Instrumentation & Measurement Magazine 25-1 - 78

put together to reconstruct the original waveform, as shown
in Fig. 1 [11].
The maximum input frequency is not limited by the sampling
rate but by the bandwidth of the analog front-end and
the sampling circuitry. Moreover, a higher voltage resolution
can be achieved with respect to the case of real time oscilloscope
as the ADC of the sampling oscilloscope can process
each sample in a longer time [10]. The ETO can provide very
low intrinsic jitter, allowing an accurate RJ measurement if the
trigger signal has low jitter, and a highest front-end bandwidth
[14]. Unlike RTOs, ETOs are more affected from trigger jitter
since multiple triggers are used to acquire a single waveform
cycle. Moreover, specific edge transitions behavior cannot be
analyzed individually. As a consequence, separating RJ from
TJ can be difficult if DJ component is present in the signal under
test [14].
Bit Error Rate Tester
The BER is a figure of merit largely used in data transmission.
BER is the ratio of the incorrectly received bits to the total
number of transmitted bits. The BERT is an instrument that
can both stimulate and receive digital signals to subsequently
compare the received signal with the expected pattern, providing
the BER value. In this way, bit reception mistakes can
be easily identified [16]. In general, the BERT operates by
first creating a data stream that is sent to the device under test
(DUT). After the transmitted data stream is received, it is compared
to the original data pattern, and the number of failing
bits is counted, providing the BER value [11]. A variety of data
stream configurations can be set, including parallel and serial
data streams, enabling the BERT to perform tests for different
devices, such as SERializer/DESerializer (SERDES) modules
[17]. A BERT can allow the identification of non-repetitive or
sporadic failures, too [16].
BERTs can provide the bathtub plot [5], showing the relationship
between the BER and the sampling location in the
Unit Interval (UI), to perform the jitter separation. This kind of
plot is fairly flat where the DJ jitter mechanisms are predominant,
at both the ends of UI, since the BER value is high. RJ
mechanisms are predominant, instead, for sampling locations
approaching the center of the UI, where the BER value rapidly
decreases [5].
Longer measurement time leads to measured BER values
with a better accuracy. Curve extrapolation techniques using
statistical jitter models can extend the measured BER to lower
values without requiring unfeasible test times, but in any case,
long test time is required to carry out RJ and DJ component
separation from the bathtub curve. This method can only be
used with a priori assumption about DJ PDF (typically double-delta
model). This approximation can lead to smaller DJ
results [6], [14].
Spectrum Analyzer
The SA is an instrument that allows the measurement of the
spectral (i.e., frequency domain) composition of an electrical
signal [16]. It is important to note that although most RTOs
78
provide a way to estimate the spectral components of a signal
by applying an FFT to the acquired signal, their measurement
accuracy is inferior to a dedicated SA instrument [11].
Whereas SAs can be used to determine clock jitter, they cannot
be used for data jitter [16], [13]. The noise floor of an SA
measurement is much lower than that of an oscilloscope. However,
this makes the SA much more sensitive to low level spurii
often hidden in the input signal noise [15].
An SA presents good performance in tracking down and
characterizing periodic features of the signal (e.g., periodic
jitter, PJ) and can also be useful and very accurate (because
of its superior signal-to-noise ratio, SNR) in cases where the
phase information is not absolutely needed, in particular for
RJ measurements [11]. Compared to conventional time domain
instruments, SA can detect signal frequency and phase
changes with greater accuracy in terms of phase noise, a
figure of merit assessed in the frequency domain, used in
band-limited applications, for example, in radio frequency
communications [2], [3], [5].
Phase noise measurements can be also carried out by using
phase noise analyzers. They operate as extremely sensitive
receivers and can separate the amplitude and phase noise components
of the signal over a limited bandwidth [10]. This kind
of instrument is essential when the RJ is required to be carefully
monitored, such as when used to examine noise-floor
mechanisms, phase-locked loops, voltage-controlled oscillators,
crystal oscillators, and other clocks and references [17].
However, low-level analysis can be carried out over a limited
bandwidth (less than 200 MHz) [17], that may not be enough
for some standards, and require the use of the other kinds of jitter
measurement instruments [17].
Some Considerations about the
Instrument Choice
Significant differences can arise in the jitter measured results
when different instruments are used, and these differences
cannot be easily traced back to a difference in the instrument
accuracy. Currently, instrumentation used to measure jitter
generally includes jitter analysis software or can work with
third-party software to carry out measurements in a short time
along with many different visualization tools. However, the
implemented algorithms usually differ from one other, and
can be protected by copyright/patents [6]. Therefore, it is important
to highlight that problems arise, not only from the
differences among the available kind of instruments, but also
from the different implemented algorithms or the different parameters
used. Moreover, the same algorithm with the same
parameters has different accuracy depending on the hardware
with which it is used [6].
Comparisons of selected instruments or algorithms, showing
the significant differences arising from the jitter measured
results when testing the same DUT have been reported in
[7], [14], [18]-[20], with most focusing on time domain measurements.
Obviously, no reference can explain all possible
disagreements in jitter measurements, because of the multitude
of different factors involved. For this reason, a universal
IEEE Instrumentation & Measurement Magazine
February 2022
Instrumentation & Measurement Magazine 25-1

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 25-1
