测控技术与仪器科技英语word范文 12页.docx

测控技术与仪器科技英语word范文 12页.docx

《测控技术与仪器科技英语word范文 12页.docx》由会员分享，可在线阅读，更多相关《测控技术与仪器科技英语word范文 12页.docx（9页珍藏版）》请在冰豆网上搜索。

测控技术与仪器科技英语word范文12页

本文部分内容来自网络整理，本司不为其真实性负责，如有异议或侵权请及时联系，本司将立即删除！

==本文为word格式，下载后可方便编辑和修改！

==

测控技术与仪器科技英语

篇一：

测控技术与仪器科技英语第四课翻译与课文

Unit4DigitalSignalProcessing（DSP）

Havingheardalotaboutdigitalsignalprocessing（DSP）technology,investigatewhyDSPispreferredtoanalogcircuitryformanytypesofoperations,anddiscoverhowtolearnenoughtodesignyourownDSPsystem.Thisarticle,thefirstofaseries,isanopportunitytotakeasubstantialfirststeptowardsfindinganswerstoyourquestion.ThisseriesisanintroductiontoDSPtopicsfromthepointofanalogsystemdesignersseekingadditionaltoolsforhandinganalogsignal.DesignersreadingthisseriescanleanaboutthepossibilitiesofDSPtodealwithanalogsignalsandwheretofindadditionalsourcesofinformationandassistance.

4.1WhatIsDSP?

Inbrief,DSPsareprocessorsormicrocomputerswhosehardware,software,andinstructionsetsareoptimizedhigh-speednumericprocessingapplications-anessentialforprocessingdigitaldatarepresentinganalogsignalsinrealtime.WhataDSPdoesisstraightforward.Whenactingasadigitalfilter,forexample,theDSPreceivesdigitalvaluesbasedonsamplesofasignal,calculatestheresultsofafilterfunctionoperatingonthesevalues,andprovidesdigitalvaluesthatrepresentthefilteroutput;itcanalsoprovidesystemcontrolsignalsbasedonpropertiesofthesevalues.TheDSP’shigh-speedarithmeticandlogicalhardwareisprogrammedtorapidlyexecutealgorithmsmodelingthefiltertransformation.

Thecombinationofdesignelementsaarithmeticoperators,memoryhandling,instructionset,parallelism,dataaddressingthatprovidethisabilityformsthekeydifferencebetweenDSPsandotherkindsofprocessors.Understandingtherelationshipbetweenreal-timesignalandDSPCalculationspeedprovidessomebackgroundonjusthowspecialthiscombinationis.Thereal-timesignalcomestotheDSPasatrainofindividualsamplesfromananalog-to-digitalconverter（ADC）.Todofilteringinreal-time,theDSPmustcompleteallthecalculationsandoperationsrequiredforprocessingeachsamples（usuallyupdatingaprocessinvolvingmanyprevioussamples）beforethenextsamplearrives.Toperformhigh-orderfilteringofreal-worldsignalshavingsignificantfrequencycontentcallsforreallyfastprocessors.

4.2WhyUseaDSP?

TogetanidealofthetypeofcalculationsofDSPdoseandgetanidealofhowananalogcircuitcompareswithaDSPsystem,onecouldcomparethetwosystemsintermsofafilterfunction.Thefamiliaranalogfilterusesresistors,capacitors,inductors,amplifiers.Itcanbecheapandeasytoassemble,butdifficulttocalibrate,modify,andmaintainadifficultythatincreasesexponentiallywithfilterorder.Formanypurposes,onecanmoreeasilydesign,modify,anddependonfiltersusingaDSPbecausethefilterfunctionontheDSPissoftware-based,flexible,andrepeatable.Further,tocreateflexiblyadjustablefilterswithhigher-orderresponserequiresonlysoftwaremodifications,withnoadditionalhardwareunlikepurelyanalogcircuits.Anidealbandpassfilter,withthefrequencyresponseshowninFig.4.1,wouldhavethefollowingcharacteristics:

?

aresponsewithinthepassbandthatiscompletelyflatwithzerophaseshift

?

infiniteattenuationinthestopband.

Usefuladditionswouldinclude:

?

passbandtuningandwidthcontrol

?

Stopbandrolloffcontrol

AsFig.4.1shows,ananalogapproachusingsecond-orderfilterswouldrequirequiteafewstaggeredhigh-Qsections;thedifficultyoftuningandadjustingitcanbe

imagined.

WithDSPsoftware,therearetwobasicapproachestofilterdesign:

finiteimpulseresponse（FIR）andinfiniteimpulseresponse（IIR）.TheFIRfilter'stimeresponsetoanimpulseisthestraightforwardweightedsumofthepresentandafinitenumberofprevious

inputsamples.Havingnofeedback,itsresponsetoagivensampleendswhenthesamplereachesthe"endoftheline"（Fig.4.2）.AnFIRfilter'sfrequencyresponsehasnopoles,onlyzeros.TheIIRfilter,bycomparison,iscalledinfinitebecauseitisarecursivefunction:

itsoutputisaweighedsumofinputsandoutputs.Sinceitisrecursive,itsresponsecancontinueindefinitely.AnIIRfilterfrequencyresponsehasbothpolesandzeros..

Thex（s）aretheinputsamples,y（s）aretheoutputsamples,a（s）areinputsampleweighings,andb（s）aresampleweighings.Nisthepresentsampletime,andMandNarethenumberofsamplesprogrammed（thefilter'sorder）.Notethatthearithmeticoperationsindicatedforbothtypesaresimplysumsandproductsinpotentiallygreatnumber.Infact,multiply-and-addisthecaseformanyDSPalgorithmsthatrepresentmathematicaloperationsofgreatsophisticationandcomplexity.

Approximatinganidealfilterconsistsofapplyingatransferfunctionwithappropriatecoefficientsandahighenoughorder,ornumberoftaps（consideringthetrainofinputsamplesastappeddelayline）.Fig.4.3showstheresponseofa90-tapFIRfiltercomparedwithsharp-cutoffChebyshevfiltersofvariousorders.The90-tapexamplesuggestshowclosethefiltercancometoapproximatinganidealfilter.WithinaDSPsystem,programminga90-tapFIRfilterliketheoneinFig.4.3isnotadifficulttask.Bycomparison,itwouldnobecost-effectivetoattemptthislevelofapproximationwithapurelyanalogcircuit.AnothercrucialpointinfavorofusingaDSPtoapproximatetheidealfillterislong-termstability.WithanFIR（oranIIRhavingsufficient

resolutiontoavoidtruncation-errorbuildup）,theprogrammableDSPachievesthesameresponse,

timeaftertime.Purelyanalogfilterresponsesofhighorderarelessstablewithtime.

MathematicaltransformtheoryandpracticearethecorerequirementforcreatingDSPapplicationandunderstandingtheirlimits.Thisarticleserieswalksthroughafewsignal-analysisand-processingexamplestointroduceDSPconcepts.Theseriesalsoprovidesreferencestotextsforfurtherstudyandidentifiessoftwaretoolsthatcasethedevelopmentofsignal-processingsoftware.

4.3SamplingReal-worldSignals

Real-worldphenomenaareanalogthecontinuouslychangingenergylevelsofphysicalprocesseslikesound,light,heat,electricity,magnetism,Atransducerconvertstheselevelsintomanageableelectricalvoltageandcurrentsignals,andanADCsamplingfrequency,oftheADCiscriticallyimportantindigitalprocessingprocessingofreal-worldsignals.

Thissamplingrateisdeterminedbytheamountofsignalinformationthatisneededforprocessingthesignaladequatelyforagivenapplication.InorderforanADCtoprovideenoughsamplestoaccuratelydescribethereal-worldsignal,thesamplingratemustbeatleasttwicethehighest-frequencycomponentoftheanalogsignal.Forexample,toaccuratelydescribeanaudiosignalcontainingfrequenciesupto20kHz,theADCmustsamplethesignalataminimumof40kHz.Sincearrivingsignalcaneasilycontaincomponentfrequenciesabove20kHz（includingnoise）,theymustberemovedbeforesamplingbyfeedingthesignalthroughalow-passfilter,isintendtoremovethefrequenciesabove20kHzthatcouldcorrupttheconvertedsignal.

However,theanti-aliasingfilterhasafinitefrequencyrolloff,soadditionalbandwidthmustbeprovidedforthefilter'stransitionband.Forexample,withaninput

signalbandwidthof20kHz,onemightallow2to4kHzofextrabandwidth.

Figure4.4depictsthefilterneededtorejectanysignalswithfrequenciesabovehalfofa48kHzsamplingrate.

Secondsample.ThetimebetweensamplesisthetimebudgetfortheDSPtopreformallprocessingtasks.Fortheaudioexample,a48kHzsampleratecorrespondstoa20.833vssamplinginterval.Fig.4.5relatesthetheanalogsignalanddigitalsamplerate.

图

NextconsidertherelationbetweenthespeedoftheDSPandcomplexityofthealgorithm（thesoftwarecontainingthetransformorothersetofnumericoperations）.Complexalgorithmrequiremoreprocessingtasks.Becausethetimebetweensamplesisfixed,thehighercomplexitycallsforfasterprocessing.Forexample,supposethatthealgorithmrequires50processingoperationstobeperformedbetweensamples.Usingthepreviousexample's48kHzsamplingrate（20.833vssamplinginterval）,onecancalculatetheminimumrequiredDSPprocessorspeed,inmillionsofoperationspersecond（MOPS）asfollow:

DSPSpeed=Operations50==2.4mopsSamplingInterval20.833vs

数字信号处理器

因为听到了很多关于数字信号处理（DSP）技术,你可能想找出是什么可以用DSP,,并发现如何学习足够你自己设计DSP系统。

本文,系列的第一,是一个机会来承担高额的第一步,

寻找答案到你的问题,。

本系列介绍DSP主题而言,从模拟系统设计师寻求额外的工具将模拟信号.设计者可以依赖阅读本系列文章对可能性的DSP处理模拟信号和在哪里能够找到更多的信息来源和援助。

4.1什么是数字信号处理器

总之,DSP处理器或者是微机的硬件、软件和指令集的优化高速数字处理必不可少的,模拟信号处理数字数据代表在真正的时间,什么是简单易懂的DSP,当作为数字,例如,DSP接收数字值基于样本的一个信号,计算结果一个过滤器的功能操作这些值,并提供数字值,代表了过滤器的输出,它还可以提供系统控制信号基于这些值的属性,DSP的高速算术和逻辑的硬件被编程来迅速执行算法建模过滤器转换。

设计元素的组合一个算术操作符,内存处理、指令集、并行性、数据处理提供这种能力形式的关键区别dsp和其他类型的处理器。

了解实时信号之间的关系和DSP计算速度提供了一些背景对这种组合是有多么特别,。

实时信号来DSP作为个人的样本训练的模数转换器（ADC）.做实时过滤,DSP必须完成所有操作所需的计算和处理每个样本（通常是更新过程涉及许多之前的样本）之前到达下一个示例。

执行高阶过滤现实世界的信号频率有重要内容要求真正的快速处理器。

4.2为什么使用一个数传信号处理器？

得到数传信号处理器剂量的计算的类型的理想而且得到一个类比线路如何与一个数传信号处理器系统相较的理想，一会在一个过滤器功能方面比较这两个系统。

熟悉类比过滤器使用电阻、电容器、授职者，喇叭筒。

集合能是廉宜、容易的，但是困难的校正，修正，而且维持一种指数地以过滤器次序增加的困难。

对于许多目的，因为在数传信号处理器上的过滤器功能是以软件为基础的、有柔性、和可重复的，所以一个罐子更容易设计，修正，而且取决于使用一个数传信号处理器的过滤器。

更进一步，用比较高次序回应产生易曲可调整过滤器s需要唯一的软件修正，藉由没有另外硬件不像纯粹地类比线路。

一个藉由在Fig.4.1被显示的频率回应的理想通带过滤器会有下列的特性：

在通带里面的一个回应哪一是完全平坦的与零状态变化

在阻带中的无限变薄。

有用的附加会包括：

通带调音和宽度控制

阻带滚边控制

作为Fig.4.1表，使用第二次序的过滤器的类比方法会需要相当多的蹒跚高Q区段；调音的困难而且调整它能被想像。

由数传信号处理器软件，有两基本方法要过滤设计：

有限的冲动回应（枞树）和无限推动回应（IIR）。

对一种冲动的枞树过滤器的时间回应是礼物的笔直的重量总数和有限数字早先的输入抽取样品。

没有回应，当样品达成"线的结束"时，它的回应对一个给定的样品结束。

（图4.2）枞树过滤器的频率回应没有杆，只有零。

因为它是一个回归的功能，所以被比较的IIR过滤器叫做无限。

因为它是回归的，它的回应能不确定继续。

IIR过滤器频率回应有杆和零。

。

x（s）是输入样品，y（s）是输出样