Sie sind auf Seite 1von 5

Display,

actuators,
Sensor Input Processing Output signals,
control

Pulse Metrology – Part 2


Part 39 in a series of tutorials on instrumentation and measurement
Nicholas G. Paulter and Donald R. Larson
Official contribution of the National Institute of Standards and Technology,
not subject to copyright in the U.S.A.

T
his article is the second part of a two-part se- Pulse Metrology
ries discussing pulse metrology. One who
Standards (continued from Part 1)
is doing pulse measurement needs to know
what has been measured, to what accuracy, and with Parameter computation: As with terms and definitions,
what uncertainty. Part 1 introduced the field of me- the purpose is to have consistent methods of comput-
trology and began an explanation of standards in ing a given parameter, which is essential to communicate
pulse metrology and some of the commonly pulse parameters and for comparison between

39
used pulse terms [1]. measurement facilities. Any term
Part 2 continues with that describes a pulse parame-
topics within the pulse me- ter and can also be computed
trology standards, with
9 is provided with a computa-
a bias toward the work
performed by the au-
4 2
Thirtynine
tion algorithm in the IEEE
Std 181-2011[2]. Adherence
thors, formerly in the 3 to the IEEE Std 181-2011 re-
in a 47
area of pulse measure-
ments in support of
the digital telecommu-
Series
56
81
20
quires that if the method
of computation used is dif-
ferent from that described
nications industries and in the IEEE Std 181-2011, then
presently in the areas of this computational method must
weapons testing, concealed be clearly described. If the source code
weapon detection, through-barrier im- of the algorithm used in the computation is not
aging, and imaging metrology. In this Part, we will available for scrutiny, this algorithm should be exhaus-
address parameter computation, test methods, and tively tested. We have observed differences between the
the use of artifacts (test objects) and conclude the IEEE Std 181-2011 computation methods and those used in
tutorial with discussions of traceability and measure- commercial software for certain pulse parameters. To max-
ment uncertainty. imize the accuracy of the comparison of parameter values,

June 2012 IEEE Instrumentation & Measurement Magazine 43


1094-6969/12/$25.00©2012IEEE
it is important to limit, or at least understand, the difference in (pulse generator) whose temporal parameters (transi-
computation methods. As an example, the effects of several dif- tion duration, for example) are of shorter duration than
ferent algorithms for determining the state levels of a step-like that of the item being measured. Much research has been
waveform were examined and shown to provide similar but not directed to the case when this is not true, that is, when the
identical results [3]. These methods were affected by noise and device or instrument under test has an impulse response
periodicity of features on the nominally static levels of step-like or output pulse with temporal parameters similar to that
waveforms. of the artifact. The following different approaches have
Test methods: Certain pulse parameters, such as impulse spec- been developed to address this issue.
trum amplitude [4], can be measured in more than one way. For •• One method [7], [8] to calibrate either the pulse genera-
impulse spectrum amplitude, for example, we had developed tor or sampler uses specially-designed semiconductor
several different methods and adopted the one exhibiting the devices based on the semiconductor’s photoconduc-
least measurement uncertainty [4] (which also happened to be tive (PC) response. This method can generate long
the one having the greatest reproducibility and repeatability to ultra-short-duration electrical pulses depending
and lowest cost to implement). Because test methods may yield on the duration of the driving optical pulses. A PC
different results, test methods are not prescribed in the IEEE pulse generator or sampler can be the artifact. The
Std 181. Instead, measurement laboratories develop their own key to making a PC system work is synchronization
systems for measuring given parameters and rely on the use of between the optical pulses used for pulse genera-
transfer standards to demonstrate equivalence of the methods. tion and sampling, which is achieved by splitting one
Standards development organizations (SDOs) can develop optical pulse into two daughter pulses. One daughter
test method standards if there is sufficient interest within the pulse is used to drive the PC pulse generator and the
user and manufacturer communities. Whatever test method is other pulse to drive the PC sampler. For a PC sampler/
selected, that method must be well documented and have an pulser, the measurement process is described math-
accurate and realistic uncertainty analysis. The test method ematically as an autocorrelation and, consequently,
and any changes must be documented to ensure that qual- phase information on these two PC functions is lost,
ity of the measurement process is maintained as described in which is a deterrent to their use in measuring devices
ISO 17025 [5]. Test methods developed by a dominant manu- with similar temporal characteristics. However, if the
facturer or a national metrology institute (NMI) often become PC device has temporal characteristics that are fast
the de facto test method standard, requiring those entities to enough relative to the temporal characteristics of what
maintain that capability as a reference for industry and other is being measured, then the PC device will have a negli-
interested parties. Furthermore, because of the globalization gible effect on the measurement.
of commerce, NMIs must perform interlaboratory compari- •• A similar approach to that just described uses a PC
sons to ensure their test methods yield consistent results and pulser and an electro-optic (EO) sampler [9]–[11],
to have their measurement capabilities listed in the Bureau In- where both functions are derived from a common opti-
ternational des Poids et Mesures (BIPM) database of Calibration cal pulse. The EO sampler, in conjunction with the
and Measurement Capabilities. optical pulse, is assumed to be so fast that the collec-
Artifacts (test objects): Artifacts are used to ensure the same tive effect on the measurement of the electrical pulse
measurement result is obtained from different measurement is negligible. In this case, the EO-sampler-measured
systems or from the same measurement system but at different electrical pulse becomes the reference pulse and the
times or under different conditions. Artifacts in pulse metrology PC pulser becomes the artifact. In 1984, NIST imple-
are physical devices, such as reference pulse generators, delay mented the EO-sampler method to measure the output
lines or generators, impedance standards, samplers, etc. The of optical detectors [12]. Since about 1990, the National
choice of artifacts depends on the parameter to be tested. For ex- Physical Laboratory (NPL) has focused on such
ample, in the NIST reference ballistic chronograph (a device for EO-based approaches [9] and, more recently, so has the
measuring bullet velocity) and the electroshock weapons (ESW) Physikalisch-Technische Bundesanstalt (PTB) [10]. In
measurement systems (developed by the authors), synthesized 2005, NIST management decided to migrate from the
frequency sources are used to calibrate the waveform recorder nose-to-nose (NTN) method to the EO method, and
timebase; these sources are traceable to the second. The applica- now relies exclusively on EO sampling [13] to calibrate
tion of pulse metrology to the reference ballistic chronograph is pulse generators.
described in [6]. A short discussion of the different artifacts used •• Another approach similar to the above is the use of
in pulse metrology is now provided. a non-optically driven pulse generator (such as an
◗◗ Pulse generators and/or samplers - Most often the pulse avalanche diode or other nonlinear semiconductor
generator or sampler is adequate to act as an artifact. Cali- device) and either a PC or EO sampler to measure the
brated pulse generators are used as artifacts for sampler output of this generator [14], [15]. This method also
calibration and calibrated samplers are used as arti- requires synchronization between the pulse gener-
facts for pulse generator calibration. The artifact should ator and the sampler, which could be accomplished
possess an impulse response (sampler) or output pulse using pulses from a pulse laser (acting as the master

44 IEEE Instrumentation & Measurement Magazine June 2012


Display,
actuators,
Sensor Input Processing Output signals,
control

oscillator) to minimize jitter. This jitter may be signif- accurately provide a measure of a long delay with short-
icant, being similar in magnitude to, and perhaps in duration sampling intervals.
some cases greater than, the transition duration of the ◗◗ Impedance - Impedance artifacts are important in the cali-
generated electrical pulse. bration of pulse generators and waveform recorders [27].
•• Another method is the nose-to-nose (NTN) method The impedance artifact may be either a well-character-
for determining the impulse response of sampling ized transmission line or wide-bandwidth resistive load.
heads [16]. This method is based on a specific type of Resistive loads are convenient because they are lumped
sampling head that, when strobed by the trigger signal elements and are not as subject to mechanical damage
with a non-zero offset, generates a pulse (called the as is a transmission line. We use resistive loads to cali-
“kick-out” pulse) that has temporal features resem- brate our ESW measurement system. However, resistive
bling that of the sampling aperture. The kick-out loads at high frequencies exhibit parasitic features that
pulse can be used as the pulse artifact or its impulse cause their impedance to vary from design values. Trans-
response as the sampler artifact. This measurement mission line impedance artifacts, on the other hand,
produces a result that is mathematically described especially coaxial ones, have accurate formulas that
as a convolution, so phase information is not lost. describe their impedance as a function of frequency for
Because the kick-out pulses are nominally identical, the geometry and materials of construction. However,
the ability to accurately measure the kick-out pulse transmission lines can be easily damaged. Also, transmis-
and assign uncertainties to its amplitude profile is sion lines have a given length and, consequently, a fixed
important. We developed an uncertainty analysis round trip propagation time. This allows the transmis-
for the NTN method [17], [18] that provides bounds sion line to be used as an impedance calibration only for
for the uncertainties of the kick-out pulse generator a measurement epoch equal to or less than the round-trip
and sampler waveforms and spectra (magnitude and pulse propagation time of the transmission line.
phase). Other national and industrial labs use and
study the nose-to-nose method (see for example, [19]– Traceability
[23]), however, as noted above, several metrology labs Traceability, specifically metrological traceability [28], is “a
are pursuing the EO method because of its potential property of a measurement result whereby the result can be
to provide a traceable measurement process [9], [10], related to a reference through a documented unbroken chain
[13], [24]. of calibrations, each contributing to the measurement un-
◗◗ Delays - Delay artifacts can be accurately measured certainty.” Therefore, traceability is a general measurement
using the calibrated timebase of a waveform recorder characteristic and provides the means by which different en-
or by comparing it to a reference delay, which is accom- tities can ensure that comparisons of quantities are accurate,
plished using an electronic device or a transmission line consistent, and reproducible. Traceability may be accomplished
(including variable-delay trombone lines). Electronic using the fundamental SI units or measurements of artifact stan-
delays include time mark generators and electronic dards. To use an artifact standard for traceability, the artifact
delay lines. Both of these devices are subject to jitter and is sent between a reference laboratory and a subordinate lab-
so cannot be used to accurately measure short delays. oratory whose measurement results are not as accurate as or
However, for long delays where the jitter is inconse- uncertainties are not as small as that of the reference lab.
quential relative to the measured delay (but it still must Pulse metrology is concerned with amplitude levels and the
be measured and included in the delay uncertainty), instants that these levels change. The amplitude can be traced
electronic delays can be used. Transmission line delays to the volt, an SI “derived unit” with appropriate instrumenta-
are based on the pulse propagation velocity in the trans- tion (an example is given in [29]). The durations are traced to the
mission line and the length of the transmission line. SI unit of the second via calibration of the waveform recorder
Because of these facts, transmission lines are useful only timebase using known frequencies (an example is given in [30].)
for short delays. However, if optical pulses are used to Pulse parameter interlaboratory comparisons [31], [32] are
achieve the delay, fairly long optical fibers can be used essential to assess the consistency of pulse measurement and
to provide microseconds of reference delay. Examples of to improve and advance pulse metrology. These comparisons
measuring delays and establishing reference delays can must have an established procedure for: transferring the arti-
be found in [25] and [26]. fact standard; defining the measurement process, including
◗◗ Frequency synthesizer - The timebase of the waveform processing, data formats, data precision, etc.; and defining the
recorder can be calibrated using a frequency synthesizer policy with which measurement results are disseminated and
whose output is traceable to the second. The synthesizer measurement differences addressed.
allows calibration of any epoch and consequently of any
delay. The accuracy of this delay artifact depends on the Measurement Uncertainty
number of periods of the sinusoidal signal captured in Measurement uncertainty provides information on how likely a
a waveform epoch relative to the required delay. This measured value is to the actual value (reference). Elements con-
method requires that several frequencies be used to tributing to pulse parameter uncertainty include: the response

June 2012 IEEE Instrumentation & Measurement Magazine 45


Tables of the pulse parameters
corresponding to these wave-
forms are given in [18].

Summary
Pulse metrology is a measure-
ment science that provides
reproducible and repeatable
measurements of pulse signals
with defensible uncertainties.
These uncertainties describe
the sensitivity of the measur-
and, the thing for which one
wants to find a number, to
various parameters and ef-
fects. Pulse metrology affects
the commercially-important
telecommunications, data
communications, and com-
puting industries. For more
information on pulse measure-
ment service offerings by NIST
see [36], by PTB see [37], and by
Fig. 1. Reconstructed waveforms showing uncertainties due to uncertainties in the magnitude (ux) NPL see [38].
and phase (uθ) of the spectrum. Although this Part 2 and
the previous Part 1 are only a
of sensors and instrumentation, background effects (tempera- partial introduction to pulse metrology, hopefully they dem-
ture, humidity, electromagnetic interference), data extraction onstrate the importance of this work and the challenge to
algorithms, human variability for manually-operated systems, continuously provide the manufacturing and user communities
calibration uncertainty, artifact uncertainty, and connection with measurement capability that exceeds their present require-
variability. The effect of the deconvolution process on the re- ments and that can be expanded to meet future requirements.
sultant (reconstructed) waveform must also be considered. The
reconstructed waveform represents the best estimate for the References
waveform that would have been measured by an instrument [1] N. G. Paulter and D. R. Larson, “Pulse metrology: Part 1,” IEEE
having an ideal impulse response. Instrum. Meas. Mag., vol. 14, no. 3, pp. 39–44, 2011.
◗◗ Pulse parameter uncertainty: There are many pulse param- [2] IEEE Standard on Transitions, Pulses, and Related Waveforms, IEEE
eters defined in the IEEE Std 181, and all of these can be Standard 181-2011.
subjected to a measurement uncertainty analysis. However, [3] O. M. Solomon, D. R. Larson, and N.G. Paulter, “Comparison of
the underlying uncertainty processes are often much some algorithms to estimate the low and high state level of pulses,”
more complicated than expected and, therefore, uncer- Proc. IEEE Instrumentation and Measurement Technology Conference
tainty analyses have been developed only for the most 2001, Budapest, Hungary, pp. 96–101, 2001.
commercially-important parameters. As an example of the [4] N. G. Paulter and D. R. Larson, “Impulse spectrum amplitude
complexity of pulse parameter uncertainty analysis see [33]. uncertainty analysis,” Metrologia, vol. 43, pp. 477–485, 2006.
◗◗ Waveform uncertainty: One of the most difficult uncertain- [5] ISO/IEC General requirements for the competence of testing and
ties to determine is the sample-by-sample uncertainty of calibration laboratories, ISO/IEC 17025:2005 (2005).
the reconstructed waveform. To do this requires that the [6] N. G. Paulter and D. R. Larson, “Reference ballistic chronograph,”
impulse response of the reference sampler or the output Optical Engineering, vol. 48, pp. 043602-1–043602-7, Apr. 2009.
of the reference pulse generator be known. NIST has [7] D. H. Auston, “Picosecond optoelectronic switching and gating in
developed a covariance based method for computing silicon,” Appl. Phys. Lett., vol. 26, pp. 101–103, 1975.
measurement uncertainties [34], [35]. [8] N. G. Paulter, “High-speed optoelectronic pulse generation and
Fig. 1 shows waveforms in which the magnitude of the sampling system,” IEEE Trans. Instrum. Meas. IM-37, pp. 449–453,
spectrum of these waveforms has been changed by adding 1988.
various amounts (given by n) of a ratioed magnitude [9] D. Henderson, A. G. Roddie, and A. J. A. Smith, “Recent
uncertainty (ux) and in which the phase of the spectrum developments in the calibration of fast sampling oscilloscopes,”
of these waveforms has been changed by adding various IEE Proceedings-A, vol. 139, pp. 254–260, Sept. 1992.
amounts (given by m) of a ratioed phase uncertainty (uθ). [10] S. Seitz, M. Bieler, M. Spitzer, K. Pierz, G. Hein, and U. Siegner,

46 IEEE Instrumentation & Measurement Magazine June 2012


Display,
actuators,
Sensor Input Processing Output signals,
control

“Optoelectronic measurement of the transfer function and time [26] D. R. Larson and N. G. Paulter, “Nanosecond delay with
response of a 70 GHz sampling oscilloscope,” Meas. Sci. Technol., subpicosecond uncertainty,” Review of Scientific Instruments, vol. 78,
vol. 16, no. 10, pp. L7–L9, Oct. 2005. pp. 084701-1–084701-5, Sept. 2007.
[11] J. A. Valdmanis and G. A. Mourou, “Subpicosecond electrooptic [27] N. G. Paulter and D. R. Larson, “Impedance of transfer standards
sampling: Principles and applications,” IEEE J. Quantum Electron., used in the calibration of high-speed samplers and pulse
vol. QE-22, no. 1, pp. 69–78, Jan. 1986. generators,” IEEE Trans. Instrum. Meas., vol. 56, pp. 1930–1934, Oct.
[12] R. J. Phelan, Jr., D. R. Larson, N. V. Frederick, and D. L. Franzen, 2007.
“Detectors for Picosecond Optical Power Measurements,” in [28] ISO-IEC International vocabulary of metrology — Basic and general
Optical Radiation Measurements, A. A. Sanders, Ed., San Diego, CA: concepts and associated terms (VIM), ISO-IEC Guide 99.
International Society for Optical Engineering, Aug., 1984. [29] T. M. Souders, B. C.Waltrip, O. B. Laug, and J. P. Deyst, “A wide-
[13] D. F. Williams, P. D. Hale, T. S. Clement, and J. M. Morgan, band sampling voltmeter,” IEEE Trans. Instrum. Meas., vol. 46, no. 4,
“Calibrating electro-optic sampling systems,” IEEE MTT-S Int. pp. 947–953, Aug. 1997.
Microwave Symp. Dig., pp. 1527–1530, May 2001. [30] G. N. Stenbakken and J. P. Deyst, “Time-base nonlinearity
[14] M. G. Li, E.A. Chauchard, Chi H. Lee, and H-L.A. Hung, determination using iterated sine-fit analysis,” IEEE Trans. Instrum.
“Intermixing optical and microwave signals in GaAs microstrip Meas., vol. 47, pp. 1056–1061, Oct. 1998.
circuits for phase locking applications,” IEEE Trans. Microwave [31] N. G. Paulter, A. J. A. Smith, D. R. Larson, T. M. Souders, and
Theory and Techniques, vol. 38, 1924–1931, Dec. 1990. A. G. Roddie, “NIST-NPL interlaboratory pulse measurement
[15] K. J. Weingarten, M. J. W. Rodwell, and D. M. Bloom, “Picosecond comparison,” IEEE Trans. Instrum. Meas., vol. 52, pp. 1825–1833,
optical sampling of GaAs integrated circuits,” IEEE. J. Quantum. Dec. 2003.
Elect., vol. 24, pp. 198–220, 1988. [32] M. Bieler, S. Seitz, M. Spitzer, G. Hein, K. Pierz, U. Siegner, M. A.
[16] K. Rush, S. Draving, and J. Kerley, “Characterizing high-speed Basu, A. J. A. Smith, and M. R. Harper, “Rise-time calibration of
oscilloscopes,” IEEE Spectrum, pp. 38–39, 1990. 50-GHz sampling oscilloscopes:intercomparison between PTB and
[17] N. G. Paulter and D. R. Larson, “Sources of uncertainty in the nose- NPL,” IEEE Trans. on Instrum. Meas., vol. 56, pp. 266–270, Apr. 2007.
to-nose sampler calibration method,” IEEE Trans. Instrum. Meas., [33] N. G. Paulter and D. R. Larson, “Pulse parameter uncertainty
vol. 52, pp. 1618–1626, Oct. 2003. analysis,” Metrologia, vol. 39, 2002, pp. 143–155.
[18] N. G. Paulter and D. R. Larson, “Uncertainty in the nose-to-nose [34] P. D. Hale, A. Dienstfrey, C. M. Wang, D. F. Williams, A.
sampler calibration method due to the assertion that the generator Lewandowski, D. A. Keenan, and T. S. Clement, “Traceable
and receiver functions are identical,” IEEE Trans. Instrum. Meas., waveform calibration with a covariance-based uncertainty
vol. 56, pp. 2570–2576, Dec. 2007. analysis,” IEEE Trans. Instrum. Meas., vol. 58, no. 10, pp. 3554–3568,
[19] Jiang-miao Zhu, Ming-liang Liu, and Feng Lu, “The theoretical Oct. 2009.
analysis of nose-to-nose calibration procedure,” Acta Electronica [35] C. M. Wang, P. D. Hale, and D. F. Williams, “Uncertainty of
Sinica, vol. 33, no.3, pp. 480–483, Mar. 2005. timebase corrections,” IEEE Trans. Instrum. Meas., vol. 58, no. 10,
[20] M. El Yaagoubi, G. Neveux, D. Barataud, T. Reveyrand, J.-M.l pp. 3468–3472, Oct. 2009.
Nebus, F. Verbeyst, F. Gizard, and J. Puech, “Time-domain [36] “High-Speed Measurements,” NIST, Physical Measurement
calibrated measurements of wideband multisines using a large- Laboratory. [Online] Available: http://www.nist.gov/pml/
signal network analyzer,” IEEE Trans. Microwave Theory and div686/sources_detectors/measurements.cfm.
Techniques, vol. 56, pp. 1180–1192, May 1980.
[21] J. B. Scott, “Rapid millimeter-wave sampler response Nicholas G. Paulter (paulter@nist.gov) began his career in
characterization to well beyond 120 GHz using an improved pulse metrology at the Los Alamos National Laboratory in
nose-to-nose method,” IEEE MTT-S Int. Microwave Symp. Dig., pp. 1980 where he worked on high-speed photoconductors. In
1511–1514, June 2003. 1989, he joined the National Institute of Standards and Tech-
[22] J.-G. Lee, J.-S. Kang, T.-W. Kang, and S.-Ho Won, “Microwave scope nology (NIST) in Colorado and later in Gaithersburg, MD,
response characterization using a nose-to-nose method,” 34th Int. to develop pulse measurement techniques and analysis.
Conf. Infrared, Millimeter, and Terahertz Waves, 2009. IRMMW-THz In 2006, he left the Quantum Electrical Metrology Division
2009, Sept. 2009. to become a program manager with the Law Enforcement
[23] X. Qinghua, L. Maoliu, and Z. Yichi, “Phase response of arbitrary Standards Office at NIST, Gaithersburg, overseeing the
frequency grid reconstruction of sampling oscilloscopes based on application of pulsed terahertz and microwaves to law en-
the NTN calibration,” Chinese J. Scientific Instrument, 2011-05. forcement and homeland security. He is a Fellow of the IEEE.
[24] L. Maoliu, “Progess in high-speed sampling oscilloscope
calibration technique - on quantity transfer chain for traceable Donald R. Larson has been involved in pulse metrology most of
waveform standard calibration,” Foreign Electronic Measurement his career starting in the Optoelectronics Division of the NIST,
Technology, 2010-10. Boulder, CO, from 1976 until 1998. In 1998, he moved to the
[25] D. R. Larson, N. G. Paulter, and K. C. Blaney, “Characterization and Quantum Electrical Metrology Division at NIST, Gaithersburg,
calibration of an optical time domain reflectometry calibrator,” in MD. Since 2006, he has been with the NIST Law Enforcement
Proc. NCSLI 2006 Workshop and Symposium, Nashville, TN, USA, Standards Office (OLES). He is a Senior Member of both the
Aug. 2006. OSA and the IEEE.

June 2012 IEEE Instrumentation & Measurement Magazine 47

Das könnte Ihnen auch gefallen