stri ps of very soft conducti ve gaskets to the door of an Fig 62 i ndustri al cabi net, onl y to fi nd that i t becomes al most i mpossi bl e to cl ose, and once cl osed i t bends l i ke a banana openi ng up l arge gaps that defeat the purpose of the gasketti ng. t has been menti oned many ti mes i n the above text and fi gures that gaskets requi re hi ghl y-conducti ve metal surfaces to make connecti on on both of thei r si des. t i s al ways recommended that metal parts are checked for the conducti vi ty of any EMC- rel ated surfaces when del i vered by thei r suppl i er, before bei ng accepted i nto store. Such tests shoul d use very bl unt, smooth probes appl i ed wi th a very l i ght pressure, to di scover whether the suppl i er has ' acci dental l y' appl i ed a pol ymer passi vati on coati ng as they someti mes seem to. t i s easy to desi gn and make a probe based on a battery, buzzer and two smooth spri ng- l oaded contacts, that can be used l i ke a rubber stamp to de-ski l l metal surface conducti vi ty checki ng at goods-i n.
EMC gaskets must conti nue to be effecti ve over the l i fecyl ce of the cabi net despi te exposure to the physi cal envi ronment at the si te where i t i s i nstal l ed. Thi s i ncl udes such i ssues as: Mechani cal (e. g. shock and vi brati on) Cl i mati c (e. g. ai r temperature, pressure and humi di ty) Chemi cal (e. g. exposure to condensati on, l i qui ds, sprays, mi sts, vapours and dusts of vari ous types) Bi ol ogi cal (e. g. moul d growth) Wear and tear from normal use, cl eani ng, mai ntenance, etc.
Good EMC gasket manufacturers provi de
a weal th of data and appl i cati on assi stance (for exampl e [27]), coveri ng the correct choi ce of gasket materi al s and styl es for parti cul ar appl i cati ons, taki ng i nto account thei r physi cal envi ronments, and the data requi red for mechani cal desi gn to achi eve the correct compressi on wi thout di storti on of the cabi net or any of i ts parts. Gal vani c corrosi on of the gaskets or the pl ati ng they bond to i s a real possi bi l i ty, and shoul d be avoi ded by fol l owi ng the gui dance i n secti on 6. 6 Preventi ng gaI vani c corrosi on Al l of the techni ques descri bed above rel y for thei r effecti veness on achi evi ng very l ow-i mpedance metal -to-metal connecti ons over the l i fecycl e of the cabi net, despi te i ts physi cal envi ronment. The contact resi stance at each RF bond must not be permi tted to i ncrease too much over the l i fecycl e, ei ther due to oxi dati on of the metal s used, or due to gal vani c corrosi on the subj ect of thi s secti on. Di fferent metal s have di fferent posi ti ons i n the el ectro-chemi cal seri es, so when connected by an el ectri cal l y-conducti ve l i qui d (cal l ed an el ectrol yte, for exampl e ordi nary water) they form an 'acci dental battery' and a sel f-generated current fl ows i n them. The most anodi c of the metal s gets eaten away by thi s current, eventual l y di sappeari ng (or turni ng i nto non-conducti ve or semi -conducti ve corrosi on products) al together. f the choi ce of metal s i s poor for the envi ronment, gal vani c corrosi on can compl etel y destroy an el ectri cal connecti on very qui ckl y i ndeed, maybe i n j ust a few weeks. More anodi c (most easi l y corroded) Group 1 Magnesi um Group 2 Al umi ni um and i ts al l oys, zi nc, cadmi um Group 3 Carbon steel , i ron, l ead, ti n, ti n-l ead sol der Group 4 Ni ckel , chromi um, stai nl ess steel Group 5 Copper, si l ver, gol d, pl ati num, ti tani um 67 68 Fi gure 63 shows an exampl e of a si mul ated l i fecycl e test usi ng standard metal bl anks to test the gal vani c compati bi l i ty of di fferent types of conducti ve EMC gasket. [20] has a very good chapter on preventi ng gal vani c corrosi on, whi ch i s summari sed very bri efl y bel ow. Cl assi fi cati on of metal s by thei r posi ti on i n the gal vani c seri es: Fig 63 more cathodi c (l east easi l y corroded) The i dea i s that the gal vani c vol tage di fferences between the materi al s i n each group are l ow enough to al l ow them to be used i n contact wi th each other regardl ess of the envi ronment. However, i n very aggressi ve envi ronments (such as the deck of an ocean-goi ng vessel ) i t i s probabl y best to make sure that onl y i denti cal metal s (or, i f they are al l oys, i denti cal composi ti ons) are used i n contact. Coati ng or pl ati ng mati ng parts wi th the same metal (for exampl e, zi nc, ti n, or ni ckel ) hel ps keep the di ssi mi l ar metal s protected from the el ectrol yte, preventi ng gal vani c corrosi on, but depends on the qual i ty of the pl ati ng. A pi nhol e or scratch i n the pl ati ng can al l ow the metal underneath the pl ati ng to get eaten away. Fi gure 64 i s a useful tabl e gi vi ng gui dance on the combi nati ons of the metal s i n the above fi ve groups, dependi ng on thei r envi ronment, and was extracted from [20] . The fl ow of DC or AC current through an el ectri cal bond al so hastens gal vani c corrosi on, maki ng i t a more i mportant consi derati on for i ndustri al cabi nets contai ni ng el ectri cal /el ectroni c ci rcui ts. Vapour-phase corrosi on i nhi bi ti on i s a recentl y devel oped technol ogy [33] that cl ai ms to use smal l quanti ti es of a sol i d materi al that subl i mes, rel easi ng a vapour that coats nearby metal parts wi th an i nsul ati ng fi l m j ust a few mol ecul es thi ck. The fi l m i s supposed to be too weak to prevent el ectri cal bonds from bei ng made between di fferent parts, but suffi ci ent to prevent oxi dati on or gal vani c corrosi on. 7 References and further readi ng [1 ] European Uni on Di recti ve 89/336/EEC (as amended) on El ectromagneti c Compati bi l i ty. The Di recti ve' s offi ci al EU homepage i ncl udes a downl oadabl e versi on of the current EMC Di recti ve and i ts successor; a tabl e of al l the EN standards l i sted under the Di recti ve; a gui dance document on how to appl y the Di recti ve; l i sts of appoi nted EMC Competent Bodi es; etc. , al l at: http: //europa. eu. i nt/comm/enterpri se/ el ectr_equi pment/emc/i ndex. htm. The new European Uni on Di recti ve 2004/1 08/EC on El ectromagneti c Compati bi l i ty (2nd Edi ti on): http: //europa. eu. i nt/eur-l ex/l ex/LexUri Serv/si te/en/oj /2004/l _390/l _3902004 1 231 en00240037. pdf See [2] for detai l s of the transi ti on from 89/336/EEC to 2004/1 08/EC. [2] "2004/108/EC: Systems, Installations and Good Engineering Practices, Kei th Armstrong, The EMC Journal , September 2006, avai l abl e from www. compl i ance- cl ub. com (search by 'Kei th Armstrong' ) [3] "EMC for Systems and Installations, Ti m Wi l l i ams and Kei th Armstrong, Newnes, 2000, SBN 0 7506 41 67 3, RS Components Part No. 377-6463 [4] Seventeen EMC Gui des on EM phenomena, l egal compl i ance and EMC testi ng have been publ i shed by REO (UK) Ltd. They are very readabl e and practi cal , and are avai l abl e vi a the Publ i cati ons and Downl oads pages at www. cherrycl ough. com [5] A number of useful and practi cal documents on compl yi ng wi th the EMC Di recti ve are avai l abl e from the ' Publ i cati ons and Downl oads' pages at www. cherrycl ough. com [6] EC 61 508: "Functional Safety of Electrical, Electronic and Programmable Electronic Systems (seven parts) [7] EC 61 51 1 : "Functional safety: Safety instrumented systems for the process industry sector [8] EC 62061 : "Safety of Machinery Functional safety of electrical, electronic and programmable control systems for machinery [9] The 'Low Vol tage Di recti ve': 73/23/EEC "Council Directive of 1 February 1973 on the harmonization of the laws of Member States relating to electrical equipment designed for use within certain voltage limits as amended by 93/68/EEC [1 0] The 'Machi nery Safety Di recti ve': "Directive 98/37/EC of the European Parliament and of the Council of 22 June 1998 on the approximation of the laws of the Member States relating to machinery The new Machi nery Safety Di recti ve: "Directive 2006/42/EC of the European Parliament and of the Council of 17 May 2006 on machinery, and amending Directive 95/16/EC (recast) [1 1 ] "List of Resources on EMC and Functional Safety, The ET, http: //www. i ee. org/OnComms/PN/emc/ EMCandFuncti onal Safety. cfm
[1 2] "EMC Testing, a seri es i n si x parts by Ti m Wi l l i ams and Kei th Armstrong, publ i shed i n the EMC Compl i ance Journal duri ng 2001 -2, and avai l abl e vi a the 'Publ i cati ons and Downl oads' pages at www. cherrycl ough. com [1 3] "On-Site EMC Test Methods, Kei th Armstrong, EMC Test Labs Associ ati on (www. emctl a. co. uk) Techni cal Gui dance Note No. TGN 49, avai l abl e from the 'Publ i cati ons and Downl oads' pages at www. cherrycl ough. com 69 70 Fig 64 [1 4] "CE + CE does not equal CE what to do instead avai l abl e from the 'Publ i cati ons and Downl oads' pages at www. cherrycl ough. com [1 5] "Assessing an EMEnvironment, Kei th Armstrong, EMC Test Labs Associ ati on (www. emctl a. co. uk) Techni cal Gui dance Note No. TGN 47, avai l abl e from the 'Publ i cati ons and Downl oads' pages at www. cherrycl ough. com [1 6] "Electromagnetic Compatibility in Heavy Power Installations EE Col l oqui um, Mi ddl esborough U. K. , 23 rd
February 1 999, EE Col l oqui um Di gest No. 99/0666 ( SSN 0963-3308) avai l abl e from EE Sal es for around 20, sal es@i ee. org. uk or from the EE Li brary: www. i ee. org. uk/Li brary, l i bdesk@i ee. org. uk [1 7] "Achieving EMC Directive Compliance with a Spreadsheet, Kei th Armstrong, Conformi ty, February 2006, from the archi ves at www. conformi ty. com [1 8] EC 61 000-5-2: 1 997 "Electromagnetic Compatibility (EMC) Part 5: Installation and Mitigation Guidelines Section 2: Earthing and cabling [1 9] PD EC TR 61 000-5-6: 2002 "Electromagnetic Compatibility (EMC) Installation and mitigation guidelines Mitigation of external EMinfluences" [20] NAVA R AD 1 1 5 "Electromagnetic Compatibility Design Guide for Avionics and Related Ground Equipment, 3 rd
Edi ti on, June 1 988, Naval Ai r Systems Command, Department of the Navy, Washi ngton DC, USA [21 ] Defence Standard 59-41 , Part 6, ssue 1 26th August 1 994: "Electromagnetic Compatibility Code of Practice for Military Vehicles Installation Guidelines, downl oad from www. dstan. mod. uk [22] Defence Standard 59-41 , Part 7 ssue 1 1 0 th November 1 995: "Electromagnetic Compatibility Code of Practice for HM ships, Installation Guidelines, downl oad from www. dstan. mod. uk [23] "Electromagnetic Compatibility Installation Guide, Eurotherm Control s, http: //downl oad. eurotherm. co. uk/downl oad- s/DL/EMC_025464_1 . pdf [24] "EMC: Electromagnetic Compatibility, Jacques Del abal l e, Schnei der El ectri c, www. schnei der-el ectri c. com/cahi er_techni - que/en/pdf/ect1 49. pdf [25] "EMC-Compatible Enclosure Assembly, Ri ttal , www. ri ttal . co. uk [26] The author's contact detai l s are provi ded at www. cherrycl ough. com [27] Lai rd Technol ogi es Techni cal Notes, vi si t: http: //www. l ai rdtech. com/pages/ catal ogs/emi . asp and scrol l down the page. [28] "Design Techniques for EMC- Part 2: Cables and Connectors, Kei th Armstrong, The EMC Journal , Apri l 1 999, pages 7-1 6, avai l abl e vi a the ' Publ i cati ons and Downl oads' pages at www. cherrycl ough. com. Thi s arti cl e was updated and i mproved wi th consi derabl y i ncreased detai l i n The EMC Journal , May 2006 pages 31 -41 and Jul y 2006 pages 25- 38, avai l abl e from www. compl i ance- cl ub. com (search by Kei th Armstrong) [29] "Design Techniques for EMC- Part 3: Filters and surge protection devices, Kei th Armstrong, The UK EMC Journal , June 1 999, pages 9-1 5, avai l abl e vi a the ' Publ i cati ons and Downl oads' pages at www. cherrycl ough. com. Thi s arti cl e was updated and i mproved wi th consi derabl y i ncreased detai l i n The EMC Journal , September and November 2006, avai l abl e from www. compl i ance-cl ub. com (search by ' Kei th Armstrong') [30] "Design Techniques for EMC- Part 4: Shielding, Kei th Armstrong, The UK EMC Journal , August 1 999, pages 1 0-20, avai l abl e vi a the ' Publ i cati ons and Downl oads' pages at www. cherrycl ough. com. An updated, revi sed and augmented versi on wi l l be publ i shed i n The EMC Journal duri ng 2007, and wi l l be avai l abl e from www. compl i ance-cl ub. com (search by Kei th Armstrong) [31 ] "Design Techniques for EMC- Part 1: Circuit Design and Choice of Components, Kei th Armstrong, The EMC Journal , February 1 999, avai l abl e vi a the ' Publ i cati ons and Downl oads' pages at www. cherrycl ough. com. Thi s arti cl e was updated and i mproved wi th consi derabl y i ncreased detai l i n The EMC Journal , January and March 2006, avai l abl e from www. compl i ance-cl ub. com (search by ' Kei th Armstrong') [32] "Choosing and Installing Mains Filters, Kei th Armstrong and Ti m Wi l l i ams, Compl i ance Engi neeri ng magazi ne, January/February 2000, pages 68 75. Avai l abl e al ong wi th some other useful arti cl es on choosi ng fi l ters vi a the ' Publ i cati ons and Downl oads' pages at www. cherrycl ough. com [33] Vi si t www. cortecVpC . com for detai l s. The author has no di rect experi ence of thi s techni que and makes no cl ai ms for i ts effecti veness. 71 72 REO i s an ori gi nal manufacturer of hi gh qual i ty power equi pment, i ncl udi ng el ectroni c control l ers, components and el ectri cal regul ators, al l backed by the appl i cati on experti se demanded by speci al i sed, i ndustri al sectors, such as .. . Control l ers desi gned speci fi cal l y for use i n the parts and materi al s handl i ng i ndustry, together wi th a wi de range of el ectromagnets for dri vi ng vi bratory feeders. Power control l ers for adj usti ng and regul ati ng vol tage, current, frequency or power, as wel l as i ts l ong establ i shed vari abl e transformers (vari acs) up to 1 MVA and sl i di ng resi stors of al l types. These are compl emented by a range of modern, el ectroni c, vari abl e power suppl i es. Components for adapti ng vari abl e speed dri ves empl oyed i n non-standard appl i cati ons; i ncl udi ng i nductors, EMC fi l ters and braki ng resi stors. The range of i nducti ve devi ces extends i nto rai l way components for el ectri cal tracti on and rol l i ng stock, whi ch i ncl udes chokes and hi gh-frequency transformers. Speci al , toroi dal transformers used i n safety, medi cal and energy-savi ng systems pl us hi gh-frequency transfomers used i n swi tch-mode power suppl i es. Test equi pment such as l oad banks and vari abl e AC/DC power suppl i es, REO acti vel y searches for devel opment partners, parti cul arl y i n ni che markets, and consi ders thi s to be an essenti al sti mul us for creati ng new and ori gi nal i deas. SoI uti ons from REO Kei th Armstrong graduated i n el ectri cal engi neeri ng wi th a B. Sc (Hons. ) from mperi al Col l ege London i n 1 972, maj ori ng i n anal ogue ci rcui t desi gn and el ectromagneti c fi el d theory, wi th a Upper Second Cl ass Honours (Cum Laude). Much of hi s l i fe si nce then has i nvol ved control l i ng real -l i fe i nterference probl ems i n hi gh-technol ogy products, systems, and i nstal l ati ons, for a vari ety of compani es and organi sati ons i n a range of i ndustri es. Kei th has been a Chartered El ectri cal Engi neer (UK) si nce 1 978, a Group 1 European Engi neer si nce 1 988, and has wri tten and presented a great many papers on EMC. He i s a past chai rman of the EE' s Professi onal Group (E2) on El ectromagneti c Compati bi l i ty, i s a member of the EEE' s EMC Soci ety, and chai rs the EE' s Worki ng Group on ' EMC and Functi onal Safety' . Contact: Kei th Armstrong by emai l at kei th. armstrong@cherrycl ough. com or vi si t the Cherry Cl ough websi te www. cherrycl ough. com Keith Armstrong from Cherry Clough Consultants See www.reo. co. uk/ guides for lots of other practical guides, similar to this one, relating to Electromagnetic (EM) phenomena The Author EMC Fi I ter Si ngl e phase, 250 V, hi gh performance uni t sui tabl e for most appl i cati ons 3 phase, 3 x 440 V, 3 l i ne mai ns fi l ter wi th very hi gh attenuati on EMC Fi I ter EMC Fi I ter EMC Fi I ter 3 phase, 3 x 480 V bookcase styl e fi l ters, wi th very hi gh attenuati on 3 phase, 3 l i ne mai ns fi l ter wi th i ncreased attenuati on Vi ew further products on-I i ne @www. reo. co. uk Product ExampI es 73 74 3 phase, motor choke upto 1 200 A P64, 900 V Braki ng resi stors upto 1 2000 watts Hi gh-current RF fi l ters wi th i ncreased attenuati on upto 1 600 A 400 V, 3 l i ne si nusoi dal fi l ter for use wi th VSD Braki ng Resi stor Motor Choke EMC Fi I ter Si nusoi daI Fi I ter CI assi cs Automati on Systems Motor ControI Systems Communi cati on Systems RenewabI e Systems Control l ers for vi bratory feeders Fi el d bus and gsm Rheostats and vari acs Sol ar transformers Soft-starts Dri ve Systems Trai n Systems Test Systems Medi caI Systems I nducti ve Components Power EI ectroni cs Chokes and hi gh frequency transformers Power suppl i es and l oad banks Fi l ters and braki ng resi stors REO - Market Sectors Medi cal Transformers Phase-angl e and frequency control l ers Chokes, resi stors and transformers