THE OCCURRENCE OF TRANSIENT FIELDS AND ESD IN TYPICAL SELECTED AREAS

Size: px

Start display at page:

Download "THE OCCURRENCE OF TRANSIENT FIELDS AND ESD IN TYPICAL SELECTED AREAS"

Job Lee
6 years ago
Views:

1 THE OCCURRENCE OF TRANSIENT FIELDS AND ESD IN TYPICAL SELECTED AREAS Stepha FREI Techical Uiversity Berli, Istitute of Electrical Power Egieerig Eisteiufer, 587 Berli, Germay The occurrece rate of Electrostatic Discharges (ESD) depeds etremely o the eviromet. Several parameters ifluece ESD ad the severity of this iterferece. A special measuremet system to measure the trasiet fields of ESD was desiged. Measuremets were doe i differet eviromets. A theoretical model was developed, helpig to predict ad to evaluate the severity level of a eviromet towards ESD.. INTRODUCTION Electrostatic Discharges (ESD) ca cause temporary disturbaces or eve lastig destructio of electroic devices. The critical compoets of a ESD-evet are both the curret ad the always associated trasiet E- ad H-Field. The short s of the ESD-curret ca cause strog, steep-risig fields eve at larger distaces from the discharge locatio. Due to the stochastic ature of ESD it is very difficult to determie the currets ad the fields of ESD occurrig uder practice coditios. May parameters ifluece the occurrece rate ad the itesity of real ESD. The direct eviromet of a device ad the activities which happe i the eviromet determie the ESD threat. Each eviromet has its ow average hazard potetial. I the test stadard IEC -4-2 there are little cosideratios cocerig the eviromet depedecy of the ESD thread: It is however desirable to improve the ability to differetiate better the actually possible etet of the evirometal hazard potetial. If it is sure, that a certai threshold of a disturbace variable does ot or oly occur with ifiitesimal probability, i certai cases the protectio level ca be reduced. However, oly with kowledge of the occurrece rates ad the itesities a decisio ca be made. I reality ESD pulses ca have a higher disturbace itesity tha eve the strogest pulses, which are prescribed with a test to meet the IEC -4-2 stadard. I the stadard a egligible occurrece frequecy of these strog disturbaces is assumed. This is hardly provable so far. Apart from study [] published at the begiig of the 8's, there is o detailed iformatio about ESD i practice so far. The metioed study eamied oly two differet eviromets ad did ot cosider the fields of ESD. Due to the small umber of eamied eviromets ad the disregard of the ESD field effects, the results of this ivestigatio are icomplete ad caot be geeralized. I order to icrease the kowledge base cocerig the occurrece rate ad the itesity of ESD evets that really occurs i practice, log-term measuremets of the trasiet fields at three differet places were made by meas of a ew developed, comple measuremet system [2]. The large variatio of ESD ad the ecessity for further tools to determie the occurrece of ESD uder practice coditios was cofirmed. A uiversal stochastic model of ESD ca be helpful. It has to cosider the substatial parameters of ESD, ad it should be able to characterize ESD with a few, simple measuremets ad geeral cosideratios of the eviromet. With the help of a stochastic model it would be possible with little effort to receive the ESD hazard potetial that is possible i a eviromet. The stochastic model helps to determie suitable test procedures ad also to select the really ecessary protectio measures. A stochastic model for the occurrece rate of ESD cosiderig the itesity distributio is preseted here. 2. EXPERIMENTAL METHOD The developed ad applied measuremet system cosists of a specially desiged circuit for the evaluatio of fast trasiet pulses. The measuremet system is

2 Frequecy distributio Frequecy distributio Frequecy distributio equipped with several field sesors. It is optimized for ESD log term measuremets. The E- ad H-fields are measured separately. This system automatically registers the amplitude of impulsive fields withi the rage of 2 V/m up to V/m or.5 A/m up to 2.7 A/m. The pulse width ca thereby be smaller tha s. Apart from the pulse amplitudes the system records importat frequecy compoets, evirometal coditios (temperature, relative humidity) ad time. The maimum repetitio rate is higher tha 25 khz. A detailed descriptio of the developed measuremet system ca be foud i [2]. Fig. : Capacitive sesor coected to the power lie for burst detectio The ofte occurrig burst fields (switchig actios propagated by the power lies) ca be mistake for ESD evets. For this reaso a differetial measuremet method was implemeted. Burst pulses were detected by a capacitive sesor placed o the power lie of the measuremet system. If this burst sesor measures a pulse at the same time with the field sesors ad a certai relatio betwee the sigals is eceeded, the measuremet system rejects this evet. With this method, the measuremet of ESD by the observatio of the trasiet fields is possible. The measuremets preseted here were doe without this burst sesor. The criterio for Burst rejectio was the level of the pulses. 3. EXPERIMENTAL RESULTS By meas of three idetical measuremet systems at three differet measurig locatios log-term measuremets of the impulsive fields were doe. Oe locatio, the uiversity library of the Techical Uiversity Berli, was selected due to ESD promotig coditios there. ESD occur there very ofte. The device was placed i a arbitrarily chose book shelf. The secod measurig locatio was uder a workstatio, o the floor i a studet electroics laboratory. The studets there work with a lot of ESDsesitive electroic compoets. The system was located i the middle of the room. The third locatio was a computer room equipped with 8 PCs. Here the device was placed i the middle of the room o the floor. The relative humidity durig the etire measurig period raged betwee 3% ad 4%, the temperature was early costat with appro. 2 C. Frequecy (peak values H-field) Frequecy (peak values E-field). Frequecy (peak values H-field) Frequecy (peak values E-field).... a) > c) >. Frequecy (peak values H-field) Frequecy (peak values E-field) Fig. 2: Frequecy ad itesity of impulsive field evets i three differet eviromets > a) Library; duratio of measuremet 2 days; 862 evets; b) Computer room; duratio of measuremet 4 days, 2488 evets c) Electroics laboratory, duratio of measuremet 9 days, 62 evets b)

3 Site pulses/day pulses >5 V/m/day pulse >5/377 A/m/day otes Histograms of the results of the measuremets ca be foud i figure 2. The histograms show the relative frequecy of the field pulses as a fuctio of the amplitude. As epected i all three cases the weak field pulses occurred substatially more frequetly tha the stroger pulses. Amog the may impulsive evets, which were recorded by the system, also may Burst (switchig) actios ca be foud. Eve though the ESD fields ofte differ from the fields geerated by switchig actios (frequecy spectrum, repeatig rate), evertheless i practice it is very comple to eecute this differetiatio. The limited dyamic rage of the used measuremet setup ad iterferece with the eviromet cause problems. With larger field stregths the ecessary dyamic becomes available. I ormal eviromets i larger distaces large field stregths ca be produced oly by ESD. A miimum distace of m was esured betwee the measurig istrumets ad other electrical equipmet. From ivestigatios it could be determied that the switchig actios occurrig i the eviromets cause field stregths of less tha 25 V/m or 25/Z A/m. With the described differetial measuremets or a evaluatio of the data with eural etworks, a better distictio would be possible I the eamied library a ESD promotig carpet is laid. A high activity of studets ad staff further promotes the frequet occurrig of ESD. Here a average umber of 93 pulses per day was measured. I the computer room ad the electroics laboratory a older lioleum floor is laid. This floor does ot particularly promote ESD. Also less activities geeratig triboelectricity ca be foud there. Oly 78 (computer room) or 29 (electroics work space) impulsive evets per day could be measured i average. A lot of the pulses are geerated by switchig actios. of the strog discharges were probably caused by a discharge i a door hadle located i a distace of approimately 2 m from the measuremet istrumet. The E-field probe was shielded i this directio by a metal book shelf. I table the eviromets are compared. I the uiversity library a average umber of 9 to 28 ESD serious evets per day occur, which cause a field stregth of more tha 5 V/m or 5/377 A/m at a quite arbitrarily selected measurig poit. I the other eviromets less tha serious ESD evet could be measured i average per day. The ESD promotig coditios i the library like the floor ad the activities caused such big differeces. E-field peak, (far field) kid of discharge K properties of the discharge locatio D discharge curret amplitude, humidity h distace d orietatio f eviromet C temperature T speed of approach v charge voltage Uo distace walked before discharge s H-field peak, (far field) walkig speed v floorig F waitig period betwee walkig ad the discharge ESD quatity evirometparameter userparameter fieldparameter shoeproperties S Fig.3: Depedecy of ESD from the eviromet library may activities -floorig computer room electroic laboratory 78,9,4 -less discharge locatios 29,7,8 -less critical activities Table : Compariso of the average umber of strog field pulses i the differet sites The higher frequecy of large H-field amplitudes measured i the library compared to the frequecy of E- field amplitudes ca be eplaied by the fact that most 3. A STOCHASTIC MODEL OF THE OCCURRENCE OF ESD The actually iterestig parameters of ESD are the currets ad the fields, but the always precedig chargig process ca be characterized by the geerated chargig voltage. To describe the severity of a ESD, the chargig voltage is however a usuitable quatity. For the same chargig voltage, e.g. the amplitudes of the currets ca vary to up to three orders of magitude. The curret derivatives, the amplitudes ad the derivatives of the fields ca vary eve up to three orders of magitude [3].

4 Parameter Abbreviatio Eample Possible Probability Desity Fuctio f() distace walked before discharge s -2 m Epoetial Distributio 2 walkig speed v L - m/s Normal Distributio 3 floorig F 4 shoe-, garmet properties S furiture properties... 5 delay betwee walkig ad the discharge t 6 rel. humidity h -9% Normal Distributio 7 temperature T -35 C Normal Distributio 8 charge voltage U -2 kv Epoetial Distributio s,v L,F,S,t,h,T 9 kid of discharge K figer, metal piece speed of approach v Normal Distributio properties of the discharge locatio D impedace Z of the curret path 2 discharge curret di Epoetial Distributio U i o,h,t,k,v,d ma dt ma 3 distace d -3m Epoetial Distributio 4 orietatio -9 Normal Distributio 5 eviromet C reflectio 6 E-field (far field) de Epoetial Distributio i,d,,c Ema dt ma 7 H-field (far field) dh H ma dt ma Epoetial Distributio i,d,,c Table 2: Parameter ifluecig the occurrece rate ad the itesity of ESD (assumptio: electrificatio by walkig) With some additioal assumptios statemets cocerig the severity level of ESD ca be derived from the chargig voltage distributio. ESD depeds o may parameters. The table prepared by Ryser [4] was eteded i table 2 with further importat parameters. I figure 3 the depedecies are preseted graphically. It must be kept i mid that it is impossible to cosider all ifluecig parameters. Oly the most importat oes are specified here. The geeratio of charge ad the chargig voltage maily deped o the parameters specified i table, lie -7. The eact fuctioal depedecies betwee the parameters ad the value of the chargig voltage ca hardly be determied by aalytical calculatio, sice the coditios are too comple. However based o measuremets the ecessary depedecies ca be determied. By measuremets a table ca be made with the approimate chargig voltage as a fuctio of the idividual parameters. A fuctio: U f ( s, v L, F, S, h, t ) ca be geerated from this table, which determies U. The occurrece of each sigle parameter, specified i table 2, is determied by the respective probability desity distributio for each eviromet. 4. Calculatio of the resultig probability fuctio from the idividual probability fuctios If the probability desity fuctios f( i ) of the idividual parameters are kow, the total probability fuctio ca be determied by simple multiplicatio of the fuctios (), due to the stochastic idepedece of the idividual ESD ifluecig parameter:, 2,..., ) f ( i ) i f ( () A area of the -dimesioal desity fuctio may represet, e.g. the probability of a certai curret amplitude. I may cases the distributio fuctio is of special iterest, i.e. the fuctio, which determies the probability that a disturbace quatity g(,..., ) is bigger tha a certai threshold g. If a costat distributio of the desities is assumed, the distributio fuctio ca be determied (2): F ( g g ) P( g )... g (, 2,..., ) g f (, 2,..., ) d d 2... d (2)

5 5. DISCUSSION The measured distributios i figure ca be approimated by a epoetially droppig fuctio. This correspod with the assumptio that the differet chargig voltages occur with a epoetial distributio (acceptace: the walkig distaces are epoetially distributed ad o the average are the chargig voltage is a fuctio of the distace). This depedecy was already determied empirically i []. It is also obvious that ESD evets far away from the measurig poit appear more ofte tha ESD evets close to the measuremet system (uiform distributio of the discharge locatios assumed). This is just determied by the available space. Eve if a epoetial distributio is assumed here, the resultig fuctio is agai a epoetial distributio. The measured depedecy ca thus be eplaied by theoretical cosideratios. Cosiderig ESD as a comple stochastic process it is possible to determie the itesity ad frequecy of the ESD i a eviromet with a few measuremets ad observatios. A ecessary prerequisite is the kowledge of the idividual depedecies. 7. REFERENCES [] R. B. Simoic, Electrostatic Furiture Discharge Evet Rates for Metallic-Covered, Floor-Stadig Iformatio Processig Machies, IEEE Iteratioal Symposium o EMC, 982 [2] S. Frei, D. Pommereke, A Study of the Impulsive Field Occurrece Rate ad Itesity, IEEE Iteratioal Symposium o EMC, 997 [3] S. Frei, D. Pommereke, Wirkug vo trasiete Felder auf digitale Systeme, EMV 96 Karlsruhe, 996 [4] H. Ryser, The Relatioship Betwee ESD Test Voltage ad Persoel Charge Voltage, EOS/ESD Symp., 99 Eve if the accuracy of such a estimatio will ot be very well i may cases, this will cause less problems. The variatios that occur with almost all measuremets cocerig ESD, they do ot ifluece a stochastic model. As log as the kid of distributio, the epected value ad the variace correspod, the result does ot chage or it chages oly isigificatly. This agreemet ca be achieved with some measuremets ad basic cosideratios. 6. CONCLUSION At selected locatios measuremets of the trasiet fields were doe. A stochastic model for ESD was preseted. With some measuremets ad basic cosideratios the average itesity ad frequecy of ESD ca be determied. A lot of the data, which is ecessary for the modelig of ESD as a stochastic process, does ufortuately ot yet eist. This will be a importat aspect of our future work. Further a simulator program, which describes a eviromet cocerig its ESD behavior, will be developed. ACKNOWLEDGMENT The author wishes to thak the Germa Research Foudatio (DFG) for the fiacial support of this project.

Measurement of Equivalent Input Distortion AN 20

Measurement of Equivalent Input Distortion AN 20 Measuremet of Equivalet Iput Distortio AN 2 Applicatio Note to the R&D SYSTEM Traditioal measuremets of harmoic distortio performed o loudspeakers reveal ot oly the symptoms of the oliearities but also