Passive Imd

download Passive Imd

of 9

Transcript of Passive Imd

  • 8/3/2019 Passive Imd

    1/9

  • 8/3/2019 Passive Imd

    2/9

    arising from the generation of IMproducts. If f1 = 930 Mhz and f2 = 955Mhz, then fIM = 905 Mhz, which fallswithin the uplink (receive) band.

    Figure 1. GSM Band Plan

    Simplex systems, where a singleantenna is used for both transmit andreceive are naturally the mostsusceptible to PIM, however, they arethe most economical. Duplex antennasystems help reduce the risk and crosspolarization of the transmit/receiveantennas increases the isolation evenfurther, but costs naturally increase.

    Causes: As stated earlier, in the

    general sense, non-linear mixingcauses the generation of PIM.Specifically, there are 3 major causesof PIM in passive devices:

    1. Poor contact junctions2. Components made with, or plated

    with materials that exhibit somelevel of hysteresis

    3. ContaminationContact nonlinearities occur when a

    current carrying contact zone becomesseparated. This usually occurs on amicroscopic level and can be due toinsufficient contact pressure, irregularcontact surfaces, oxidation causing ametal/oxide junction, contact impuritiesor corrosion. This small separation of

    contact surfaces can generate avoltage potential barrier, where electrontunneling (known as the diode effect) ormicroscopic arcing may take placeresulting in a nonlinear voltage to

    current ratio.

    The use of ferromagnetic materialssuch as nickel or steel within thecurrent path, especially at high powerlevels, can also generate PIM due tothe nonlinear voltage to current ratioand hysteresis effect of these materials.

    Contaminants such as metal particlesfrom machining operations that touch

    current carrying surfaces can alsocause intermittent nonlinearities,generating PIM.

    Specifying PIM: PIM is normallyspecified in terms of dBm or dBc. dBmis a measure of power relative to 1milliwatt. Zero dBm is 1 milliwatt into a50 ohm load. dBc is a measure of dBbelow a specified carrier level. For

    example, 20 watts, or 43 dBm is atypical input power level specified fortesting passive devices. A normal testrequirement for allowable PIM might be

    110 dBm. This would make thespecification 110-43, or -153 dBc.The typical range seen today is 100dBm to 120 dBm with two +43 dBmcarriers.

    Measuring PIM: Acceptable levels ofPIM are extremely low and thereforedifficult to measure accurately. Ascompared to S parameters, PIM cannotbe simulated or predicted using HFSSor other analytical software. The onlyway to determine if a device generates

  • 8/3/2019 Passive Imd

    3/9

    PIM, and to what level, is to measure it.Even testing to AQL levels isdangerous because PIM is veryunpredictable and can be generated inwhat appear to be perfectly designed

    and constructed devices. 100% testingis suggested.

    The International ElectrotechnicalCommission, Technical Committee 46,Working Group 6 has developed ageneric test procedure for measuringPIM. See Figure 2.Two methods are described: Reflectedand Transmitted.

    The two different frequencies areamplified, then filtered and combined ina diplexer. The composite signal isapplied to the device under test (DUT)

    through a duplexer. For the reflectedmode, a one port device or terminated2 port device is placed on the output ofthe diplexer and the PIM that isgenerated is filtered in the duplexer anddetected by the receiver. In thetransmitted mode, the 2 port DUToutput is connected to a secondduplexer.

    Figure 2. IEC Test Set-up [1]

  • 8/3/2019 Passive Imd

    4/9

    For electrically short devices such asadapters, there will be no measurabledifference in PIM as a function offrequency whether tested in thereflected or transmitted mode.

    However, as the device becomeselectrically long, such as for cableassemblies, the PIM signal at Port 1 inthe reflected mode is a combination ofthe Port 1 response and the phaseshifted response from the IM sources atPort 2. In the transmitted mode, the

    PIM signals always arrive in phase atport 2 in the transmitted mode.Therefore, transmitted measurementscan be made at a fixed pair offrequencies. See Figure 3. In practice,

    system cables are generally used toprevent damage to test port connectorsresulting in more complex models. Thiseffectively eliminates the fixedfrequency measurement as an optionsince the model is only correct for thisone specific configuration.

    Figure 3. Cable Assembly PIM Model [2]

    Further investigation has also shownthat when testing well matched, lowloss devices such as cable assembliesand adapters, harmonics of the inputsignals can be transmitted through to

    the second duplexer. Since theduplexer is not well matched atharmonic frequencies, the signals canreflect back into the DUT. This cangenerate additional PIM where withother orders may contribute to false

    readings. Filtering can help, but asstated earlier, swept, reflectedmeasurements are recommended.

    Sweeping can be accomplished in

    several manners. Several pairs oftransmit frequencies can be chosen togive equal steps through the receiveband. Another common method usedis to fix one of the transmit frequencies,(f1) at the bottom of the band and

  • 8/3/2019 Passive Imd

    5/9

    sweep the other frequency, (f2) down.Then reverse the operation by fixing f2at the top of the band and sweeping f1up.

    It is essential that a low IM load beused when testing 2 port devices. Withcurrently available test equipment, thespectrum analyzer or receiver noisefloor should be approximately 140dBm. System residual PIM of the testset components should be between-125 and -135 dBm. A poor load willmake it impossible to measure low PIMdevices. The most common type ofload is a cable load consisting of semi-

    rigid cable soldered to a high qualityconnector. Typically, cable generatesextremely low PIM, and carefulconnector design, as will be discussedlater, can result in a termination thatgenerates PIM at levels less than 125dBm.

    Commercial turnkey systems can bepurchased for the different frequencybands. See Figure 4. In addition, test

    systems can be built using individualcomponents as described in the IECtest procedure. See Figure 5.

    Figure 4. Commercial TurnkeySystem [3]

    Figure 5. System built usingdiscrete components

    Minimizing PIM in Practice: Thereare six facets of connector design thatmust be evaluated when consideringlow PIM design. They are: contactdesign, connector mating interface,connector internal junctions, cableattachment, materials and plating.

    Contact design: Contact design andconstruction is one of the mostimportant, yet often overlooked aspectsin suppressing PIM generation.Contact junctions must be designed toprovide high contact pressures at thepoint of desired current flow. The jointsmust be capable of performing over thelife of the system especially duringdynamic flexing and vibrations.

    There are typically five different types ofcontact junction designs used in joiningtwo conductors together. Theseinclude solder, butt contact, springfingers, clamp and crimp. Theconnector itself can be divided intothree general regions in order to identify

  • 8/3/2019 Passive Imd

    6/9

    the region of contact junctions. Theseinclude the connector mating face,connector internal junctions, and theconnector to cable attachment.

    Connector Mating Interface: The twomost common interfaces used for lowPIM applications are the Type N andDIN 7/16. The Type N is very popularin North America and is used commonlyamong radio manufacturers, while theDIN 7/16 (DIN standing for DeutscheIndustrie Normen) originated in Europeand has become very popular on digitalcellular systems due to its low PIM andhigh power handling capabilities. The

    7/16 interface is very robust andprovides a large contact surface areaand mating force resulting in lowcontact resistance. Although not incommon usage, TNC and even SMAhave seen some usage in low power orreceiver applications.

    Connector Internal Junctions: Acommon axiom states that more isbetter. When designing low PIM

    connectors, less is better, at least whenit comes to internal junctions. One-piece center contacts or solder joints, ifone-piece designs are impractical, arebest. One-piece bodies are alsodesirable. Threads and press-fitconfigurations can lead to problems.

    Cable Attachment: Mechanical stabilityof the cable/connector junction is ofutmost importance. Small movements

    caused by flexing can be translated intosignificant PIM. Center conductorsshould be soldered, not crimped.Braided cables are especiallysusceptible to PIM. The cable of choiceby far is a copper jacketed cable that iseither seamless, such as semi-rigid

    types, or welded, corrugated annularand helical styles. Once again,soldering the outer conductor is themethod most likely to prevail in the longrun. Cables with a diameter of 1/2 and

    larger are fairly rigid and flex less thansmaller diameter cables. Therefore inapplications using large diametercables, mechanical clamping to theouter conductor may be tolerated.

    Materials: Ferromagnetic materialssuch as nickel or steel must beeliminated from the current path due totheir non-linear characteristics. Brassand copper alloys are generally

    accepted as linear materials. Testshave shown that nickel plate under goldon the center contact will typically resultin a 40 to 50 dB increase in PIM.Stainless Steel in the body will usuallygive a 10-20 dB increase in PIM.Plating: Connector components areplated for electrical and environmentalpurposes. Silver or white bronze isused for its high conductivity in order tosignificantly reduce contact junction

    resistance. Gold does not oxidize in airand this makes it popular for use inspring contact junctions where highcontact pressures are not attainable.Typically, Type N and Din 7/16 willhave silver plated contacts. Due to theskin effect, the RF current densitiesonly reside within a few skin depths atthe surface of the conductor. It istherefore important to ensure sufficientplating thickness such that the majority

    of the current travels within the platingmaterial. Approximately 98% of thecurrent density travels within 4.6 skindepths. At 1 Ghz, on a silver platedsurface, the skin depth is .002mm or 78uin [4].

  • 8/3/2019 Passive Imd

    7/9

    Cleanliness: Extreme care (and thiscannot be overemphasized) must betaken to ensure no metal particles orchips of any kind from the assemblyprocess end up inside the assembly.

    This relates to both assembly of theconnector, trimming of the cable andsoldering operations. Excess flux willattract other contaminants. Pieces ofcopper jacket or braid will compromisethe performance and cause intermittentoperation.

    Cable Considerations: The connector,is considered the most likely suspectwhen troubleshooting problem cable

    assemblies. Well made cable willgenerate extremely low levels of PIM,but poorly made cable can, and willcause problems. One area of concernis the weld along the seam in annularand corrugated cables. Another area ofconcern is the adhesion of the copperto the center conductor. Copper cladaluminum is a common centerconductor material. Poor clean roomand assembly practices can lead to

    copper flaking off the aluminum andgenerating significant PIM.

    Other Influences: Mechanical,environmental and electrical factors caninfluence the performance of thedevice. PIM can be generated in anotherwise statically, well performingcable assembly. It is not uncommon toput an assembly on a PIM test set upand see excellent results in a static

    condition. In the field, any mechanicalflexing due to wind and vibration canhave a devastating impact. In order totest for this before the assembly isinstalled, dynamic PIM tests are beingdeveloped by the IEC. Raw cable andfinal cable assemblies will be subjected

    to flexing and rotational movementusing a fixture as shown in Figure 6.

    Figure 6. Cable Flexing Fixture [1]

    Connectors will be assembled on veryshort cable assemblies and subjectedto an impact test from a free fallingweight as shown in Figure 7.Temperature cycling and thermal shockcan also have a significant effect onPIM and will likely be included inindividual customer specifications.

    Figure 7. Connector Test Set-up [1]

    Questions have arisen as to the effectof impedance or VSWR on PIM.Experiments have been performed anddata presented to show that VSWR haslittle or no influence on PIM. SeeFigure 8.

  • 8/3/2019 Passive Imd

    8/9

    Figure 8. PIM vs. Return Loss

    Sometimes, it is desired to be able topredict the performance of a device at a

    higher power level than available. Highpower amplifiers are expensive anddifficult to obtain. The device can betested at a lower power level and theperformance at a higher power levelcan be approximated. In simplebroadband devices terminated into abroadband termination, one wouldexpect the PIM to change by 3 dB forevery one dB change in input power.Experimentation (See Figure 9) has

    shown that this relationship is notalways as expected but 3 isreasonable. Some of the possiblereasons are:

    1. High VSWR at harmonicfrequencies (ie. n*f1 or m*f2)

    2. Non linear behavior ofelectromechanical junctions as theyapproach a breakdown potential

    3. Coefficients of higher order than 3

    that occur at the frequency of thethird order [3,7]

    Figure 9. PIM vs. Input Power Level

    In general, for connectors and cableassemblies, PIM is not frequencydependent. However, there has beensome discussion that testing a cable atdifferent frequencies might yielddifferent results due to the skin depth.For example, the current density at1800 Mhz would be twice as high as at900 Mhz. No experimentation to datehas shown consistent differences due

    to frequency effects. Assuming theinsertion loss of the cable does notchange appreciably and the sources ofPIM do not vary dramatically withfrequency, it is likely that the PIMmeasured at 900 Mhz will be similar tothat measured at 1800 Mhz.

    PIM can be time dependent in that theRF heating effect will vary themechanical junctions inside the

    assembly. In such cases, dependingon the design and assembly conditionof the device, the PIM can eitherincrease or even decrease over time.Well designed and properly madecomponents should not vary with time.

  • 8/3/2019 Passive Imd

    9/9

    The Future: It is a certainty that theairwaves will become more crowded.Additional methods will have to bedeveloped to accommodate theproliferation of signals. Two tone PIM

    testing will likely give way to multi tonetests which are more reflective of realworld conditions. Currently, multi tonetesting is extremely expensive andrarely performed. Additional orders ofPIM, and therefore more sensitive testsystems will also be needed. As moreservices compete for space, testing for2nd, 5th, 7th, 9th and higher orders of PIMbecome essential. High power testingat the level of +48 dBm is being

    discussed. The future of PIM testingwill present numerous challenges.

    References:

    1. RF Connectors, connector cableassemblies and cables-Intermodulation level measurementIEC 62047

    2. Deats and Hartman, Measuring the

    Passive IM Performance of RFCable Assemblies, Microwaves andRF, March 1997

    3. http://www.summitekinstruments.com/4. Theodore Moreno, Microwave

    Transmission Design Data , DoverPublications, 1948

    5. Amphenol Signals, Issue No. 2,February 1997

    6. Amphenol Signals, Issue No. 3,August 1997

    7. Hartmut Gohdes, Impact of PowerVariation on 3rd Order PassiveIntermodulation of Coaxial RF-Cables and Their Connectors,International Wire and CableSymposium Proceedings, 1997

    Biography: David Weinstein receivedhis BSEE from the City College of NewYork in 1975. Over the past 32 years,he has held various laboratory,engineering and management positions

    with several connector companies. Hecurrently has 4 patents and severalpublications relating to connectors.Since 1997 he has been the PrincipalRF Engineer with AmphenolCorporation, Communications andNetwork Products Division, specializingin Passive Intermodulation distortionand the RF analysis of connectors andother devices using HFSS software.