OmniBER 718communicationsperformance analyzer

Product Note

Protection-switching timemeasurements:Understanding the OmniBER analyzer’smeasurement technique and thealternative test methods

2

The OmniBER analyzer’s servicedisruption measurement is designed toaccurately measure the time taken by atransmission system to perform anautomatic protection switch when atransmission defect is detected by thesystem. The measurement supportstesting of all protection switchingarchitectures deployed in today’stransmission networks. Result accuracyand reliability is based on a simple testprinciple, coupled with a speciallydesigned error detector that measuresthe duration of error bursts associatedwith a protection switch event.

Contents

Introduction ..................................... 2Test configuration and setup ........ 3Measurement principle ................. 3Service disruption test results ..... 4Understanding the test results ..... 4Case study of protection-switching time measurementson a STM-64 dual-ring ................... 5System configuration ..................... 5OmniBER setup .............................. 5Summary of test procedureand results ...................................... 6Appendix 1: Protectionswitching time test methodsand associated error sources ....... 7Protection switching timemeasurements - the alternativesolution ............................................ 8(1) Service disruption timewith simulated LOS failure(minimize detection time) ............. 8(2) Service disruption time withSF trigger by excessive errors(eliminate detection time) ..........10Summary and conclusions ..........11

Introduction

3

Test configuration and setup

Figure 1: Test configuration for measuring

protection-switching time

As shown in figure 1, the OmniBERanalyzer measures protection-switchingtime from the tributary-side of thetransmission system under test. Thispoint is significant for two reasons:

1. The measurement is totallyindependent of the protectionswitching architecture and alloptional settings – it thereforesupports all configurations

2. The performance of the systemunder test cannot be affected by thetest set since results are obtainedthrough passively monitoring a PRBSfor errors

The measurement is performed byinserting a PRBS test pattern into atributary feeding the transmissionsystem under test, looping this signal atthe associated drop-side tributary, thenmonitoring the PRBS for errors. ThisPRBS can be inserted either as part of aPDH signal or as a mapped servicewithin a SDH signal.

Measurement principle

The protection-switching timemeasurement is an out-of-service testthat is typically performed duringverification, installation and commissioningof new transmission systems. As such,the measurement principle assumes thatthe system under test will be operatingerror-free before the test is performed.This assumption is important since thepresence of background errors can affect

test results (see Understanding the TestResults for details).

The measurement is performed by, first,connecting the OmniBER analyzer aspreviously described and verifying error-free reception of the PRBS test pattern.Then, invoking a protection switch on aworking section of the transmissionsystem that is transporting the PRBS.Typical methods used to trigger thisprotection switch are disconnecting afibre1, 2 to simulate a fibre-break, orremoving power from a transmissionelement to simulate a node-failure. Aswill be shown in the following STM-64Ring case study, there is value in verifyingthe system’s performance when testedseparately using a simulated fibre-breakand a simulated node-failure.

1 Exercise extreme caution whendisconnecting an optical fibre –follow your organization’s standardsafety procedures.

2. Refer to Appendix 1 for a discussionon the issues associated withgenerating a loss-of-signal bydisconnecting a fibre.

Irrespective of how the protectionswitch is triggered, it results in thePRBS test pattern being corrupted for ashort period.

As shown in figure 2, the duration of thiscorruption is controlled by three factors:

1. The system’s fault detection time2. The protection-switching time3. The time taken by the OmniBER

analyzer to re-align to the pointers(SDH tributary only) and test pattern

Since the objective is to measure atransmission system’s protection-switching time (specified as less than 50ms in ITU-T standards), it is essential thatthe contributions made by these additionalfactors are minimized. For fault detectiontime, this is achieved by triggering theprotection switch using a failure thatresults in a LOS defect. Although ITU-TG.783 (2000) defines LOS detection time asbeing “in the province of regionalstandards”, it provides an example basedon a value of less than 100 µs (less than0.2% of the maximum acceptableprotection-switching time). In the case ofpointer and pattern acquisition, the requiredtimes are as shown in figure 2. Thesevalues represent the following percentageerrors for PDH and SDH tributaries (with140M & 2M mappings) when related to themaximum acceptable switching time:

PDH: +0.1%SDH (140M): +0.35% to +0.85%SDH (2M): +1.35% to 3.85%

Figure 2: Contributors to the protection switching time as measured by OmniBER

4

When measuring a system’s protection-switching time, the above discussionshows that the total systematic errorassociated with the OmniBER analyzer’sservice disruption measurement can berestricted to between +0.3% to +4.05% ofthe maximum acceptable switching time.Consequently, it can be relied on toaccurately evaluate this important systemspecification.

Service disruption test results

Three separate results are provided by theOmniBER analyzer’s existing servicedisruption measurement:

Longest: The duration of the longest errorburst detected during the testShortest: The duration of the shortesterror burst detected during the testLast: The duration of the most recenterror burst detected during the test

These result fields are reset to 0 ms atthe beginning of a protection-switchingtime test by pressing the instrument’sSTART button. When the protectionswitch is triggered, the duration of theresulting error burst is measured anddisplayed. For the system under test topass, a single1 error burst of duration lessthan 50 ms should be detected. Detectionof a single error burst is indicated by anidentical value being displayed in thethree result fields.

Why are three separate results provided?During a detailed investigation into themeasurement requirements, it wasfound that some transmission systemsexhibited a characteristic similar toswitch-bounce during a protection-switching event. This results in multipledistinct error bursts being present on thereceived test pattern. By providing threeseparate results, the OmniBER analyzer’sservice disruption measurement clearlyidentifies the presence of this unwantedoperating characteristic.

New developments in the OmniBER 718analyzer have been designed to makeanalysis of a protection-switch fasterand easier, including:

Timestamping: Displaying a relativetimestamp of the beginning of eachservice disruption event.

History: A record of the first ten servicedisruption measurements.

While an Alarm Indication Signal (AIS),is not a pure protection-switchingmeasurement, it is closely related. AIS isactivated as a result of any physical layerfailure, such as a fibre break. TheOmniBER 718 can now provide details ofAIS duration measurements,timestamping and history. When used inconjunction with the OmniBER analyzer’sservice disruption measurements, itbecomes possible to show a relationshipbetween alarms within a device andautomatic protection switches in anetwork. This means you can quicklyand accurately debug network elements.For example, using the timestampingand AIS duration measurements,makes it easy to see if a device fails dueto AIS not being raised quickly enough, ortaking too long to be removed after theswitch has taken place.

Understanding the test results

As is true for any measurement, having aworking understanding of how themeasured values are derived is helpfulwhen attempting to identify the root-cause of unexpected results. In the caseof the OmniBER analyzer’s servicedisruption measurement this meansunderstanding the rules associated withthe analysis of error-burst duration.

The OmniBER analyzer’s servicedisruption test measures the elapsedtime between the first and last error inan error-burst that consists of two ormore errors. The error-burst is taken ashaving ended when no errors are detectedduring a period of greater than 200 to 300ms following the last error. Single errorsthat are separated by more than 200 to300 ms are not considered as being partof an error-burst (no result is returned).

Figure 3 illustrates the affect thesesimple rules have on measurementresults when different error distributionsare present in the received test pattern.In case 1 and case 2, there are singleerrors due to a low background error ratein the transmission system, plus anerror-burst associated with a protection-switching event.

Figure 3: The OmniBER analyzer’s error burst analysis

5

However, in case 1 the measuredprotection-switching time is not affectedby the background errors since theseoccur outside the 200 to 300 ms periodused to define the end of the error-burst.In contrast, the result obtained in case 2is affected due to a single backgrounderror being present less than 200 msbefore the error-burst actually starts.This leads to an artificially highprotection-switching time being reported,and consequently emphasizes theimportance of ensuring that the systemunder test is error-free before performingthe measurement.

Cases 3 and 4 are based on a scenariowhere the system under test generatestwo error-bursts when performing aprotection switch. The point beingillustrated here is that the resultsobtained will be affected by theseparation of these two error-bursts.In case 3 a result for each error-burstwill be reported (since they are more than300 ms apart), while in case 4 onlya single high value will be reported(since they are less than 200 ms apart).However, in both cases the reportedresults will indicate that a problem existsin the system under test.

System configuration

� MSPRING protection switching� Wait-to-Restore period set at 5 minutes� Under normal working conditions,

the OmniBER analyzer’s test patternis transmitted via the links shown assolid lines on the diagram

� The two rings are connected viaSTM-16 tributaries ports

OmniBER setup

� Connected at an STM-16 tributary port� Unframed PRBS test pattern

inserted as a mapped 140 Mb/ssignal within a selected VC-4 channel

� The OmniBER analyzer measurementrestarted after each protection-switching event

Case study of protection-switching time measurementson a STM-64 dual-ring

The following case study illustrates howthe OmniBER analyzer is enabling theprotection-switching time performanceof STM-64 systems to be evaluatedbefore deployment in today’s operationalnetworks. This particular exampledocuments results obtained during anevaluation of the dual-ring transmissionsystem shown in figure 4.

Figure 4: STM-64 System Under Test

6

Initially, the wide variation in the resultsobtained from these tests causedquestions to be raised concerning thereliability of the OmniBER’s analyzer’sservice disruption measurement. Whenexamined more closely, however, with anunderstanding of the measurementtechnique (as explained in this ProductNote), it became clear that the analyzerwas accurately measuring the error burstactivity associated with each test. Andin doing so, the OmniBER analyzer’sservice disruption measurement providedthe evaluation team with a detailedknowledge of the actual protection-switching performance supported by thesystem under test.

Summary of test procedure and results

Protection Switch Transmission Direction MeasuredTrigger Event Of Test Pattern (1) Results

Normal operation R1A1� R1A4 � R2A1 � R2A3 -

Node failure (2): R1A1� R1A2 � R1A3 � R2A2 100 ms (4)

Ring #1, ADM 4 � R2A1 � R2A3

Node failure (2): R1A1� R1A2� R1A3 � R2A2 30 msRing#2, ADM 1 � R2A3

Node Recovery (3): R1A1� R1A4� R1A3� R2A2 174 & 18 ms (5)

Ring #1, ADM 4 � R2A3 then 34 ms (6)

Node Recovery (3): R1A1� R1A4� R2A1� R2A3 14 & 11 ms (5)

Ring #2, ADM 1 then 34 ms (6)

Fibre-disconnect: R1A1� R1A2� R1A3� R1A4 30 msBetween ADM 1 & � R2A1� R2A3

4 on Ring#1

Fibre-reconnect: R1A1� R1A4� R2A1� R2A3 30 msBetween ADM 1 &

4 on Ring#1

Notes:1. Indicates the direction of transmission of the test pattern after completion of the

protection switch [Rn = Ring number (1 or 2); An = ADM number (1 to 4 on R1;1 to 3 on R2)]. Go and return paths for the test pattern are the same(i.e. bi-directional protection switching is used)

2. Caused by removing power from the selected ADM3. Power restored to failed node4. Customer accepted this as being acceptable due to need for switching events to

occur on both rings in order to restore error-free transmission5. These results were measured approximately 90 seconds after power was restored

to the ADM – before the wait-to-restore period was complete. On furtherexamination using OmniBER’s graphical measurement results, it was found theseunexpected results were caused by an AU-AIS alarm being output for a short timeperiod at the tributary connected to OmniBER’s receiver. The only possibleexplanation for this is that the ring’s operation is being corrupted during theprocess of the failed node re-establish itself as an active node on the ring.

6. This result was measured 5 minutes after power was restored to the ADM – it istherefore the actual protecting-switching result for this test.

7

Appendix 1:Protection switching time test methods and associated error sources

Protection switching timemeasurement – the problem

Many of today’s high-reliabilitytransmission networks are founded onSDH/SONET technology, which providesbuilt-in fault restoration known asAutomatic Protection Switching (APS).This general term covers a range ofdifferent protection schemes designedfor use in Linear and Ring networktopologies, and includes linear MultiplexSection Protection (MSP), MultiplexSection Protected Rings (MSPRING) andPath Protection. However, regardless of thenetwork topology and specific protectionscheme, the basic principles behind theOmniBER analyzer’s service disruptionmeasurement, and its application inverifying a transmission system’sprotection switching time, remain valid.

In Figure A1, two network nodes (e.g.ADMs) are shown with a single workingcircuit and a single protection circuitbetween them. In linear systems,working and protection circuits may bepaired (1+1 protection as shown), or oneprotection circuit may be shared amongseveral working circuits (1:n protection).

This example shows the state of thenodes after a switch has taken place.The typical sequence of events is:

� The tail-end node detects thefailure and signals the head-end to request a protectionswitch.

� The head-end performs abridge or bridge and switchoperation, and sends back anacknowledgement.

� The tail-end receives theacknowledgement andperforms a bridge and switchoperation, then finishes bysending a status message tothe head-end.

Figure A1

Figure A2 shows the two maincomponents of a service disruptionfollowing any kind of failure:

The first part is the time taken to detecta failure. Protection switching can beinitiated by either of two events:

1. Signal Fail (SF): usually loss ofsignal, loss of framing, or a very higherror ratio such as 1 x 10-3 or greater.

2. Signal Degrade (SD): a persistentbackground error rate that exceeds aprovisioned threshold in the range

1 x 10-5 to 1 x 10-9. Note that, at themultiplex section level, ITU-T G.806(October 2000 draft) specifies the‘detection time’ for these error ratesas 1 second for 1 x 10-5 to 10,000seconds for 1 x 10-9.

The second part is the time taken forthe actual switching process tocomplete as described above. This isnormally dominated by the protocolprocessing time at each node on theProtection Circuit.

Figure A2

� The head-end finishes byperforming a switchoperation if necessary.

Following a failure, full service is notrestored until all the bridge and switchoperations are completed. A key designgoal for Network EquipmentManufacturers (NEMs) is to keep thisservice disruption a short as possible, astheir customers (Network Operators)will demand that all system deployed inthe network meet or exceed thespecification published by the governingstandards body (Telcordia or ITU-T). Thisappendix deals with the challenge ofmaking meaningful and repeatablemeasurements of protection switch time.

8

Although users of protected services aremore interested in the total servicedisruption time, the ITU-T standardsprovide separate specifications forprotection switch time and the ‘detectiontimes’ associated with the various SFand SD conditions. While this division ofservice disruption time into itscomponent part may seem unhelpful, it isnecessary due to the wide variation indetection time associated with thedifferent SF/SD conditions. These rangefrom approximately 100 µs for a LOSfailure to 10,000 seconds for a signaldegrade that has a provisioned thresholdof 1 x 10-9 error rate. Another goodreason for treating detection timeseparately is that the nature of somefault conditions can be veryunpredictable. For example, when a fibreis damaged during construction work itmay not break cleanly. Instead, theoptical signal may fade over several tensof milliseconds or vary erratically beforefinally disappearing. So the ITU-Tstandards require that, once SF/SD isdetected, a protection switch event mustbe completed in 50 milliseconds or less.This is a tough requirement, but if it ismet, end-users will not normally notice aprotection switch event even allowing fora realistic SF/SD detection time.

Now, having looked at the generalprocesses associated with a ProtectionSwitch event, it is time to address thecentral question: How can thepotection switch time of a transmissionsystem be measured to ensure that itmeets the ITU-T specification (equal toor less than 50 millisecond)?

Ideally this would be achieved bymeasuring the time from “SF/SDdetected” to “switch completed”.However, the “SF/SD detected” eventcannot be seen, and is difficult to infer,from signals outside the NetworkElements (NE’s) in the system undertest. So the question remains, ‘how can

you obtain a reliable measure of asystem’s protection switching time’.

Protection switching timemeasurements - the alternativesolutions

Two different approaches can be usedto obtain measurement results that are(or should be) closely related to asystem’s protection switching time,namely, measure the service disruptiontime associated with a SF/SDcondition that either:

1. Minimizes the ‘detection time’(create a LOS failure – typicallydetected in less than 100 µs), or

2. Eliminates the ‘detection time’(generate control parity errors* onthe entity being protected)

* B2 errors for multiplex section protectedsystems; HP-B3 errors for High-order Pathprotected systems;LP-B3/BIP-2 errors for Low-order Pathprotected systems

This document now provides adiscussion on the merits of each ofthese approaches.

(1) Service disruption time withsimulated LOS failure (minimizedetection time)

As discussed earlier in this productnote, the principles behind the servicedisruption measurement are simple:

Figure A3

� Insert a PRBS test pattern at atributary port feeding the protectedtransmission system

� Perform a loopback on this testpattern at the associated drop-sidetributary port

� Induce a failure on the ‘working’ linesystem that affects the test pattern

� Measure the duration of theresulting error-burst in the PRBStest pattern at the tributary port.

Although ITU-T G.783 (2000) definesLOS detection time as being “in theprovince of regional standards”, itprovides an example based on a valueof 100 µs or less. This is less than thedetection times specified for all otherSF conditions. Consequently, if it isassumed that the system under testdetects LOS within this ‘example’time, then inducing an ‘instantaneous’LOS failure and measuring theresulting service disruption time willprovide an accurate estimate of thesystem’s protection switching time. Intheory, this method will yield a resultthat over-estimates the protectionswitching time by a maximum of100 µs, since the measured servicedisruption time will include both theLOS detection time and the protectionswitching time.

9

While the theory behind this test methodis simple, there is a significant practicalissue that must be addressed, namely,how do you induce an ‘instantaneous’LOS. This is more difficult than itappears at first glance.

As illustrated in Figure A4, threedifferent methods can be used togenerate a LOS failure. The simplest ofthese is to manually disconnect a fibreon the working circuit (Warning:Exercise extreme caution whendisconnecting an optical fibre – followyour organization’s standard safetyprocedures). While this approach isattractive due to its simplicity,it does not result in the generation of an

‘instantaneous’ LOS condition. This isdue to the finite time associated with

As shown in Figure 5, the optical powerat the receiving end of the link will notdisappear instantaneously when a fibre isdisconnected. Instead the power levelwill roll-off over the time taken toperform the disconnection. A consequenceof this is that the receiving NE willexperience a short period of time whereerrors are generated before it enters the

LOS condition. And since these errorswill be measured as part of the servicedisruption time, the period of time theypersist is a source of ‘over-estimation’of a system’s protection switching time.It should be noted here that the durationof this error period will, in many cases,be related to the time taken todisconnect the fibre – the faster the

disconnection, the shorter the errorperiod. Consequently, variation in the‘speed’ of manual disconnection canlead to poor result repeatability.

Inserting a programmable opticalattenuator in to the working circuitprovides a more predictable method ofinducing a LOS condition. In this case theLOS condition is induced by switching-inhigh attenuation in the transmissionpath. While this approach will almostcertainly provide repeatable results, itmay not fully address the issue of‘measurement error’ due to the opticalpower level rolling-off over a finite periodof time. This particular point relates tothe fact that most programmable opticalattenuators control the level ofattenuation by mechanicallycontrolling the position of a prism.And as with any mechanical processthere will be a finite response timeassociated with this control. Note that‘response time’ is a typical technicalspecification for most programmableoptical attenuators.

the process of manually disconnectingthe fibre.

Figure A5

Figure A4

10

The final method discussed here ofinducing a LOS condition is via a testset connected in thru-mode on theworking circuit. In this case the LOScondition is induced by eitherswitching off the test set’s lasertransmitter or using its alarmgeneration controls to transmit LOS.Both of these controls will yield apredictable and instantaneous LOScondition, and consequently enablerepeatable and accurate protectionswitching time measurements to beperformed. In principle, the onlysource of measurement errorassociated with this method should bedue to the LOS detection time beingincluded in the service disruptiontime result.

From the above discussion, it is clearthat the method used to induce a LOScondition will affect both therepeatability and accuracy of protectionswitching time results as measured bythe service disruption time technique.However, in all cases the measurementerror leads to results that are an ‘over-estimate’ of a system’s actualprotection switching time performance.Consequently, if the measured servicedisruption time is less than 50milliseconds, then you can have fullconfidence that the system undertest truly complies with thepublished standards for protectionswitching time.

(2) Service disruption time withSF trigger by excessive errors(eliminate detection time)

In this method, a test set connected inthru-mode is used to inject a high-rate oferrors in the parity-check byte(s)associated with the protection systemunder test. In the case of a multiplexsection protected system B2 parityerrors are used, while HP-B3 and LP-B3/BIP-2 parity errors are used for high-order path and low-order path protectedsystem respectively. For simplicity, thefollowing discussion will assume thatthe system under test is protected at themultiplex section level.

The technique discussed here formeasuring protection switching time isbased on creating a Signal Fail in the

system under test that is caused by anexcessive error condition. In ourmultiplex section protected system thismeans injecting a B2 error rate thatexceeds the receiving NE’s provisionedthreshold for the excessive errorcondition. To ensure that you alwaysexceed the provisioned error threshold,set the thru-mode test set to inject themaximum error rate supported by theparity-check bytes (in our case –continuously error all bits of all B2bytes). Since these errors are onlyinjected in to the B2 parity bytes they willnot affect the traffic being carried in any ofthe payload channels. Consequently, noerrors will be added to the PRBS testpattern that is transmitted and monitoredby the second test set connected at atributary port of the system under test, andused to measure service disruption time.

Figure A6

11

Within 10 ms of injecting the B2 errors,the tail-end node (the NE receiving theB2 errors) will detect the excessiveerror condition. This causes the NE todeclare a SF and to initiate a protectionswitch sequence. In addition, the tail-end node is required to insert an AIS

alarm in all down-stream trafficchannels within 250 µs of declaring SF.And since this AIS will overwrite thePRBS test pattern that is transmittedand monitored by test set#2, it causesthe service disruption measurement tobe triggered (started).

For standards compliant networkelements, this method will yieldaccurate and repeatable results whenmeasuring protection switching time.Its main advantage over the ‘LOSmethods’ discussed earlier is that iteliminates the ‘SF detection time’ errorfrom the measured result. The onlytechnical drawback is that its resultsslightly under-estimate a system’sprotection switching time – but only byup to 250 µs (assuming that the tail-endnode inserts the downstream AIS withinthe 250 µs period specified in ITU-TG.783). Possibly the most serious‘drawback’ associated with thismeasurement method is a commercialone – it requires two transmission testsets (one covering the requiredtributary rates, the other coveringrequired line rates).

Figure A7

Summary and conclusions

While service disruption measurementscan be measured as described in thisnote, it is worthwhile looking at theoverall picture from a systemperspective, as illustrated in figure A8.This represents an idealized model ofwhat occurs in the physical layer during aswitch. Following error free operation, anetwork element detects an event suchas a fibre break that may give reason toperform a protection switch. Followingthe event detection, an AIS may beinitiated by processes within the networkelement. The physical switch takesplace, then a few moments later theAIS is removed. After a period ofsynchronization on the protection signalpath, error free operation is resumed.

The service disruption measurementwithin the OmniBER analyzer is definedas being the time between the end oferror free operation, before the switchtakes place, to the beginning of error freeoperation after the switch has occurred.

The OmniBER analyzer’s servicedisruption measurements are carried outby inserting a PRBS on the tributary sideof the device under test, looping it backon itself on the corresponding drop sidetributary, and monitoring this PRBS forerrors as a switch occurs. ITU-Trecommends that a protection switchingprocess (i.e. the ”switch” section offigure A8) should be 50 milliseconds orless. While this is a difficult standard tomeet, a large part of the problem is ininitiating the protection switch. There aretwo methods to do this effectively:

� Create a LOS failure, which willtypically be detected in less than 100 µs.

� Generate control parity errors on theprotection system.

While each method has its ownadvantages, it is worth remembering thatwhatever device you are trying to varify,the OmniBER analyzer provides the bestprotection-switch analysis capabilitiescurrently available.

Figure A8: Protection switch processes

Agilent Technologies’ Test and MeasurementSupport, Services, and Assistance

Agilent Technologies aims to maximize the valueyou receive, while minimizing your risk andproblems. We strive to ensure that you get the testand measurement capabilities you paid for andobtain the support you need. Our extensive supportresources and services can help you choose theright Agilent products for your applications andapply them successfully. Every instrument andsystem we sell has a global warranty. Support isavailable for at least five years beyond theproduction life of the product. Two conceptsunderlie Agilent’s overall support policy: “OurPromise” and “Your Advantage.”

Our Promise

Our Promise means your Agilent test andmeasurement equipment will meet its advertisedperformance and functionality. When you arechoosing new equipment, we will help you withproduct information, including realistic performancespecifications and practical recommendations fromexperienced test engineers. When you use Agilentequipment, we can verify that it works properly,help with product operation, and provide basicmeasurement assistance for the use of specifiedcapabilities, at no extra cost upon request. Manyself-help tools are available.

Your Advantage

Your Advantage means that Agilent offers a widerange of additional expert test and measurementservices, which you can purchase according to yourunique technical and business needs. Solveproblems efficiently and gain a competitive edge bycontracting with us for calibration, extra-costupgrades, out-of-warranty repairs, and on-siteeducation and training, as well as design, systemintegration, project management, and otherprofessional engineering services. ExperiencedAgilent engineers and technicians worldwide canhelp you maximize your productivity, optimize thereturn on investment of your Agilent instrumentsand systems, and obtain dependable measurementaccuracy for the life of those products.

By internet, phone, or fax, get assistance with all your test& measurement needs

Online assistance:

www.agilent.com/find/assist

Phone or FaxUnited States:(tel) 1 800 452 4844

Canada:(tel) 1 877 894 4414(fax) (905) 282 6495

China:(tel) 800 810 0189(fax) 1 0800 650 0121

Europe:(tel) (31 20) 547 2323(fax) (31 20) 547 2390

Japan:(tel) (81) 426 56 7832(fax) (81) 426 56 7840

Korea:(tel) (82 2) 2004 5004(fax) (82 2) 2004 5115

Latin America:(tel) (305) 269 7500(fax) (305) 269 7599

Taiwan:(tel) 080 004 7866(fax) (886 2) 2545 6723

Other Asia Pacific Countries:(tel) (65) 375 8100(fax) (65) 836 0252Email: tm_asia@agilent.com

Product specifications and descriptionsin this document subject to changewithout notice.

© Agilent Technologies UK LTD. 2002

Printed in USA, 8 March, 2002

5988-5122EN