免费科普：传播延迟差(SKEW)+延迟（Delay)
07
传播延迟差(SKEW)
SKEW是指在不同的讯号线上，讯号到达接收端的时间差，也就是Delay的差值.
典型的延时差异测试图形
常见的Delay可分为二种：
一.差分对内延迟差(Inter-pair Skew)，是指输入差分讯号下，同一对线内两导体线之Single-end Delay的差值(相减)；是在TDR上设定Differential讯号，一次直接可以量得.
二.差分对间延迟差(Inter-pair Skew)，是指不同对线间之Differential Delay的差值(相减)，（差分）对间延迟差(Inter-pair Skew)，是指不同对线间之Differential Delay相减；是分2次以上量测再计算得到的.
位(时)差-- SKEW 单位：ps/ns
LVDS 是靠+/-一对讯号线在做讯号传输讯号是成双成对的，若其中一根线较长或其他高频效应使讯号跑得不等速，造成讯号在接收端接收的讯号重组时造成还原错误的现象。
同一对讯号线产生的称Intra Skew
不同对讯号线产生的称Inter Skew
关于SKEW
如果Delay Skew数值越小时，表示讯号传输的时间差越小，线材的传输特性较一致。反之，Delay Skew数值越大时，表示讯号传输的时间差越多，线材的传输特性较不一致,当差分线对之间不平衡时，就会引入共模噪声。控制共模噪声的最好方法是使差分线对的两根线尽量在长度上相等、再走线方式上一致、两根线之间的间距尽量保持一致，从而使两根线之间处于平衡状态。另外，一对差分线中的两根信号线之间的长度不同时，除了会造成skew外，还会在接收端造成抖动，这两点在高频讯号传输的过程中都是要时时注意的，还有一点要注意的是，一个信号线的总长度要尽量避免等于信号波长四分之一的整数倍的情况。
08
延迟和反射产生的原因
虽然可以达到光速的量级，但毕竟是有限量，随着数字系统时钟频率的不断提高，以至于信号在互连线上传播的时延TD（延迟）与时钟周期相比拟，这时延迟噪声不再能被忽略。在设计数字系统时序容限时，延迟是最为重要的一环，好在延迟只需要很简单的计算就可以预测到；而反射产生的主要原因是阻抗不匹配，下面我们将SIMON公司整理研究的原文分享给大家，下面的翻译仅供参考！
To several telecommunications professionals, conceptssuch as 'propagation delay' and 'delay skew' bring to mind painful memories ofhigh school physics class. In reality, the effects of delay and delay skew onsignal transmission are easily explained and understood.
Delay is a property that is known to exist for alltypes of transmission media. The propagation delay is equivalent to the amountof time that passes between when a signal is transmitted and when it isreceived on the other end of a cabling channel. The effect is akin to the delayin time between when lightning strikes and thunder is heard-except thatelectrical signals travel much faster than sound. The actual delay value fortwisted-pair cabling is a function of the nominal velocity of propagation(NVP), length and frequency.
NVP varies according to the dielectric materials usedin the cable and is expressed as a percentage of the speed of light. Forexample, most category 5 polyethylene (FRPE) constructions have NVP ranges from0.65c to 0.70c (where "c" represents thespeed of light ~3 x 108 m/s) whenmeasured on finished cable. Teflon (FEP) cable constructions range from 0.69c to 0.73c, whereas cables made of PVC are in the 0.60c to 0.64c range.
Lower NVP values will contribute to additional delayfor a given length of cable, just as an increase in end-to-end cable lengthwill cause a proportionate increase in the end-to-end delay. As with most othertransmission parameters, delay values are frequency dependent.
When multiple pairs in the same cable exhibitdifferent delay performance, the result is delay skew. Delay skew is determinedby measuring the difference between the pair with the least delay and the pairwith the most delay. Factors that affect delay skew performance include materialselection, such as conductor insulation, and physical design, such asdifferences in twist rates from pair to pair.
Although all twisted-pair cables exhibit delay skewto some degree, cables that are conscientiously designed to allow for variancesin the NVP and pair-to-pair length differences will have acceptable delay skewfor standard-compliant horizontal channel configurations. Some of thecharacteristics that could adversely affect delay skew performance includecables with poorly designed dielectric constructions and those with extremedifferences in pair-to-pair twist rates.
Propagation delay and delay skew performance arespecified by some local area network (LAN) standards for worst case 100 m channel configurations to ensure propersignal transmission. Transmission problems associated with excessive delay anddelay skew include increased jitter and bit error rates. Based on IEEE802-series LAN specifications, a maximum propagation delay of 570 ns/100m at 1 MHz and a maximum delay skew of 45ns/100m up to 100 MHz are under consideration byTIA for category 3, 4 and 5, 4-pair cables. TIA Working Group TR41.8.1 is alsoconsidering development of requirements to assess propagation delay and delayskew for 100 ohm horizontal links and channels that are constructed inaccordance with ANSI/TIA/EIA-568-A. As a result of TIA committee "LetterBallot" TR-41:94-4 (PN-3772) it was decided during the September 1996meeting to issue an "Industry Ballot" on a revised draft prior torelease. Still unresolved is the issue of whether or not the categorydesignations will change (e.g., category 5.1), to reflect differences betweencables that are tested for additional delay/delay skew requirements, and thosethat are not.
Although propagation delay and delay skew arereceiving much attention, it is important to note that the most significantcabling performance issue for most LAN applications remains to be attenuationto crosstalk ratio (ACR). Whereas ACR margins improve signal to noise ratiosand thereby reduce the incidence of bit errors, system performance is not asdirectly affected by cabling channels with significant delay skew margins. Forexample, 15 ns delay skew for a cabling channel will typically not result inany better network performance than 45 ns, for a system designed to tolerate upto 50 ns of delay skew.
For this reason, the use of cables with significantdelay skew margins are more valuable for the insurance they provide againstinstallation practices or other factors that may otherwise push delay skew overthe limit, rather than the promise of better system performance compared to achannel that only meets the system delay skew limits by several nanoseconds.
Because cables that use different dielectricmaterials for different pairs have been found to cause problems with delayskew, there has been recent controversy over the use of mixed dielectricmaterials in cable construction. Terms like "2 by 2" (a cable havingtwo pairs with dielectric material "A" and two pairs with material"B") or "4 by 0" (a cable having all four pairs made fromeither material A, or material B) that are more suggestive of lumber thancable, are sometimes used to describe dielectric construction.
Despite commercial hype that may mislead one tobelieve that only constructions having a single type of dielectric materialwill exhibit acceptable delay skew performance, the fact is that properlydesigned cables having either one dielectric material, or multiple dielectricmaterials are equally capable of satisfying even the most severe channel delayskew requirements specified by applications standards and those underconsideration by the TIA.
Under some conditions, mixed dielectric constructionsmay even be used to offset delay skew differences that result from differenttwist rates. Figures 1 and 2 illustrate representative delay and skew valuesobtained from a randomly selected 100 meter cable sample having a "2 by2" (FRPE/FEP) construction. Note that the maximum propagation delay anddelay skew for this sample are 511 ns/100m and 34 ns, respectively in the frequency rangefrom 1 MHz to 100 MHz
Fig. 1: Cable Propagation Delay
Fig. 2: Cable Delay Skew
Now that the issue of delay skew has been brought tolight, cabling system providers are taking great pains to ensure that theirproducts will not contribute to delay skew problems in the field. Manyinstallers ask: "What about my previously installed cabling?" Basedon the fact that the Siemon Cabling System has only approved qualified cablemanufacturers with the most advanced technical competence and manufacturingtechnology, we are able to provide full assurance that the channel requirementsspecified by all cat. 3, 4 and 5 signaling applications for delay and delayskew are met by all existing and new Siemon Cabling System installationscovered by our 16 year application and component warranty.
Regarding future standards developments on the issuesof delay and delay skew, The Siemon Company and its Cable Partners are fullycommitted to providing telecommunications cabling solutions that meet or exceedapplicable standards requirements.
传输延迟是所有类型的传输媒体都存在的一种特性 。传输延迟等于信号从线缆信道的一端发送在另一端接收所经过的传输时间。这就像先看到闪电后听到雷声这其中的时间差，只不过电信号传播得比声音快得多。对于双绞线来说，实际的传输延迟值是由相速度，线缆长度和频率决定的。
相速度的变化与线缆的绝缘材料有关。我们用光速的百分比来表示。例如：大部分的CAT5结构的速度变化范围是从0.65c到0.70c（c代表光在真空中的传播速度3×108m/s）铁氟龙（FEP）线缆的结构可以达到0.69c到0.73c，而PVC材质只有0.60c到0.64c的范围
对于一段长度一定的线缆来说，低相速度值将会引起额外的延迟。就象增加线缆端到端的长度就会增加相应的延迟。和其它很多传输参数一样，延迟是由频率决定的。
当多线对在同一个线缆显示不同的延迟特性的结果就是延迟差异。延迟差异是由测量线对间的最小延迟和最大延迟的差值来表征的。影响延迟差异特性的因素包括材质的选择，比如导体绝缘，物理设计，比如线对捻入率的差异。
传输延迟和延迟差异是由一些局域网标准制定的标准，测量在最坏情况下100米信道结构特性，来保证信号正确无误地传输。传输延迟和延迟差异增大所带来的传输问题包括抖动和误码率的增加。
基于IEEE 802系列局域网规范，TIA正在考虑四线对的三类，四类和五类网络线在1MHz时的最高传输延迟为570 ns/100m，频率高达100MHz时最大延迟差异为45ns/100m。TIA工作组TR41.8.1也在考虑发展要求，以评估基于ANSI/TIA/EIA-568-A的100欧姆横向链路和信道的传播延迟和延迟差异。由于TIA委员会的TR-41:94-4 (PN-3772)投票，决定在1999年九月发起对之前颁发的修正草案进行工业投票。为了反映要求测试的传输延迟和延迟差异有额外要求的线缆和没有要求的线缆的差别，是否改变类别名称（比如cat5.1），这仍旧是一个悬而未决的问题。
虽然传输延迟和延迟差异正在引起关注，但对大部分LAN应用来说，最重要的关于线缆方面的要求还是衰减串扰比（ACR）。然而衰减串扰比的差值能提高信噪比，这样就能减少误码率的发生，系统性能是不直接受明显的线缆信道延迟差异值影响的。例如，对于一个容许延迟差异高达50ns的系统来说，15ns的延迟差异与45ns的延迟差异相比，线缆信道并不会有好的网络特性
为此，使用传输延迟余量大的线缆对于保证实际加工或者其它因素导致延迟差异超出规定要求更有价值，与仅仅比规定的延迟差异相差几纳秒的线缆相比，余量大的线缆可以提供更好的系统特性。
因为我们发现不同的线对用不同的绝缘材质会出现延迟差异的问题，而今在线缆中使用混合绝缘材料引起了争议，像2对2（线缆的两对线用A绝缘料，另两对用B绝缘料）或4对0（线缆四对线要么全用A绝缘料，要么用B绝缘料）这些术语用来描述绝缘结构。
尽管商业宣传误导我们认为仅仅只用单一的绝缘料才会有可接受的延迟差异特性，而实际上，只要我们设计合理，不管是用一种绝缘料，还是多种绝缘料混合，只要我们的设计满足规范标准和TIA规定的最差的信道延迟要求，就都是可取的。
在一些情况下，混合绝缘料甚至可能用来抵消由于不同捻入率所产生的延迟差异的不同，图1和图2分别表示传输延迟和延迟差异，这是随机测量100米“2对2”结构（FREP/FEP）样品线得出的测量结果，注意到这条样品线在频率从1MHZ到100MHZ的最大传输延迟和延迟差异分别是511ns/100和34ns。
既然我们已经发现延迟差异的问题，综合布线系统供应商正在花大力以确保它们的产品不会在这一领域导致延迟差异问题，许多施工人员问：“以前的施工布线怎么办？”基于这个事实，西蒙综合布线系统只核准拥有最先进的技术能力和生产技术的合格电缆生产商，对于现存的和新的西蒙布线系统设施，规定的三类，四类及五类线的传输延迟和延迟差异要求，我们能够提供充分保证，这涵盖我们的应用。考虑到将来传输延迟和延迟差异的标准会更新发展，西蒙公司和他的合作公司全力致力于提供通信电缆的解决方案来满足或超过标准要求。
END