Jitter Measurement Conundrum: Different Testers, Different Results

-- Test & Measurement World, 3/1/2004

—Ronnie Neil, Agilent Technologies

Introduction

A major measurement issue in the optical transport market today is- why do companies get significantly different jitter results from various vendor testers, and which result should they believe? The answer is that jitter tester performance varies greatly in some key aspects, due to the differences in implementation of measurement circuitry. Distinct implementations arise due to differences in the understanding of how jitter manifests itself in a real network, how that needs to be tested and general expertise in the field of jitter measurement.

The topic of jitter is a complex subject at the best of times and the accurate measurement of low levels of jitter on high bit rate optical systems e.g. 10 Gb/s, pushes technology to the limit.

When engineers use test equipment from different vendors, and see various measurement results when testing the same Device Under Test (DUT), which set of results should they believe? What can be done to identify which tester is right and which is wrong? What is being done about this situation within the industry?

For some jitter tests, such as transient peak detection, a simple test that uses only the tester itself with no external equipment can be performed quickly to root out the offending test set.

For other tests, such as measuring low levels of output jitter from a device under test, a simple resolution has not been readily available – until now. Recent work just completed on ITU-T O.172, jitter and wander measuring equipment for digital systems based on the synchronous digital hierarchy (SDH), is aimed at providing a standardized method of characterizing the performance of jitter test equipment. This should help resolve the problem of ‘different tester, different result.’

Jitter Transient Peak Detection

The need to measure peak-to-peak jitter has been an integral part of jitter standards since the early days of PDH systems. Measurements of jitter output from a DUT use a peak detector to record when a selectable threshold level of jitter is exceeded. The jitter standard for test equipment, ITU-T O.172, requires counting the number of occasions and the period of time for which a selectable threshold is exceeded. Any instance of measured jitter exceeding a pre-defined threshold is termed a jitter hit.

Figure 1. Click here to view a larger image.

When test equipment from different vendors show various results for jitter hit detection, it is a simple matter to confirm the test set functionality without the need for external test equipment.

Most jitter test equipment comprises a variable source, normally used for testing jitter tolerance of network equipment.  By setting the jitter source to output a known value and using the testers own receiver to measure the transmitter output, verification of a proper jitter hit detection can be determined.

An example of tester settings to use to confirm functionality is shown in the diagram Figure 2.

Figure 2. Click here to view a larger image.

Intrinsic Jitter Generation

When dealing with jitter measuring equipment, the term ‘intrinsic jitter’ refers to the amount of jitter the electronics of the measuring equipment contributes to the results. It follows that the lower the intrinsic jitter of measuring equipment, the more accurate will be the final evaluation.

Currently, the ITU-T Recommendation O.172 sets fixed intrinsic error limits for jitter measurement receivers; however, no definition is given to validate the various calibration schemes employed by different test set vendors. Without common understanding or control, there is inconsistency between test sets, and at worst the measurement could be completely invalid.

Intuitively one might expect that to measure a low-noise signal, it would be necessary to use low- noise measurement equipment. Indeed, the premise of ITU-T O.172 started with the concept of a test equipment residual of one-third of the DUT performance limit. Subsequently this requirement has been relaxed somewhat as for example in ITU-T G.783 where a test equipment requirement of one-half the DUT performance limit is realized. Nevertheless, it is important the measurement equipment’s own noise be significantly less than the limit being tested to guarantee accuracy.

The test equipment fixed error term or receiver intrinsic jitter is a limit put on the performance of the test equipment itself. If the receiver noise floor exceeds this amount then the ability to give a truly representative reading is impaired.

While it is generally agreed that calibration of a jitter receiver is useful, it should also be recognized that a receiver unable to make inherently low-noise optical data pattern jitter measurements, cannot be made standards compliant by software correction to give artificially low readings. Since the statistics of jitter being measured are unknown, any wholesale subtraction of calibration terms won’t give reliable readings. In fact there is a high probability of misreading the DUT jitter.

Figure 3. Click here to view a larger image.

It can be seen when measuring low levels of jitter, as required by SONET and SDH standards, that evaluating uncertainty can be a significant contributor to the final measured value. The diagram helps to highlight some key requirements for accurate, repeatable and consistent jitter measurements.

• Test equipment from different vendors is likely to have different values of intrinsic jitter due to differences in circuit designs. This causes different vendors equipment to display different results for the same DUT.

•  High intrinsic jitter increases the possibility of producing a ‘FAIL’ result for a DUT which should really be a ‘PASS’.

• Where high intrinsic jitter causes the measured reading to be on or just above the ‘PASS’ limit, it will be difficult to obtain repeatable results leading to a loss of confidence in the design of the DUT, the test equipment itself or both.

• The test equipment’s intrinsic jitter can be additive or subtractive.

• Lower intrinsic jitter reduces measurement uncertainty.

Recent Changes to ITU-T O.172

When comparing measurement results from different test equipment, getting a true understanding of the equipments real performance is not possible just from looking at a traditional paper specification. Different designs use several calibration techniques to deal with design imperfections.

While some form of calibration is generally useful, the fundamental requirement to accurately measure low levels of jitter is for a receiver with inherently low levels of intrinsic jitter. A poor receiver cannot be made standards compliant through software calibration alone.

The only way of comparing different receivers is by characterizing them against a jitter source with known negligible jitter. By plotting ‘characterization maps’ it will then be possible to objectively compare the performance of different measuring receivers.

Agilent Technologies has been instrumental in promoting a standard method of characterizing jitter test equipment, and this will be included as an appendix in the next version of ITU-T O.172.

The method provides a ‘gold standard’ optical data signal that can be used to isolate the error contributed by the test equipment from the signal being measured.

The main building blocks of the method shown in Figure 4 are:

• A reference source with negligible jitter

• A high quality optical transmitter and pattern generator

• Calibration

Figure 4. Click here to view a larger image.

It should be noted that this methodology could be used to improve the design of a jitter receiver. It cannot be used to improve a given receiver through calibration. 

Conclusion

Accurate and repeatable jitter measurement has been a test requirement since the first days of digital networking. The understanding and control of jitter has evolved as networks have grown in physical size and bandwidth. Standards continue to evolve to help resolve challenges that continually appear and the latest changes to ITU-T O.172 make it possible to eliminate some of the problems related to measuring low levels of jitter due to today’s high-capacity optical networks.