Data processing: measuring – calibrating – or testing – Calibration or correction system – Signal frequency or phase correction
Reexamination Certificate
2003-01-28
2004-03-02
Barlow, John (Department: 2863)
Data processing: measuring, calibrating, or testing
Calibration or correction system
Signal frequency or phase correction
Reexamination Certificate
active
06701269
ABSTRACT:
TECHNICAL FIELD
The invention relates generally to jitter analysis within a signal and more particularly to determining jitter characteristics of a device, wherein a “device” is broadly defined to include multi-component systems.
BACKGROUND ART
As the speeds of communications systems increase, the adverse effects of jitter on the performance of such a system also increase. The term “jitter” is defined herein as deviation of the significant instants of a signal from their ideal position in time. Thus, jitter within a clock signal or within a data signal will cause phase variations from the ideal. As examples, the “significant instants” of a data signal may be the rising and falling edges of data bits or the “significant instants” of a clock signal may be the rising edge of each pulse or may be the center of each clock pulse.
Jitter is one of the major causes of errors within a communications system. Jitter may be divided into a number of different categories. “Deterministic jitter” (DJ) is defined as the category of jitter that is predictable and constant. Deterministic jitter has specific sources. “Random jitter” (RJ) is defined as the category of jitter that will vary from measurement to measurement. Random jitter adds in a root sum of square (RSS) basis. One source of random jitter is thermal noise in electrical components. Random jitter is typically assumed to have a Gaussian distribution. As is know in the art, a Gaussian random variable will exceed fourteen times its standard deviation (i.e., 14×S.D.) only one time in 10
12
. As a consequence, if exceeding this span causes a bit error in a communications system, the system has a bit error rate of 10
−12
. “Data dependent jitter” (DDJ) is defined herein as jitter that will vary in accordance with the pattern of data within a data signal.
The characterization of a digital communications system may be performed using two separate measurement devices, namely a bit error ratio tester (BERT) and a digital communications analyzer (DCA). The BERT is an error performance analyzer, while the DCA is a sampling oscilloscope. The DCA is utilized to generate eye mask measurements. This may be a 30 second test in which the sampling rate is 40 kHz, which corresponds to more than one million samples. The eye mask test detects distortions that include overshoot, rise time, and jitter. The DCA is superior to the BERT with respect to measuring DDJ, since the DCA has a frequency response that more closely approaches the ideal Thompson-Bessel response. A DCA adds less distortion to the measurement of a device under test than does the BERT. However, while the DCA can acquire parametric measurements that are in some aspects more accurate than the BERT, the DCA operates at a lower sampling rate, so that test times are significantly longer.
A BERT error detector has decision circuits that operate at a high bit rate. Every bit in a sequence is measured, but the BERT returns only a binary result (i.e., either the bit is “correct” or it is “in error”). In comparison, the DCA samples the bit stream at a relatively low rate (e.g., 40 kHz), but the amplitude of the signal is measured to 10 bits, for example. In some standards, jitter is tested by measuring the bit error ratio as a function of sampling time. The graphing of such measurements provides what is referred to as the “bathtub jitter curve,” on the basis of the shape of the curve. By mapping the measurements of the bit error rate versus sampling point delay, jitter can be equated from the bathtub curve and viewed as a histogram.
The BERT measurements at a speed of three Gb/second to an error rate of 10
−12
may take approximately five minutes. In order to decrease the time necessary to test a number of devices, the BER performance at the vertical edges of the “eye” may be extrapolated (Q measurement). In an application published under the Patent Cooperation Treaty (WO 99/39216), Wilstrup et al. describe a method of using a time interval analyzer to extrapolate jitter measurements. The described method includes obtaining measurements of the spans of a signal from a device under test, generating variation measurements for each of the spans, transforming the variation estimates from a time domain to a frequency domain, and determining the random component and the deterministic component of the jitter. As with other extrapolation approaches, the assumption of a Gaussian distribution of total jitter is required, but the efforts to group all jitter components for an extrapolation lead to inaccurate predictions in some applications.
As is known in the art, finite bandwidth in the jitter measurement devices causes degradation of the eye opening. This adds pattern dependent jitter and results in a slowing of the transition times in the eye. The DCA suffers from the bandwidth limitation much less than the BERT. Another cause of the closing of the eye in time is the fact that jitter is inherent to the measurement device. U.S. Pat. No. 6,185,509 to Wilstrup et al. addresses this second concern by using a time interval analysis approach to reducing the effects of inherent measurement jitter. However, the issue of finite frequency response is not addressed. Thus, while advancements in jitter analysis are being made, further improvements are possible.
SUMMARY OF THE INVENTION
Extrapolation of bit error ratio-related information acquired using a bit error ratio tester (BERT) is enabled by “calibrating” the extrapolation process. The extrapolation initialization involves utilizing at least one measurement of jitter that is superior to the corresponding measurement available via the BERT. A digital communications analyzer (DCA) may be used to obtain the measurement or measurements considered to be superior to those acquired via the BERT.
As one possible implementation, the DCA is connected to a device under test and is configured to maximize its accuracy with regard to measuring deterministic jitter (DJ). A “device under test” is defined broadly herein as including multi-component systems, such as a digital communications system. A reliable measure of DJ of a transceiver under test may be acquired by utilizing a quadrature time base and pattern trigger. The quadrature timebase provides very low random jitter (RJ) in the measurement, unlike typical DCA measurements. The worst-case data transitions are identified. The “worst-case” data transitions in a testing signal from the device under test are bit transition pairs (rising and falling edges) that yield the most closed “eye” in the mid-range decision point voltage. Analysis of eye diagrams is known in the art.
The BERT is one that has error location capability. Consequently, the worst-case data transitions identified by the DCA can be remeasured via the BERT, with the remeasuring of the worst-case data transitions being performed as a function of sampling time down to a first error rate. For example, the first error rate may be 10
−8
.
In “correcting” the BERT data, the data dependent jitter (DDJ) measurement by the DCA may be considered to be correct, since it is superior to the DDJ measurement capability of the BERT. With this assumption, the bit error ratio data of individual pattern transitions are offset in time. The offset may be by the difference of the DDJ measurements on the BERT and on the DCA, but other approaches may be used. With the corrected bit error ratio data, the BERT bathtub jitter curve (bit error ratio as a function of sampling point delay) for the device under test is calculated. DJ and RJ are evaluated from this bathtub jitter curve and the data is extrapolated to a second error rate (for example, 10
−12
) that is lower than the first error rate. Subsequent extrapolations of BERT-acquired data may be performed using the same offset in time, so that the use of the DCA is not necessary, particularly if the type of device under test and the test pattern (for example, a stress pattern of finite length) remain the same.
In another implementation, the BERT is used to first identify the worst-case data transitions,
Camnitz Lovell H.
Jungerman Roger L.
Agilent Technologie,s Inc.
Barlow John
Lau Tung
LandOfFree
Jitter measurement extrapolation and calibration for bit... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Jitter measurement extrapolation and calibration for bit..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Jitter measurement extrapolation and calibration for bit... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3241261