Comparison of Injected and Radiated EMC Testing of

Comparison of Injected and Radiated EMC Testing of Active Implanted Cardiac Medical Comparison of Injected and Radiated EMC Testing Devices at the Boundary Frequency of 450 MHz of Active Implanted Cardiac Devices at Howard Bassen, Life Fellow IEEE, U. S. FoodMedical and Drug Administration Center for Devices andFrequency Radiological Health, Gonzalo Mendoza, CNI the Boundary of 450 MHz Technical Services Howard Bassen, Life Fellow IEEE, U. S. Food and Drug Administration Center for Devices

Abstract--We compared testing via radiated vs. injected susceptibility lower frequency electromagnetic field emitters [4]. EMC and Radiological Gonzalo CNI Technical Services methods specified in the ISOHealth, 14117 standard for Mendoza, electromagnetic testing of implanted cardiac devices is needed because of the compatibility of implantable cardiac medical devices. Injected testing concern for interference from EM emitters such as metal is specified at and below 450 MHz and radiated testing is specified at detectors, handheld radio transceivers, cellular phones, antiand above this boundary frequency. Experimental and computational ic field emitters [4]. EMC testing of implanted cardiac devices is needed Abstract--We compared testing via radiated vs. injected susceptibility theft/electronic article surveillance systems, and studiesmethods were performed determine in a modelcomof because of the concern for interference from EM emitters such as specified intothe ISO 14117voltages standardinduced for electromagnetic radiofrequency identification devices (RFID). The ISO an implant. The device under test (DUT) a model with a metal metal detectors, handheld radio transceivers,standard cellular phones, anti-theft/ patibility of implantable cardiac medicalwas devices. Injected testing is 14708-3 implantable neurostimulator has concerns case (5.0 x 4.5 x 3.4 cm high) containing a diode detector and a fiber article surveillance systems, and radiofrequency identificaspecified at and below 450 MHz and radiated testing is specified at and forelectronic this type of EMC and uses the methods in ISO 14117 for optic link plus an insulated wire simulating a unipolar lead. We tion devices (RFID). 14708-3 implantable above this boundary Experimental and computational stud- radiated testing. ForThe theISO frequency range ofneurostimulator 450 MHz to stan3000 evaluated induced voltagefrequency. in the model for radiated versus injected dard has concerns for this type of EMC and uses the ISOISO ies were performed to determine voltages induced in a model of an testing. The voltage from experimental measurements agreed with MHz a radiated susceptibility test is prescribed methods in both inthe 14117and for radiated testing. For the frequency range of 450 MHz to 3000 implant. Thewithin device ±1.4 underdB testor(DUT) a modeltesting with a with metal12.25 case computed values less. was Radiated 14117 ISO 14708-3 standards. Watts (5.0 delivered to cm thehigh) dipole specified in the standard thelink A MHz a radiated susceptibility test prescribed both ISOstandard 14117 x 4.5 x 3.4 containing a diode detector and induced a fiber optic test configuration is used asisshown in infig. 1.the The Vp-pevaluated ). This is in same voltage as the injected test requirement (14 We and ISO 14708-3 plus an insulated wire simulating a unipolar lead. specifies use ofstandards. a torso simulator consisting of a nonstronginduced contrastvoltage to theinspecified testversus power of 120 mW The or the model radiated for radiated injected testing. conductive plastic container filled with saline (measuring at even the optional worst-case radiated test level of 8 Watts. Lateral A test shown fig. 1. The standard specifiesin voltage from experimental measurements agreed with computed values least 51 configuration cm × 36 cmis ×used 14 as cm) andintwo plastic grids placed displacement of the dipole by 2 cm changed the voltage induced in the container. The saline had conductivity at 450 MHz of 0.33 use of a torso simulator consisting of a non-conductive plastic conwithin ±1.4 dB or less. Radiated testing with 12.25 Watts delivered to the DUT by 20% or less. For depth changes from 5 mm to 10 mm S/m (375 ohm-cm at 1 kHz) and a dielectric constant of 79. tainer filled with saline (measuring at least 51 cm × 36 cm × 14 cm) the dipole specified in the standard induced the same voltage as the voltage variations were 12.5%. There is poor agreement (by a factor The grids contain square holes are approximately 1.27 cm and two plastic grids placed in thethat container. The saline had conductest requirement (14 Vp-p). This is intesting strong contrast the of 9.9injected or more) for radiated vs. injected at the to border (0.5 in)aton grid isatpositioned the saline frequency of radiated 450 MHz separating theor two methods. Our tivity 450each MHzside. of 0.33The S/mlower (375 ohm-cm 1 kHz) and aindielectric specified test power of 120 mW even the optional worstcomputational model usedLateral at other frequencies anddipole for by andconstant allowsofthe saline pass through thethat holes. It supports 79. The gridstocontain square holes are approximately case radiated testcan levelalso of 8 be Watts. displacement of the EMC studies of other types of active implants that use the ISO14117 device testside. (DUT) . Thegrid upper grid is inpositioned cm (0.5under in) on each The lower is positioned the saline 2 cm changed the voltage induced in the DUT by 20% or less. For depth the1.27 standard’s radiated test method, such as neurostimulators. the saline and to supports a dipole antenna. For efficiency and allows the saline pass through the holes. It supports the device changes below saline surface from 5 mm to 10 mm voltage variations above onlytest used one saline) inabove the the computational (DUT). Thegrid upper(in gridthe is positioned saline and 12.5%. There is poor compatibility, agreement (by ainjected factor of 9.9 or more) for weunder Index were terms-electromagnetic testing, ISO omission of efficiency the upper grid supportsThe a dipole antenna. For we only usedchanged one grid (in the the vs.radiated injected testing testing at the border frequency of 450 MHz sepa- model. 14117radiated standard, results of induced voltage in ofthe saline) in the computational model. The omission theDUT upper by gridless rating the two methods. Our computational model can also be used at computational 2.5 %.the computational results of induced voltage in the DUT by changed other frequencies and for EMC studies of other types of active implants than I. INTRODUCTION less than 2.5 %. that usereports the ISO14117 standard’s radiated test method, as neuroThis paper on studies related to EMC testingsuch standards stimulators. for medical devices. It deals specifically with two related

methods for testing active implantable cardiac devices Indexthat terms-electromagnetic testing, ISO 14117 (devices utilize electroniccompatibility, circuitry) injected as specified in ISO radiatedtesting testing is specified at and below 450 MHz 14117standard, [1]. Injected and radiated testing is specified at and above this boundary frequency. Testing to this standard is performed by all major manufacturers in the U.S. and other counties. Other I. Introduction implantable device EMC standards such as ISO 14708-3 for implantable [2]to EMC specify test This paperneurostimulators reports on studies related testing similar standards for medmethodologies ISOspecifically 14117. The predecessor to 14117 is ical devices.to It deals with two related methods for testing 1. 1. The configuration: AAMI/ANSI PC69 [3]. It was developed to address concerns active implantable cardiac devices (devices that utilize electronic cir- Fig.Fig. TheISO ISO14117 14117standard standardradiated radiatedsusceptibility susceptibility test test configuratorso simulator, plastic grid, dipole antenna, device under test (DUT). about cuitry) interference of implanted cardiac pacing and as specified in ISO 14117 [1]. Injected testing is specified at and tion: torso simulator, plastic grid, dipole antenna, (DUT). defibrillation devices from testing cellular phonesatand numerous below 450 MHz and radiated is specified and above this The dipole is driven by a radiofrequency (RF) generator. The _________________ boundary frequency. Testing to this standard is performed by all major device The dipole driven by a radiofrequency The (DUT)ofis the underis test (DUT) is positioned(RF) ongenerator. the top surface manufacturers in the U.S. and other counties. Other implantable device lower positioned on the top surface of the lower grid at the center of the torso grid at the center of the torso simulator. Testing is “The mention of commercial products, their sources, or their use in EMC standards such as ISO 14708-3 for implantable neurostimulators simulator. Testing is performed with 120 mW net power to the performed with 120 mW net power to the dipoledipole for for the connection with material reported herein is not to be construed as either [2] anspecify actual test or implied endorsement products the methodologies similaroftosuch ISO 14117. Theby predecessor to frequency the frequency range MHz ≤≤ ƒƒ< < 1 000 MHz.MHz. For worst-case testing range 450450MHz 1 000 For worst-case Department Health andPC69 Human 14117 isofAAMI/ANSI [3]. ItServices.” was developed to address concerns an optional level of 8level W is used. Theisdipole testing an optional of 8 W used.antenna is placed on the top

64

about interference of implanted cardiac pacing and defibrillation devices from cellular phones and numerous lower frequency electromagnet-

grid (above the saline) with the axis of the antenna elements parallel to the X-axis.

“The mention of commercial products, their sources, or their use in connection with material reported herein is not to be construed as either an actual or implied endorsement of such products by the Department of Health and Human Services.”

For frequencies at and below 450 MHz the ISO standard prescribes an injected susceptibility technique. No antennas or leads of the DUT are involved in the injected test. A "tissue injection net-

©2016 IEEE Electromagnetic Compatibility Magazine – Volume 5 – Quarter 4

For frequencies at and below 450 MHz the ISO standard surface of the saline. We used a lead with a length of 58.5cm, Forprescribes frequencies at and susceptibility below 450 MHz the ISO of the of saline. We used a lead with a length of 58.5cm, an injected technique. No standard antennas or surface composed a solid-conductor insulated wire with a diameter prescribes an injected susceptibility technique. No antennas or composed of a solid-conductor insulated wire with diameter leads of the DUT are involved in the injected test. A "tissue of 1.6 mm (fig. 3). Another configuration used aa short lead (4 leads of the DUT are involved in the injected test. A "tissue (fig.The 3). outer Another configuration used a short lead (4 4.6 injection network" is specified by the 14117 standard and is of 1.6 cm mm long). diameter of the insulated wire was injection network" is specified by the 14117 standard and is cm long). The outer diameter of the insulated wire was 4.6 (4 cm long). The outer diameter of the insulated wire was 4.6 mm. work" is specified by the 14117 standard is composed of coaxcomposed of coaxial components. For and frequencies of 10 - 450 mm. The spiral lead configuration on the grid is specified by composed coaxial components. For -network 450 mm. lead configuration on isthe by lid The spiral lead configuration on grid specified byspecified the ISO the ial components. For frequencies of 10frequencies - 450 MHz, aaof signal MHz, aofsignal generator is connected through 5010 ΩgeneratheThe ISOspiral 14117 standard. Thetheposition ofgrid the is lead below MHz, a signal generator isofconnected through 50 Ω2). network to tor the generator cardiac device A peakofto the14117 standard. The position of the lead below the lid of the ISO 14117 standard. The position of the lead below the lid ispulse connected through athe 50 Ω network to thea(fig. pulse generator of the implant was 8.5 mm. At the distal end of the lead, 3mm to the pulse generator of device (fig. 2). peak voltage of 14 V2). cardiac (open circuit) is injected intoto the of implant was 8.5 mm. AtAt thethe distal enddistal of the lead, was the cardiac device (fig.the A peak to peak voltage of A 14 peak V (open the was 8.5 mm. Atproximal the end lead,inside 3mm the wasimplant stripped bare. end ofof3mm allthe leads, peakconnectors voltage of 14 V (open circuit) is injected into the of eachinto setthe of connectors leads of the DUT.setPer Annex-A stripped bare. At the proximal end of all leads, inside the case, circuit) is injected of each of leads of the of was stripped bare. At the proximal end of all leads, inside case, a connection to the RF sensor was made. For allathe cases, connectors of each of leads ofofthethe DUT. Per Annex-A of ISO thesetmagnitude voltage based connection to the RF was made.was For all cases,For bothallexperiDUT.14117 Per Annex-A of ISO 14117 the magnitude of the is voltage is on case, a connection tosensor the sensor made. both experimental andRF computational, the lead for cases, the DUT ISOcalculations 14117 theformagnitude of the voltage is based on magnetic field strengths that were estimated mental and computational, the lead for thethe was positioned at based on calculations for magnetic field strengths that were esti- both experimental lead for the DUT was positioned atand thecomputational, centerline of theDUT dipole. calculations for magnetic field strengths that were estimated formated worst-case RF emitters. the centerline of the dipole. for worst-case RF emitters. was positioned at the centerline of the dipole. for worst-case RF emitters.

Fig.Fig. 2. Tissue injection network. 2. Tissue injection network. Fig. 2. Tissue injection network.

The goals of our study included addressing the following. The goals ofour our study study included addressing thethe following. One The included following. Onegoals goal of was to determine the addressing correlation between radiated goal was totodetermine thethe correlation between radiated testing Onetesting goal was determine correlation between radiated versus injected testing as prescribed by the ISO 14117 versus injected testing as prescribed by the ISO standard. testing versus injected testing prescribed by the14117 ISOfrequencies 14117 standard. This was done inas the narrow range of This was done in the narrow range of frequencies (425 and standard. This was done in the narrow range of frequencies (425 and 450 MHz) that separate these two distinct450 tests MHz) that separate thesethe twovoltage distinct teststwo methods. We tests evalu(425methods. and 450 MHz) that separate these distinct We evaluated induced in a DUT at the methods. evaluated the power voltage involtage. a DUT at the ated We the voltage in a DUT atinjected the prescribed dipole input 3. Experimental system - simulated implant: metal case, insulated solid prescribed dipoleinduced input orinduced This was Fig. 3. 3. Experimental system simulated implant: metal case, solid conductor wire, spiral lead- configuration - unipolar lead. prescribed dipole input power or injected voltage. This was power or injected voltage. This was done for radiated testing by a Fig.Fig. Experimental system - simulated implant: metal case,insulated insulated done for radiated testing by using an experimental model of conductor wire, spiral lead configuration - unipolar lead. done for radiated testing by using an experimental model of a using an experimental model of a simulated implant to validate a wire, spiral lead configuration - unipolar lead. A. conductor Experimental studies simulated implant to validate a computational model that we solid Experimental studies simulated implant to validate a computational thatthe we computational model that we developed. Oncemodel validated, developed. Once validated, the computational model would A. The purpose of the experimental system for radiated testing developed. Oncemore validated, theallow computational modelimplant wouldofand The purpose of the experimental system for radiated testing computational model would many configurations allow many configurations of themore simulated was to validate the computational model of the system. The allow many more configurations of Another the simulated implant and the was the simulated implant andtoleads. goal was to determine, A.experimental Experimental Studies to validate the computational model the system. The leads. Another goal was determine, for radiated testing, model was limited to of a few configurations, leads. Another goal was to determine, for radiated testing, thee.g. experimental model was limited to a few configurations, for radiated testing, the variability when slight deviations from variability when slight deviations from the standard exist, whereas the validated computational model could be variability when slight deviations from the exist, e.g.ofthe whereas the ofmore validated model could be a the standard exist, e.g. theimplant saline, depth The purpose the experimental system different for radiated testing wasused dipole height above the dipole saline,height depthabove of standard the below configured easilycomputational in many ways. We dipole height above the saline, depth of the implant below the configured more easily in many different ways. We used a 36 the implant below the surface of the saline, and lateral position to validate the computational model of the system. The experimensurface of the saline, and lateral position of the dipole with dipole antenna placed above a saline-filled tank, (51 cm × surface of to the saline, and lateral position of the dipole with of dipole antenna placed above a saline-filled tank, (51 cm × 36 respect the DUT. Another goal is to determine theto effects of the dipole with respect to the DUT. Another goal is detertalcm model was limited to a few configurations, whereas the validat× 14 cm) and a submerged simulated implant containing a respect to the DUT. Another goal isvoltage to determine thevoltage of cmed×computational 14 optic cm) and amodel submerged containing different impedances on induced ineffects an implant, mine theinput effects of different input impedances on could besimulated configured more in many aand fiber voltage monitoring link. Forimplant botheasily experimental different inputininjected impedances voltagetesting. induced in an implant,To fiber optic voltage monitoring link. For both experimental and on forinduced both andforon radiated To address an implant, both injected and radiated testing.these different ways. We used a dipole antenna placed above a computational studied the DUT was positioned salinedirectly for questions both injected and radiated testing. modeling. To address these studied the DUT was positioned directly on in performed The radiated computational address we these questionscomputational we performed computational modeling. filled tank, (51 cm × 36 cm × 14 cm) and a submerged simulated center under the dipole. We followed the method specified questions we performed computational modeling. radiated under the dipole. We followed the method specified in the model was validated experiments for aThe subset of the The radiated model wasby validated by experiments for a subset of center implant containing a fiber optic voltage monitoring link. For both ISO 14117 for frequencies above 450 MHz. We evaluated model was validated byDUT experiments for testing. a subsetFor of injected the ISO 14117 for frequencies above 450 MHz. We evaluated the configurations of the the configurations of the DUTfor forradiated radiated testing. For injected experimental and computational studied the DUT was positioned induced voltage at 450 MHz and 425 MHz at the input to the configurations of computational the DUT for radiated testing. For injected voltage atunder 450 MHz 425 MHz atthe themethod input toslightly the testing, modeling was since used since the induced implant model to see if and significant changes occurspectesting,only only computational modeling was used the configdirectly on center the dipole. We followed testing, only computational modeling was used since the model toforsee iffrequency significant changes occur slightlyThe configuration of physical below the14117 prescribed range for radiated testing. uration of physical electronicelectronic componentscomponents is quite simpleisandquite implant ified in ISO frequencies above 450 MHz. We evaluated configuration of physical electronic components is quite below the prescribed frequency range for radiated testing. The simple and computed are reliable. dipole antenna (Stoddart Electro Systems 91598-2) was results results are results reliable. the induced voltage at 450 MHz and 425 MHz at the input to the simplecomputed and computed are reliable. dipole antenna (Stoddart Electro 91598-2) was adjusted so to each its arms was Systems a quarter wavelength implant model see of if significant changes occur slightly below of the adjusted so each of its used. arms was a quarter wavelength of the II. METHODS AND MATERIALS specific frequency driven by aThe radiofrequency the prescribed frequency rangeItforwas radiated testing. dipole II. METHODS AND MATERIALS specific frequency used. It was driven by a radiofrequency (RF) signal with 91598-2) 1 kHz was amplitude modulation The test systems experimental and computational) for antenna (Stoddartgenerator Electro Systems adjusted so each II. Methods and (both Materials signal generator IFR with3416), 1 kHz amplitude modulation The test systems (both experimental andantenna computational) for a (RF) and RF frequency power amplifier radiated testing consisted of a dipole placed above of(Aeroflex, its arms wasModel: a quarter wavelength of the an specific (Aeroflex, Model: IFR 3416), and10W1000). an RF power amplifier radiated testing consisted of a dipole antenna placed above a (Amplifier Research, Model: A 20 dB saline-filled tank, and above a submerged model of an implant used. It was driven by a radiofrequency (RF) signal generator with dual The test systems (both experimental and computational) for radiat(Amplifier Research, Model: 10W1000). A 20wasdBused dual saline-filled tank, above a submerged model of an coupler (ANZAC, Model: CH-132) (figs. 1 and 3).and The height of antenna the dipole above theaimplant surface 1 directional kHz amplitude modulation (Aeroflex, Model: IFR 3416), and an RFalong ed testing consisted of a dipole placed above saline- of directional coupler (ANZAC, Model: CH-132) was used along (figs. 1 and 3). The height of the dipole above the surface of with coaxial cables and two power meters (Hewlett Packard thefilled saline mm or 20model mm of to an study the(figs. effects of with power amplifier (Amplifier Research, Model: 10W1000). A 20 dB tank,was andeither above 15 a submerged implant 1 and coaxialThis cables andwas twoused power meters forward (Hewlettand Packard 8482A). setup to monitor reflected the this saline was either 15 dipole mm orabove 20height mm tofollows studyof the effects of test dual directional coupler (ANZAC, Model: CH-132) wasand usedreflected along 3). parameter. The height ofThe the the surface the was 20 mm ISOsaline 14117 8482A). This setup was used to monitor forward power into the dipole. We delivered a series of net powers thisspecifications parameter. mm height follows ISO 14117 with coaxial cables and two metersa (Hewlett Packard either 15 mmThe orfor 2020 mm to study the effects of this parameter. The radiated fields above 450 MHz.test The power into the dipole. Wepower delivered series of net powers from 1 mW to 500 mW to the dipole antenna during testing. specifications for radiated fields 450(5.0 MHz. 8482A). This to setup used to forward andduring reflected 20 mm height follows ISO 14117 test specifications forcm radiated experimental model consisted of a above metal case x The 4.5 cm from 1 mW 500was mW themonitor dipole testing. Multiple power levelsto were used antenna since the detector in our experimental model consisted of a metal case (5.0 cm x 4.5 cm power into the dipole. We delivered a series of net powers 1 fields above 450 MHz. The experimental model consisted of a x 3.4 cm high) with an insulated solid conductor wire to Multiple power levels were used since the detector from into our simulated implant was slightly nonlinear in its input output x 3.4 cm high) with an insulated solid conductor wire to mW to 500 mW to the dipole antenna during testing. Multiple metal case (5.0 cm x 4.5 cm x 3.4 cm high) with an insulated solid represent a unipolar lead that penetrated the case through a simulated implant was slightly nonlinear in its input to output response. The DUT was a simulated pacemaker/ defibrillator represent a unipolar thatvoltage the case throughthe power levels the detector in our simulated conductor wire tolead represent apenetrated unipolar lead that penetrated waterproof hole. An RF sensor was connected ata the response. The were DUTused wassince a simulated pacemaker/ defibrillator waterproof hole. aAn RF voltage connected at conthe implant was slightly nonlinear in its input to output response. The case through waterproof hole.sensor An RF was voltage sensor was nected at the proximal end of the lead, inside the case. The implant case rested on a grid and the top of the case was 5 mm below the surface of the saline. We used a lead with a length of 58.5cm, composed of a solid-conductor insulated wire with a diameter of 1.6 mm (fig. 3). Another configuration used a short lead

DUT was a simulated pacemaker/ defibrillator but also could represent a neurostimulator or other medical implant. It consists of a metal case radiofrequency detection circuitry and a simulated lead. The top of the DUT was positioned in the saline filled tank at a depth of 5 mm and 10 mm.

©2016 IEEE Electromagnetic Compatibility Magazine – Volume 5 – Quarter 4

65

circuitry andrepresent simulated lead. The The top top the medical DUT was wasDUT analog voltagedesigned data out outanalog theoptical salinelink tank via aa fiber fiber optic optic and aa simulated lead. the DUT analog voltage data ofof the saline tank via butcircuitry also could a neurostimulator orofof other to a custom (Z-Axis Phelps, positioned thesaline filledcase tankradiofrequency depthofof55detection mmand and10 10NYcable. cable. The transmitter 0.6inch aluminum cube, withan an positioned ininthe filled tank atataadepth mm The transmitter isisaa0.6 aluminum cube, with implant. It consists ofsaline a metal 14532, U.S.). The transmitter ofinch this link carries detector , and an input bandwidth of 2 Hz – 50 mm. input range of 10 m V p-p , and inputtank bandwidth of 2 Hz – 50 mm. and a simulated lead. The top of the DUT was analog inputvoltage range ofdata 10 mout Vp-p circuitry of the an saline via a fiber optic KHz. The receiverfor for the fiber optic linkwas was placed 5meters meters Thesaline saline hadaa0.18% 0.18%tank NaCl concentration (resistivity ofcable. KHz. The receiver the fiber optic link placed 5 The had NaCl concentration (resistivity of positioned in the saline filled at a depth of 5 mm and 10 The transmitter is a 0.6 inch aluminum cube, with an bandwidth of 2 Hz – 50 KHz. The receiver for the fiber optic link was The saline had a 0.18% NaCl concentration (resistivity of 375 ohmfrom the saline tank avoid it from from RF interference. 375ohm-cm ohm-cm at11kHz) kHz)asasspecified specifiedininthe the14117 14117standard. standard.At Atinput from the toto itbandwidth 375 , and anavoid input of 2interference. Hz – 50 mm. range ofsaline 10 mthe Vtank p-p placed 5 meters from saline tank to avoid it from RFRF interference. cm at 1 kHz) asatspecified in the 14117 standard. At the proximal Calibration of the detector circuit was performed follows. the proximal end of the implant, the metallic case contained a Calibration of the detector circuit was performed follows. the proximal end of the implant, the metallic case contained a KHz. The receiver for the fiberwas optic link was placed The 5asas meters The had a 0.18% NaCl concentration (resistivity Calibration of the detector circuit performed as follows. end saline of the implant, the metallic case contained a peak detectorof The input to the detector was an SMA coaxial connector. This peak detector circuit, an optically linked analog transmitter The input to the detector was an SMA coaxial connector. This peak detector circuit, an optically linked analog transmitter saline tank avoid it connector. from RFThisinterference. 375circuit, ohm-cm at 1 kHz) as specified in the 14117 standard. At from input tothe the detector was antoSMA coaxial allowed an optically linked analog transmitter and fiber optic cables allowed calibrated net power to be fed into the 50 Ω resistor and fiber optic cables that entered the metallic case. These allowed calibrated net power to be fed into the 50 Ω resistor and fiber optic cables that entered the metallic case. These ofpower the detector circuit was performed as rest follows. the proximal end the implant, the metallic case contained a Calibration calibrated to bedetector fed into the 50 Ω resistor and the of that entered theof metallic case. These weretoencapsulated inthe parafand thenet rest the circuit. Wecalibrated calibrated thedetector detector were encapsulated in paraffin paraffin wax waterproof entireThe and the rest the detector circuit. We the were encapsulated wax to waterproof the entire input to theofof detector was an SMA coaxial connector. This peak detector circuit, in an optically linked analog transmitter the detector circuit. We calibrated the detector circuit using an fin wax to waterproof the entire assembly. The proximal end of the circuit usingan annet external signal generator (425 MHz and450 450 assembly. The proximal end the simulated lead connects circuit using external signal generator (425 MHz and The proximal end ofofthe simulated lead connects totoallowed calibrated power to be fed into the 50 Ω resistor andassembly. fiber optic cables that entered the metallic case. These signalNet generator (425 MHzthe anddetector 450 MHz).was Net power into versus simulated lead connects to metal a 50 Ω case load inside theimplant. metal case of the external MHz). power into measured a 50 Ω load inside the of the While power intocircuit. the detector was measured versus a 50 Ω load inside the metalwax casetoofwaterproof the implant. the rest Net of the detector We calibrated the detector were encapsulated in paraffin theWhile entirethe andMHz). the voltage detector was measured versus voltage out. Thedata power versus the implant. While the impedancecan avary pacemaker can out. The power power versus voltage was fitted input impedance aend pacemaker greatly, wevary used voltage out. versus voltage data was fitted toto aa input impedance ofofainput pacemaker canofvary greatly, we used using an The external signal generator (425 MHz and 450 assembly. The proximal of the simulated lead connects to aacircuit voltage data was fitted to a polynomial curve to allow interpolation greatly, we used a 50 Ω load in the experimental study because of polynomial curve to allow interpolation of measured data. 50 Ω load in the experimental study because of the following. polynomial curve into to allow interpolation of measured data. 50 Ω Ω load load inside in the experimental study because of the following. Net power the detector was measured versus a 50 the metal case of the implant. While the MHz). measured data. us This allowed us to characterize thedelivered power delivthe following. When450 delivering 450 MHz RF via coaxial cables to aload of This This allowed us to characterize the power delivered into the When delivering 450 MHz RF via coaxial cables to a allowed to characterize the power into When delivering MHz RF via coaxial cables to a load input impedance of a pacemaker can vary greatly, we used a voltage out. The power versus voltage data was fitted to athe ered into the implant versus voltage outof of the implant. With this load much higher than 50 ohms,the thereflected reflected power power will approach implant versus voltage out of the implant. With this much higher than 50 ohms, the reflected power will approach implant versus voltage out the implant. With this much higher than 50 ohms, will approach 50 Ω load in the experimental study because of the following. polynomial curve to allow interpolation of measured data. information we us computed the ratiothe of volts intoofof thevolts dipoleinto antenna 100% ofofof thethe forward power, thus making the power and information we computed the ratio volts into the dipole dipole 100% the forward power, thus making thetovoltage power andThis information we ratio the 100% forward thus making the power allowed to computed characterize the power delivered into the When delivering 450 MHzpower, RF via coaxial cables a loadand vs. volts received at the 50 Ω load of the DUT. into the load undeterminable. One end of the load is grounded to antenna vs.volts volts received the 50 loadofofthe theWith DUT. this voltage into the50 load undeterminable. Oneend end theload loadisisimplant antenna vs. received atatthe ΩΩload DUT. voltage into the load undeterminable. One the versus voltage out of 50 the implant. much higher than ohms, the reflected power willofof approach grounded thecase. case. Thevoltage voltage induced inmeasured the 50ΩΩwith loadisisinformation we computed the ratio of volts into the dipole the of case. voltage induced inthus the 50 Ω load is grounded totoforward the The induced in the 50 load 100% theThe power, making the power and B. Computational ComputationalStudies Studies B. measured with an RF diode detector circuit. The detector an RF diode detector circuit. The detector is composed of a zero measured withload an undeterminable. RF diode detectorOne circuit. The voltage into the end of thedetector load is isisantenna vs. volts received at the 50 Ω load of the DUT. RadiatedStudies testing.For Foran ananalysis analysisofofradiated radiatedimmunity immunity 1)1) Radiated testing. composed zero bias Schottky diode detector (Advanced bias Schottky detector (Advanced Semiconductor, Inc., B. Computational composed ofofdiode aazero bias Schottky diode grounded to the case. The voltage induced indetector the 50 Ω(Advanced loadASI is testing we used SEMCAD X 14.8 electromagnetics software B. Computational Studies testing we used SEMCAD X 14.8 electromagnetics software Semiconductor, Inc., ASI 3486) with axial leads plus a passive 3486) with axial leads plus a passive impedance matching circuit Semiconductor, ASI 3486) withcircuit. axial leads a passive measured with an Inc., RF diode detector The plus detector is that utilizes the finite difference time domain (FDTD) method that utilizes the finite difference time domain (FDTD) method impedance matching circuit (fig. 4). Using Using 50(Advanced loadus the 1) 1) Radiated testing. an analysis of immunity radiated immunity impedance circuit (fig. 4). aa50 ΩΩload ininto the (fig. 4). of Using a 50 Ω load in the experimental model allowed Radiated testing. For anFor analysis of radiated testing composed amatching zero bias Schottky diode detector (Schmid & Partner Engineering, AG Zurich, Switzerland). We & Partner Zurich, Switzerland). experimental model allowed to calibrate calibrate theand case radio-testing used SEMCAD X 14.8 AG electromagnetics softwareWe experimental model ususdetection to the radiocalibrate the case radio-frequency circuitry validate we(Schmid usedwe SEMCAD X 14.8Engineering, electromagnetics software that utilizes Semiconductor, Inc., ASIallowed 3486) with axial leads plus acase passive used a Windows PC workstation with hardware GPU a the Windows PC workstation with hardware GPU frequency detection circuitry and validate aitscomputational computational utilizes finite timemethod domain (FDTD) method frequency detection circuitry and validate a computational model that also had aUsing 50 Ω load inputinport. theused finite difference timedifference domain (FDTD) (Schmid & Partimpedance matching circuit (fig. 4). a 50ataΩ load the that acceleration the NVIDIA Tesla Tesla K40Switzerland). GPU Accelerator Accelerator the NVIDIA model thatmodel also had 50us load at its its input port. After(Schmid & Partner Engineering, AG Zurich, model that also aa 50 ΩΩto load atwas input port. After experimental allowed calibrate the case radioAfter validation thehad computational model changed to a10 kΩ neracceleration Engineering, AG Zurich, Switzerland). WeK40 used aGPU Windows PCWe (NVIDIA Corporation, Santa Clara, Clara,with CA,hardware U.S.). Simulations Simulations (NVIDIA Corporation, Santa CA, U.S.). validation the computational computational model waschanged a10 kΩ kΩused a Windows PC GPU workstation validation the model was totoa10 frequency detection validate achanged computational input impedance forcircuitry the rest of and the studies. workstation with hardware acceleration the NVIDIA Tesla GPU were performed at 425 MHZ and 450 MHz. Spiral andshort short were performed at 425 MHZ and 450 MHz. Spiral and input impedance for the rest of the studies. the NVIDIA Tesla K40 Accelerator inputthat impedance forathe the at studies. model also had 50rest Ω of load its input port. After acceleration K40lead GPU Accelerator (NVIDIA Corporation, SantaGPU Clara,different CA, U.S.). input configurations were evaluated with lead configurations were evaluated with different input (NVIDIA Corporation, Santa Clara, CA, U.S.). Simulations validation the computational model was changed to a10 kΩ Simulations were performed at 425 MHZ and 450 MHz. Spiral and of a impedances the implant. implant. The short lead consisted impedances inin short lead consisted of a were performed at the 425 MHZ andThe 450with MHz. Spiral and short input impedance for the rest of the studies. short lead configurations were evaluated different input straight, 40mm mmlength length thesame samewire wire and insulation the straight, 40 the insulation asasthe lead configurations wereofofevaluated withand different input impedances in theWe implant. The short lead consisted ofelectromagnetics a straight, spiral lead. used a 3D computational spiral lead. a 3D electromagnetics impedances in We the used implant. Thecomputational short lead consisted of a 40 software mm lengthmodel of the same wire and insulation as themodel spiral very lead. software model thatreplicated replicated the physical model very closely that the physical straight, 40 mm length of the same wire and insulation asclosely the We(figs. used1a1and 3D computational electromagnetics software model (figs. and 5). It consisted of a dipole and a saline filled tank 5). It consisted of a dipole and a saline filled tank spiral lead. We used a 3D computational electromagnetics that replicated theplastic physical model very closely (figs. 1 and 5).filled It with a single plastic grid in the saline. The tank was filled with with a single grid in the saline. The tank was with software model that replicated the physical model very closely consisted of a dipole and apermittivity saline tank with asaline single plastictank ofof saline with relative permittivity 79and conductivity saline with aaItrelative ofofand 79 aaconductivity (figs. 1 and 5). consisted of afilled dipole aand filled grid in the saline. The tank was filled with saline with a relative 0.33 S/mcorresponding corresponding to0.18% 0.18% NaCl at450 450 MHz. 0.33 S/m MHz. with a single plastic grid in to the saline.NaCl The at tank was filled with permittivity and a conductivity 0.33 to of saline with ofa 79 relative permittivityofof 79S/m andcorresponding a conductivity 0.18% NaCl at 450 MHz. 0.33 S/m corresponding to 0.18% NaCl at 450 MHz. Figure 3a. RF detector: hardware implementation, Figure3a. 4a. RF detector: hardware implementation, Figure RF detector: hardware implementation, Figure 3a. RF detector: hardware implementation,

Figure4a. 4a.Computational Computationalmodel modelofofimplant. implant. Figure

Thedipole dipoleantenna antennawas wasdriven drivenby byaasource sourcewith withan animpedance impedance The of 50 Ω. The software provided the reflection coefficient at the of 50 Ω. The software Figure4b. 4b.RF RFdetector: detector:circuit circuitdiagram. diagram. Figure 4a. Computational modelprovided of implant.the reflection coefficient at the Figure input to the antenna that enabled the net input power be Figure 4b. RF detector: circuit diagram. Figure 5a.to Computational implant. input the antenna thatofenabled the net input power totobe Thevoltage voltage measured withaafiber fiberoptically opticallylinked linkedanalog analogThe dipole antenna wasmodel driven by a source with an impedance The isismeasured with determined. The model of the implant included a metallic case determined. The model of thethe implant included a metallic case transmitter connected thedetector. detector.The Thedetector detectordelivers deliversaaof 50 Ω. The software provided reflection coefficient at the transmitter connected totothe Figure 4b. RF detector: circuit diagram. and unipolar pacemaker neurostimulator lead.The The case and unipolar pacemaker oror lead. case Thevoltage voltage isis1measured with afrom fiber optically linked analog transThe dipole antenna was driven by a neurostimulator source withinput an impedance demodulated 1kHz kHzsignal signal theRF RFsignal signallinked received bythe theinput toaathe antenna that enabled the net power toofbe demodulated from the received by The measured with a fiber optically analog mitter connected to the detector. The detector delivers a demodulat50 Ω. The software provided the reflection coefficient at the transmitter connected to the detector. The detector delivers a determined. The model of the implant included a metallic case ed 1 kHz signal fromsignal the RFfrom signalthe received by thereceived DUT to a custom inputa tounipolar the antenna that enabled the net input power to be pacemaker or neurostimulator lead. The case demodulated 1 kHz RF signal by the and designed analog optical link (Z-Axis Phelps, NY 14532, U.S.). The transmitter of this link carries detector analog voltage data out of the saline tank via a fiber optic cable. The transmitter is a 0.6 inch aluminum cube, with an input range of 10 m Vp-p, and an input

66

determined. The model of the implant included a metallic case and a unipolar pacemaker or neurostimulator lead. The case was penetrated by a 1.6 mm diameter wire surrounded by 4.6 mm diameter insulation.

©2016 IEEE Electromagnetic Compatibility Magazine – Volume 5 – Quarter 4

was penetrated by a 1.6 mm diameter wire surrounded by 4.6 2) Injected Testing. The ISO standard’s injection test wasdiameter penetrated by a 1.6 mm diameter wire surrounded by 4.6 specification 2) Injected uses Testing. The setup ISO as standard’s mm insulation. a coaxial shown ininjection fig. 7. test It > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION specification NUMBER (DOUBLE-CLICK HEREasTOshown EDIT)in< fig. 7.4 It mm diameter insulation. uses a coaxial setup requires monitoring only forward power with a directional requiresFor monitoring only forward power with a directional coupler. III. Results the analysis of injected immunity testing we used coupler. For the analysis injected immunity testing we used and ofcircuit design software (National was penetrated by a 1.6 mm diameter wire surrounded by 4.6 2) Multisim Injected simulation Testing. The ISO standard’s injection test Multisim simulation and circuit design software (National Instrumentsuses Corporation, Austin, U.S.). mm diameter insulation. specification a coaxial setupTX as 78759-3504, shown in fig. 7. It Instruments Corporation, Austin, power TX 78759-3504, U.S.). requires monitoring only forward with a directional A. Validation of the radiated computational model – coupler. For the analysis of injected immunity testing we used III. Results comparison with measurements III. Results Multisim simulation and circuit design software (National Instruments Corporation, TX 78759-3504, U.S.).model – A. Validation of theAustin, radiated computational In order to validate the of computational model we normalized experimenA. Validation the radiated computational model – comparison with measurements talcomparison and computational results to 1 volt peak into the dipole at 425 MHz with measurements III. Results and at 450 MHz. Then we obtained the received voltage of the implant. In order to validate the computational model we normalized results Table I. computational The table showsto the measured InValidation orderaretoshown validate theradiated model normalized experimental and computational results 1ratio voltofwe peak into –the A.The of in the computational model to computed voltages out of the DUT. This demonstrated the validity of experimental and computational results to 1 volt peak intothe the dipole at 425 MHz and at 450 MHz. Then we obtained comparison with measurements Figure 5b. Computational model of radiated test setup. the computational model. Agreement by a factor of 1.38 or less is Figure Computational model radiated setup. dipole atvoltage 425 MHz and implant. at 450 MHz. Then we Figure 5b.5b.Computational model of of radiated testtest setup. received of the The results areobtained shown the in A feed through connector for the patient lead was modeled to Inacceptable received voltage of the implant. The are shown in due totable the wide range of uncertainties inresults both computationTable I. The shows the ratio of measured to computed order to validate the computational model we normalized A feed through connectormodel. for the We patient leadvarious was modeled to Table I. The table shows the ratio of measured to computed replicate our hardware used al modeling out andand experimental Iftowe1 assign uncertainA feed through connector for the patient lead was modeledlumped to rep- experimental computational volt into of thethe of the DUT.measurements. This results demonstrated thepeak validity replicate our hardware model. We used various lumped voltages element input impedances with our computational model. voltages out of the DUT. This demonstrated the validity of the ties to both computations and measurements equally, a variation ofor less licate our hardware model. We used various lumped element input dipole at 425 MHz and at 450 MHz. Then we obtained the computational model. Agreement by a factor of 1.38 is Figure 5b. Computational model of radiated test setup. element input impedances with our computational Pacemakers implantable defibrillators have and an model. input received computational model. Agreement by a factor of 1.38 or less is ±19% results. impedances and with our computational model. Pacemakers voltage of the implant. The results are shown in acceptable due to the wide range of uncertainties in both Pacemakers and implantable defibrillators an to input acceptable due to the wide range of uncertainties in both A impedance feed through connector for the patient lead washave modeled for the sensing lead to the device's case (ground) implantable defibrillators have an input impedance for the sensing Table I. The tablemodeling shows the of measured to computed If computational andratio experimental measurements. impedance for sensing to of the device's case (ground) replicate our hardware model. We used various with values on the the case order oflead tens thousands of lumped ohms in Table computational measurements. lead to the device's (ground) with of values on the order of tensin voltages I. assign Validation ofmodeling computational model – comparison with meawe uncertainties toexperimental both computations andIf out of the DUT. Thisand demonstrated the validity of the with values on the order of tens thousands of ohms parallelinput with impedances capacitances with of 0 our to over 10 nF. The higher computational element computational model. we assign uncertainties to both computations and measurements equally, a variation of ±19% results. model. Agreement by a factor of 1.38 or less is of thousands of ohms in parallel with capacitances of 0 to over 10 surements parallelrepresent with ofdefibrillators 0filter to over 10 nF.usually Theinput higher values a feed through capacitor used acceptable Pacemakers and capacitances implantable have an measurements equally, a variation of ±19% results. in both due to the wide range of uncertainties nF. The higher values represent a feed through filter capacitor values represent a feed through filter capacitor used for electromagnetic interference (EMI) protection. We used impedance for the sensing lead to the device's case usually (ground) Table I. Validation of computational model with measurements computational modeling and experimental measurements. usually used for electromagnetic interference (EMI) protection. Height–ofcomparison Measured vs. If Frequency Depth of case for electromagnetic interference (EMI) protection. We with values on thefrom order50Ω of to tens of ohms I. Validation Depth of computational – comparison with measurements input resistances 10 of kΩthousands and capacitances of in 0used to we Table Frequency of case to model Height ofcomputations Measured vs. assign uncertainties both and We used input resistances from 50Ω to 10 kΩ and capacitances of dipole ( mm ) Computed (MHz ) below saline input resistances 50Ω 10 over kΩ and capacitances of 0 to Frequency Depthsaline of case Height Measured ( MHz ) below dipole of Computedvs. parallel capacitances of to 0resistance to 10 nF.used Thetohigher 10 nF.with The 50 from Ω input was enable a( mm variation of ±19% 0 tonF. 10 nF. The The Ω input resistance was used to enable compariRatio / (Computed dB ) surface ) ( MHz ) equally, below saline dipole results. 10 50 Ωthrough input resistance was used hardware. to enable measurements surface comparison of a50 results with our 50 ohm coaxial values represent feed filter capacitor usually used surface son of results with our 50 ohm coaxial hardware. Higher reactanc( mm ) ( mm ) Ratio / ( dB ) comparison of results with our 50 protection. ohm hardware. reactances or resistances would createcoaxial an RFWe mismatch forHigher electromagnetic interference (EMI) used Table I. Validation of computational – comparison measurements ( mm (15 mm ) with Ratio / ( dB ) 5 ) model / 2.47dB esresistances or resistances would create RF mismatch that would make it 425 425 5 15 1.33 /1.33 2.47dB Higher reactances resistances would create anpower RF of mismatch that would make itor50Ω impossible to measure net input from to 10ankΩ and capacitances 0totothe Frequency Depth of 5case Height20 of Measured vs. dB 425 5 15 1.33 / 2.47dB 425 0.93 / -0.63 that would make it impossible to measure net power to the impossible to measure net power to the diode detector on the ( MHz425 ) below saline dipole20 Computed theinput experimental device. software voltage 10diode nF. detector The 50onΩ resistance was Aused to enable 0.93/ /-0.63 -0.63dB dB 425 55 20 0.93 diode detector on theAwith experimental device. A voltage experimental device. software voltage probe was connected 5 surface5 20 0.93 / 0.93 -0.63 dB comparison of results our the 50 lead ohm coaxial hardware. probe was connected between and thesoftware ground (case) 425 425 425 20 -0.63dBdB 10 15 1.38 //2.79 ( mm )10 ( mm )15 Ratio / ( /dB ) dB wasthewith connected between the lead andan the ground (case) between lead and the ground (case) in parallel with thereceived input 425 1.38 2.79 Higher reactances orthe resistances would create RF mismatch inprobe parallel input impedance to measure 0.95 / -0.44 dB 425450 5 10 15 20 1.33 / 2.47dB in parallel with the input impedance to measure received impedance to measure received voltage. 425 5 20 0.93 / -0.63 dB voltage. that would make it impossible to measure net power to the / -0.44 425 450 5 10 20 20 0.930.95 / -0.63 dB dB voltage. The computational model useddevice. throughout this study was diode detector on the experimental A software voltage 5 20 0.93 / -0.63 dB The model used throughout thiswas study was 425425 The computational used made 10 10 15 15 1.38 / 2.79 dB dB made tocomputational match themodel experimental dimensions. The probe was connected between thethroughout leadmodel’s andthis thestudy ground (case) 425 1.38 / 2.79 made to with match experimental model’s dimensions. The model’s sizeimpedance was asdimensions. small possible to house in experimental parallel thetheinput to as measure received to match the experimental model’s The experimental 10 20 0.95 / -0.44 dB experimental model’s as small as the possible to house 450450 voltage. the components inside it.size model’s size was as small as was possible to house components 10 20 0.95 / -0.44 dB the components inside it. Wecomputational developed a model more realistic model of this a pacemaker The used throughout study was or inside it. We developed aexperimental more model dimensions. of a pacemaker neurostimulator for realistic computational simulations (thinor made to match thecase model’s The neurostimulator case for computational simulations (thin computational This had a reduced height experimental model’s wasmodel asmodel small as possible to house We developed amodel). moresize realistic of a pacemaker or neurostimulacomputational model). This model had a reduced height ofinside 9 mm versus 30.5 mm for the original the(thickness) components it. simulations tor case for computational (thin computational model). This (thickness) ofand mm realistic versus 30.5 mm for the original We developed a 9more model of aThis pacemaker computational physical models (fig. 6).versus small model had a reduced height (thickness) of 9 mm 30.5 mm height foror and physical models cardiac (fig. simulations 6). Thisneurological small(thin height neurostimulator case for and iscomputational fairly representative ofcomputational modern the original computational physical models (fig.and 6). This small is fairly We representative of received modern had cardiac andthis neurological implants. calculated the voltage of unit with computational model). This model a reduced height height is fairly representative of modern cardiac and neurological implants. Wedepth calculated the received voltage of this unit with an immersion of 5 mm. results are presented later. (thickness) of 9 mm versus 30.5 mm for the original We calculated received voltage of this unit with an animplants. immersion ofthe 5 mm. results later. computational anddepth physical models (fig.are 6).presented This small height immersion depth of 5 mm. results are presented later. is fairly representative of modern cardiac and neurological implants. We calculated the received voltage of this unit with Figure 7. Computational circuit model for injection testing. Figure 7. Computational circuit model for injection testing. an immersion depth of 5 mm. results are presented later. There are many sources of uncertainties for both There are modeling many sources uncertainties for both computational and for of experimental measurements. computational modeling and for experimental measurements. For7.computations, errors areforintroduced due to the resolution Figure Computational circuit model injection testing. Figure 7. Computational circuit for injection testing. computations, are introduced the resolution ofForthe FDTD grid. errors The model resulting voxelsdue of to objects are not of the FDTD grid. The resulting voxels of objects are not exactlyareequal in shape or sizeofto the objects being There many sources uncertainties for modeled. both There are many sources ofmeasurements, uncertainties for the both computational modelexactly equal in shape or for sizeexperimental to objects being modeled. computational modeling and measurements. For experimental there are errors and ing and forexperimental experimental measurements. Forvoltage computations, are and Figure 6. Realistic size pacemaker model. For measurements, thereto areerrors errors For computations, errors are introduced due uncertainties in the calibrations of thatthe areresolution measured. uncertainties in the calibrations ofofgrid. voltage that arevoxels measured. dueare togrid. the resolution of values the FDTD The Figure 6. Realistic size pacemaker model. Also there errors in resulting the received voltages due ofintroduced the FDTD The voxels of resulting objects are not to Figure 6. Realistic size pacemaker model. Also there are errors in the values of received voltages due to of objects are not exactly equal in shape or size to the objects being 2) Injected Testing. The ISO standard’s injection test specification exactly residual non-linearities the to detector. Finally, there are errors equal in shape orinsize the objects being modeled. residual non-linearities in the detector. Finally, there are errors modeled. For experimental measurements, there are errors and unceruses a coaxial setup as shown in fig. 7. It requires monitoring only For experimental measurements, there are errors and tainties in the calibrations of voltage that are measured. Also there are forward power with a directional coupler. For the analysis of uncertainties in the calibrations of voltage that are measured. Figureinjected 6. Realistic size pacemaker model. errorsthere in theare values of received to residualvoltages non-linearities immunity testing we used Multisim simulation and circuit Also errors in the voltages values due of received due to in the detector. Finally, there are errors due to dimensional andare positiondesign software (National Instruments Corporation, Austin, TX residual non-linearities in the detector. Finally, there errors 78759-3504, U.S.).

al uncertainties in objects within the measurement setup.

©2016 IEEE Electromagnetic Compatibility Magazine – Volume 5 – Quarter 4

67

Table IIa. Pacemaker input impedance vs. received voltage for radiated and injected testing with 120 mW into dipole R(Ω)

C ( nF )

Rad recvd1 Volt ( Vp-p )

Rad recvd Cur (Ap-p )

Inject recvd2 Volt ( Vp-p )

Inject recvd Cur (Ap-p )

Ratio rad to inject (V)

Ratio rad to inject (A)

50

0

0.581

0.011

6.96

0.14

0.08

0.08

50

10

0.001

0.020

0.010

0.28

0.10

0.07

10,000

0

1.39

0.000

13.8

0.001

0.10

0.00

10,000

10

0.001

0.020

0.010

0.28

0.10

0.07

Notes: * for 120 mW into dipole, ** for injected (7 Vp-p 50 Ω load, 14 Vp-p 10 kΩ load), Frequency = 425 MHz, Lead configuration = spiral, Depth of case below saline, Surface = 5 mm, Lead depth = 8.6 mm. 1 - Radiated received, 2- Injected received

Table IIb. Pacemaker input impedance vs. received voltage for radiated and injected testing for the optional 8 Watt radiated test conditions R(Ω)

C (nF )

Rad recvd1 Volt* (Vp-p )

Rad recvd Cur* (Ap-p )

Inject recvd2 Volt** (Vp-p )

Inject recvd Cur** (Ap-p )

Ratio rad to inject (V )

Ratio rad to inject (A )

50

0

4.74

0.092

6.96

0.14

0.68

0.66

50

10

0.0056

0.16

0.010

0.28

0.55

0.58

10k

0

11.38

0.001

13.8

0.001

0.82

0.81

10k

10

0.005

0.16

0.010

0.28

0.55

0.58

Notes: * for 8 W into dipole, ** for injected (7 Vp-p 50 Ω load, 14 Vp-p 10 kΩ load), Frequency = 425 MHz, Lead configuration = spiral, Depth of case below saline, Surface = 5 mm, Lead depth = 8.6 mm. 1 - Radiated received, 2- Injected received

B. Effects of variations of specific parameters on received voltage Once the computational model was validated, we used data from this model to study the effects of slight deviations of the test setup dimensions on the results of radiated testing that a user of the standard might experience. We studied the effects of variations of a number of parameters on the voltage received by our computational model. This was done to identify the range of uncertainties that could be encountered when the measurements prescribed by the 14117 standard are performed. These measurements were performed with various placements of the implant and the leads. The resulting uncertainties were studied. 1) Effects of depth of the implant case below and dipole height above

68

saline surface We studied the effects of the depth of the case below the saline surface on received voltage. For a depth of 5 mm vs.10 mm the ratio of the received voltage was 1.12. We concluded that this demonstrates relative insensitivity to changes in the depth of the case below the saline surface. We studied the effects of the height of the dipole above the saline surface on received voltage. The results were that for a dipole height of 15 mm vs. 20 mm the ratio of the received voltage was 1.04. This was for the depth of case below saline surface of 5 mm. We concluded that this demonstrates insensitivity to changes in the height of the dipole above the saline surface. 2) Effects of lead length on the received voltage. We studied the effects of different lead configurations on the received voltage of the implant using computational modeling. A short straight lead (4 cm long) was compared to a 58.5 cm long lead in the spiral configuration. The depth of the case below saline surface was 5 mm. The computed received voltage for the standard spiral lead was actually lower (by 5%) than the voltage received for the short lead. The ISO 14117 standard states that in vitro test studies have shown that the primary RF coupling to the implant at these frequencies is through the device connector and therefore the layout of the lead is not critical at these test frequencies. 3) Effects of frequency on received voltage. For the DUT box 5mm below the surface of the saline, with a spiral lead configuration and a dipole 2 cm above the saline surface the received voltage at 450 MHz is larger than 425 MHz voltage by a factor of 1.10. 4) Effects of implant input impedance on received voltage and current for radiated and injected testing. We studied the effects of implant input impedance on received voltage and current. Pacemakers use much higher values of input resistance than 50 Ω. In the experimental study we only used a 50 Ω input impedance. We did not use other values of resistors and capacitors since they would make it impossible to measure net power to the diode detector. In our computational modeling of radiated testing we used input resistances of 10 kΩ and 50 Ω. These were in parallel with capacitances of 0 and 10 nF, representing feed through filter capacitors used by most modern implanted active cardiac medical devices. For radiated testing the source voltage was scaled to represent 120 mW and 8 Watts delivered to a dipole antenna. These powers are prescribed by the 14117 standard for the basic (120 mW) and worst-case (optional 8 W) radiated testing. We used a spiral lead configuration with a 5 mm immersion of the top of the implant case below the saline surface. The height of the dipole above the saline surface was 2 cm (as prescribed by the standard). Results are shown in Table II. We used the received voltage results of our computational model as input data for calculating current in the load current using a simple lumped impedance model in Multisim software. For our modeling of the injected test system of the ISO 141117 standard, we used the same DUT input resistances and capacitances as in the radiated computations. We injected 7 Vp-p into a DUT with a 50 Ω input impedance and 14 Vp-p into a DUT with an

©2016 IEEE Electromagnetic Compatibility Magazine – Volume 5 – Quarter 4

input impedance of 10 kΩ (approximately open circuit). These volt- current induced at the input of the implant are indicated. These are for various load impedances and 120 mW into the dipole for ages are prescribed by the standard. The results in Table II are radiated testing. The ratio of the radiated to injected voltages presented for both injected and radiated calculations. We ranges from 0.55 to 0.82. observed the expected dramatic decrease in induced voltage into > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 6 cardiac devices with large capacitors. In general, implanted neurostimulators and other active implants other than cardiac devic- For radiated testing, comparison of measured vs. computed tests active othercapacitors. than cardiac devices often induced do not use is that of the this difference uncertainties both the es often implants do not use filter Therefore, the voltage two methods differinclude by ±1.4 dB or less. Causesfor of this filter capacitors. Therefore, voltage induced into these difference computational and experimental The uncertainties in include uncertainties for bothmodels. the computational and into these devices is relatively high the compared to those cardiac devicdevices is relatively high compared to those cardiac devices experimental the computational results include inthe of the FDTD models. The uncertainties theresolution computational es that incorporate filters. that incorporate filters. grid causing loss of voxel resolution, and inexact shapes or results include the resolution of the FDTD grid causing loss of sizes of small features of the objects being modeled. Also voxel resolution, and inexact shapes or sizes of small features of Table III. Effects on the received voltage of the implant by lateral disTable III. Effects on the received voltage of the implant by lateral there are unknown factors in the proprietary software. the objects being modeled. Also there are unknown factors in placements of the pacemaker displacements of the pacemaker Uncertainties for experimental measurements include lack of the proprietary software. Uncertainties for experimental meaperfect compensation for non-linear behavior of our detector, Y XX YY Relative surements include lack of perfect compensation for non-linear Relative Received positioning errors, and calibration of RF equipment. ((cm) cm ) (cm) ( cm ) Received of our positioningabove errors, the and computational calibration of RF model Voltage Based ondetector, the comparison X behavior voltage equipment. 0 0 1 was validated. This enabled the use of the computational 0 2

0 -2

1 0.8 0.8 0.9

model to simulate a wide range of parameters relatively easily,

Based on therequiring comparison above the computational model val-be the without physical models to be built, as was would idated. This enabled the use of the computational model to simu0.9 case for experimental studies. late2) a wide range of parameters easily, without Evaluation of resultsrelatively with respect to therequiring ISO 14117 physical models to be built, as would be the case for experimental standard. The ISO standard requires radiated testing with 120 5) Effects of various lateral implant locations for radiated 5) Effects Another of various set lateral locationswas for radiated testing. mW delivered to the dipole. With an implant with a 10 kΩ testing. of implant computations performed for the studies. Another of computations was performed2 for DUTthe in multiDUT inset multiple locations, displaced cmthe from center in input impedance and no capacitive filters present, 1.39 Vp-p is induced of atresults the input. For the optional The 8 Watts with respect to thestandard’s ISO 14117 standard. ple locations, displaced 2 cm from the center in the horizontal (X) are 2) Evaluation the horizontal (X) and vertical (Y) directions. Results is produced. The injected delivered to the dipole 11.28 V ISO standard requires radiated testing with 120 mW delivered to and vertical (Y) directions. Results are presented in Table III for p-p presented in Table III for the simulated implant immersed 5 to be delivered to the input of a DUT test prescribes 14 V the dipole. With an implant with a 10 kΩ input impedance and no the simulated implant immersed 5 mm below the saline surface p-p mm below the saline surface exposed to 450 MHz. Variations with a high (open circuit) input impedance at 425 MHz. The capacitive filters present, 1.39 Vp-p is induced at the input. For the exposed to 450 MHz. Variations of 20 % or less were seen for the of 20 % or less were seen for the various 2 cm displacements difference between the radiated and injected test is significant standard’s optional 8 Watts delivered to the dipole 11.28 Vp-p is various 2 cm displacements so it was concluded that small lateral so it was concluded that small lateral displacements of the by a factor of 9.9test or prescribes more. In 14 order thedelivered radiated The injected Vp-pfor to be to test to displacements the implant respect to the weresource not a of produced. implant with of respect to thewith dipole were notdipole a major for a 10atkΩ (high injected 14 Vcircuit) p-p requirement errors.source of errors. the match input ofthe a DUT with atest’s high (open input impedance major input impedance it would take test 12.25 425 impedance) MHz. The difference between theDUT, radiated and injected is Watts into the dipole. For both radiated and injected testing C. Results for the thin computational model significant by a factor of 9.9 or more. In order for the radiated test according to the ISO 14117 standard, the voltage induced in to match the injected test’s 14 Vp-p requirement for a 10 kΩ (high C. Results for the thin computational model the load of the implant is affected significantly by low input We calculated the received voltage of the thin (9 mm thick) impedance) input impedance DUT, it would take 12.25 Watts into impedances such as a feedthrough capacitor. Higher input model at an immersion depth of 5 mm. This was excited by a the dipole. For both radiated and injected testing according to the We calculated the received voltage of the thin (9 mm thick) model at an impedances create higher voltages in the implant and lower dipole antenna mmThis above the saline driven with 20 1V immersion depth of20 5 mm. was excited by a dipole antenna mmpeak ISO 14117 standard, the voltage induced in the load of the implant reactive impedances (larger capacitances) create lower at 425 The with received wasThe0.095 V voltage peak. In is affected significantly by low input impedances such as a above theMHz. saline driven 1 V peakvoltage at 425 MHz. received voltages in the implant. For injected testing this is especially comparison, the original computational model, when exposed feedthrough capacitor. Higher test inputdoes impedances create reflected higher was 0.095 V peak. In comparison, the original computational model, true since the injected not monitor power under the same conditions, received 0.083 V peak. This served voltages in the implant and lower reactive impedances (larger when exposed under the same conditions, received 0.083 V peak. This back from the DUT; therefore, the voltage at the input of an to validate the use of the rather thick experimental and original capacitances) create lower voltages in the implant. For injectedThus the served to validate the use of the rather thick experimental and original unknown DUT cannot be calculated or estimated. computational models (30.5 mm thick) versus real-world testing this is especially true since the injected test does not monicomputational models (30.5 mm thick) versus real-world implants. The implants. The slight increase in received voltage for the thin effects of load impedances other than 50Ω are not monitored. tor reflected power back from the DUT; therefore, the voltage at slight increase in received voltage for the thin model is due to the fact model is due to the fact that in making the case thinner, the This is in contrast to most RF measurement practices using DUT cannot be calculated or estimated. that in making the case thinner, the lead was placed higher in the saline the directional couplers. lead was placed higher in the saline than the standard model. input of an unknown Thus the effects of load impedances other than 50Ω are not monithan the standard model. The overall finding was that a thin computaThe overall finding was that a thin computational model tored. This is in contrast to most RF measurement practices using tional model received the same voltage as our original model. The origireceived the same voltage as our original model. The original V. Conclusions directional couplers. nal computational model was made thicker to replicate the experimencomputational model was made thicker to replicate the We developed a computational model of an active implanted tal model. This physical model wasphysical limited to amodel minimum sizelimited that couldto a experimental model. This was be fabricated to house the optical link and RF detector. minimum size that could be fabricated to house the optical cardiac device that was validated with an experimental model 2 -2 -2

-2 2 2

For the specified radiated test conditions, the induced voltages from a dipole antenna for two amodels differ by ±1.4 dBactive or less. We used IV. We these developed computational model of an implanted car-these to IV.Analysis Analysis and Discussion Discussion radiated with injected EMC testing according diaccompare device that was validated with an experimental model at 425 to the of the the specified ISO 14117 standard. At the border A. Comparison Comparisonofofradiated radiatedwith withinjected injectedvoltages voltages andprovisions 450 MHz. For radiated test conditions, the frequency A. of 450 MHzfrom separating the two the radiated and 1) The ratio of the radiated to injected received voltages. induced voltages a dipole antenna for methods these two models difinjected tests do not agree well (by a factor of 9.9 or more) in The ratio of the radiated to injected received voltages was fer by ±1.4 dB or less. We used these to compare radiated with 1) The ratio of the radiated to injected received voltages. The ratio terms of the voltage induced at the input of an implanted evaluated for tovarious impedances of the DUT for with our injected EMC testing according to the provisions of the ISO 14117 of the radiated injectedinput received voltages was evaluated device. This is for the prescribed injected voltage and radiated computational models. of Our in Table various input impedances thefindings DUT withare our presented computational mod- II standard. At the border frequency of 450 MHz separating the two test power for an implant with a 10 kΩ input impedance and where the voltage and current induced inputand of the methods the radiated and injected tests do not agree well (by a els. Our findings are presented in Table II whereatthethe voltage implant are indicated. These are for various load impedances no feedthrough filter capacitors. Injected testing at 425 MHz and 120 mW into the dipole for radiated testing. The ratio of with 14 Vp-p is equivalent to radiated testing at 12.25 Watts ©2016 IEEE Electromagnetic Compatibility Magazine Quarter 4 for radiated testing. into the– Volume dipole5 –antenna The radiated 69 the radiated to injected voltages ranges from 0.55 to 0.82. testing specification for the ISO 14117 standard is 120 mW link and RF detector.

at 425 and 450 MHz. V. Conclusions

factor of 9.9 or more) in terms of the voltage induced at the input of an implanted device. This is for the prescribed injected voltage and radiated test power for an implant with a 10 kΩ input impedance and no feedthrough filter capacitors. Injected testing at 425 MHz with 14 Vp-p is equivalent to radiated testing at 12.25 Watts into the dipole antenna for radiated testing. The radiated testing specification for the ISO 14117 standard is 120 mW into the dipole, with an optional worst-case test of 8 Watts. The radiated and injected tests do agree moderately well if the much higher 8 W optional power for radiated testing is used. When our injected and radiated models included filter capacitors the induced RF voltage was reduced greatly, as expected. Filter capacitors reduce RF input voltage for both radiated and conducted testing and are used in most cardiac devices, but not in all neurostimulators or other implanted devices.

Biographies

In another phase of this study, we found that a short 40 mm lead produces a voltage that was 5% greater than that produced by a long lead. This is due to the fact the majority of the longer lead is distal to the case in saline. This is where electric fields are generated by the dipole are weaker. We also studied the effects of variations of a number of parameters on the voltage received by the simulated implant using our computational model for radiated testing. These were performed with various placements of the implant and the leads. This was done to identify the magnitude of uncertainties that could be encountered when the measurements prescribed by the ISO 14117 standard are performed. For the radiated tests we found that small changes in position of the dipole and the depth of the DUT in the saline do not affect significantly the induced voltage at the input of the DUT. For lateral displacement of the dipole of 2 cm, changes in this voltage were 20% or less. For depth changes from 5 mm to 10 mm changes were 12.5%.

He was Chief Engineer of the Microwave Research Laboratory at the Walter Reed Army Institute of Research from 1985 to 1990. He returned in 1990 to FDA/CDRH to lead the Electromagnetics and Wireless Laboratory. He has published over 84 papers, including over 22 journal papers and three book chapters. He holds four patents on means for measuring, transmitting, or shielding electromagnetic fields.

Our computational model can be used to study other variations in the radiated test method such as far field exposures of the DUT or induced voltages at other test frequencies. It can also be used for other types of active implants such as neurostimulators that use the ISO14117 standard’s radiated test method.

References [1] ISO 14117:2012(E), Active implantable medical devices - Electromagnetic compatibility, EMC test protocols for implantable cardiac pacemakers, implantable cardioverter defibrillators and cardiac resynchronization devices, First edition 2012-07-15, International Organization for Standardization, 2012. [2] ANSI/AAMI/ISO 14708-3:2008/(R) 2011, Implants for surgery - Active implantable medical devices - Part 3: Implantable neurostimulators, American National Standard, Approved 14 July 2009 and reaffirmed 16 November 2011, Association for the Advancement of Medical Instrumentation and American National Standards Institute, Inc. [3] ANSI/AAMI PC69:2000, Active implantable medical devices - Electromagnetic compatibility - EMC test protocols for implantable cardiac pacemakers and implantable cardioverter defibrillators, Developed by Association for the Advancement of Medical Instrumentation , 3 May 2000, American National Standards Institute, Inc. [4] Hayes, D et al, Interference with Cardiac Pacemakers by Cellular Telephones, N Engl. J Med 1997; 336:1473-1479May 22, 1997 DOI: 10.1056/ NEJM199705223362101.

Howard I. Bassen (M’77–SM’78–F’92-LF’11) received his BS in electrical engineering from the University of Maryland, College Park, MD, USA in 1965 and his MS in management of science and technology in 1980 from George Washington University, Washington D.C., USA. From 1972 to 1980 he was an electronics engineer and a supervisory engineer in the U.S. Food and Drug Administration’s (FDA) Center for Devices and Radiological Health (CDRH). He led original research for 15 other scientists, engineers and technicians in the measurement of electromagnetic radiation induced in the human body by radiation emitting electronic products.

Mr. Bassen is a Fellow of the IEEE - Selected by the Engineering in Medicine and Biology Society. He chaired ANSI C95 Subcommittee 1-Techniques, Procedures, and Instrumentation - Non-ionizing Radiation and IEEE Standards Coordinating Committee 34, Subcommittee 2 (Cellular telephone safety dosimetry techniques). He chaired the Committee on Man and Radiation (COMAR) in the IEEE Engineering in Medicine and Biology Society. Gonzalo G. Mendoza was born in Cochabamba, Bolivia. He received the B.S. degree in Biomedical Engineering from the Central University of Cochabamba in 1997. He obtained the M.Sc. degree in Biomedical Engineering from The Catholic University of America (Washington DC, USA) in 2002. From 2000 to 2003 Mr. Mendoza was a Research Assistant and then a Research and Development Biomedical Engineer at The Catholic University of America. Since 2003, he has been a Biomedical Engineer Contractor with the U.S. Food and Drug Administration (FDA), Center for Devices and Radiological Health (CDRH), Office of Science and Engineering Laboratories (OSEL), Silver Spring, MD, US. Mr. Mendoza research interests include electromagnetic compatibility and numerical modeling of medical devices and human subject safety, in which he has authored several journal articles and conference proceedings. Contact: Howard Bassen, U. S. Food and Drug Administration Center for Devices and Radiological Health, Silver Spring, MD, 20903, USA. (e-mail: Howard. [email protected]). Gonzalo Mendoza, CNI Technical Services, Silver Spring, MD, 20903, USA. (e-mail: [email protected]).

70

©2016 IEEE Electromagnetic Compatibility Magazine – Volume 5 – Quarter 4