Lecroy WP03 Parameter Analysis User Manual

WP03 -PARAMETER ANALYSIS PACKAGE

OPERATION AND PROGRAMMING MANUAL

9300 SERIES

DIGITAL STORAGE OSCILLOSCOPES

LeCroy Corporation
700 Chestnut Ridge Road
Chestnut Ridge NY 10977-6499 USA
Phone: 1-800-5-LECROY 1-914-425-2000
FAX: [phone]
www.lecroy.com

&RUSRUDWH+HDGTXDUWHUV

700 Chestnut Ridge Road
Chestnut Ridge, NY 10977–6499
Tel: [phone], Fax: [phone]

(XURSHDQ0DQXIDFWXULQJ

2, rue du Pré-de-la-Fontaine
1217 Meyrin 1/Geneva, Switzerland
Tel: [phone], Fax: [phone]

,QWHUQHWwww.lecroy.com

Copyright © July 1998, LeCroy. All rights reserved. Information in this publication supersedes all
earlier versions. Specifications subject to change.

LeCroy, ProBus and SMART Trigger are registered trademarks of LeCroy Corporation. Centronics is
a registered trademark of Data Computer Corp. Epson is a registered trademark of Epson America

2

Inc. I

C is a trademark of Philips. MathCad is a registered trademark of MATHSOFT Inc. MATLAB is
a registered trademark of The MathWorks, Inc. Microsoft, MS and Microsoft Access are registered
trademarks, and Windows and NT trademarks, of Microsoft Corporation. PowerPC is a registered
trademark of IBM Microelectronics. DeskJet, ThinkJet, QuietJet, LaserJet, PaintJet, HP 7470 and
HP 7550 are registered trademarks of Hewlett-Packard Company.

LCXXX-WPO3-OM-E Rev D 0898

&KDSWHU³,QWURGXFWLRQ

7KH9DOXHRI'62+LVWRJUDPVDQG7UHQGV.........................1–1

&KDSWHU³+LVWRJUDPV

7KHRU\RI2SHUDWLRQ.................................................................... 2–1

&UHDWLQJDQG$QDO\]LQJ+LVWRJUDPV ....................................2–10

&KDSWHU³+LVWRJUDP3DUDPHWHUV

$YHUDJHDYJ ..........................................................................................3–1

)XOO:LGWKDW+DOI0D[LPXPIZKP .............................................3–2

)XOO:LGWKDW[[0D[LPXPIZ[[ ..............................................3–3

+LVWRJUDP$PSOLWXGHKDPSO .........................................................3–4

+LVWRJUDP%DVHKEDVH.....................................................................3–5

+LJKKLJK ................................................................................................3–6

+LVWRJUDPPHGLDQKPHGLDQ ..........................................................3–7

+LVWRJUDP5RRW0HDQ6TXDUHKUPV...........................................3–8

+LVWRJUDP7RSKWRS ..........................................................................3–9

/RZORZ................................................................................................. 3–10

0D[LPXP3RSXODWLRQPD[S ......................................................... 3–11

0RGHPRGH.......................................................................................... 3–12

3HFHQWLOHSFWO..................................................................................... 3–13

3HDNVSNV ............................................................................................ 3–14

5DQJHUDQJH........................................................................................ 3–16

6LJPDVLJPD .......................................................................................3–17

7RWDO3RSXODWLRQWRWS...................................................................... 3–18

;&RRUGLQDWHRI[[·WK3HDN[DSN .............................................. 3–19

&RQWHQWV

&KDSWHU³7UHQGLQJ

,QGH[

9LVXDOL]LQJ7UHQGV....................................................................... 4–1

7UHQG&RQILJXUDWLRQ0HQXV............................................................4–2

7UHQG&DOFXODWLRQ........................................................................ 4–6

LLL



:3,QWURGXFWLRQ

7KH9DOXHRI+LVWRJUDPVDQG7UHQGV

The value of histograms in data analysis and the
interpretation of measurement results is well known. The
WP03 option added to your oscilloscope provides this and
more for waveform parameter analysis. With WP03,
histograms and trends (
parameter measurements can be created, statistical
parameters determined, and graphic features quantified for
analysis.

Statistical parameters alone — such as mean, standard deviation
and median — are usually insufficient for determining whether
the distribution of measured data is as expected. Histograms
provide an enhanced understanding of the distribution of
measured parameters by enabling visual assessment of the
distribution. Observations based on the histogram of a param eter
can indicate:

À Distribution type: normal, non-normal, etc. This is helpful in

determining whether the signal behaves as expected.

À Distribution tails and extreme values , which can be obser ved

and may be related to noise or other infrequent and non­repetitive sources.

À Multiple modes, which can be observed and could indicate

multiple frequencies or amplitudes. These can be used to
differentiate from other sources such as jitter and noise.

see Chapter 4

) of waveform

+LVWRJUDPVRI3DUDPHWHU
0HDVXUHPHQWV

Generating histograms of wavef orm measurement parameters is
a three-step process:

1. Waveform parameters of interest are selected from the
“CURSORS/MEASURE” menu.

2. Histograms are selected and set up through the scope’s
“Math Setup” menu for the waveform parameter of interest.

3. Statistical parameters are selected for measurement of
histogram characteristics.

²

:3

+LVWRJUDP0DWK)XQFWLRQ Histograms of user-selected waveform parameters are created

using the scope’s Histogram Math function. This is done by
defining a trace (A, B, C, or D) as a math func tion, and selecting
“Histogram” as the function to be applied to the trace. As with
other traces, histogram s can be positioned and expanded using
the POSITION and ZOOM knobs on the instrument’s front panel.

Histograms are displayed based on a set of user settings,
including bin width and number of parameter events. Special
parameters are provided for determining histogram
characteristics such as mean, median, standard deviation,
number of peaks and most-populated bin.

This broad range of histogram options and controls provides a
quick and easy method of analyzing and understanding
measurement results.

The “MEASURE” “Parameters” menu is accessed by pressing
the CURSORS/MEASURE button, then selecting “Parameters”
from the top menu that appears, as shown in

Figure 1.1

.

Parameters are used to perform waveform measurements for
the section of waveform that lies between the parameter cur sors
(Annotation ➊ in this figure). The position of the parameter
cursors is set using the “from” and “to” m enus and controlled by
the associated ‘menu’ knobs.

The top trace in
parameter measurement is being performed on the waveform
(Annotation ➋) with a value of 202.442 kHz as the average
frequency. The bottom trace shows a histogram of the freq
parameter with an average frequency of 201.89 kHz (Annotation
➌), which is the average frequency of the data contained within
the parameter cursors.

Figure 1.1

shows a sine waveform. A freq

²

,QWURGXFWLRQ

1

2
3

Figure 1.1

Selection of “Custom” from the “mode” menu and then
“CHANGE PARAMETERS” displays the “CHANGE PARAM”
menu group, shown in
can be selected, with each displayed on its own line below the
waveform display grid. Parameter m easurements can then also
be selected from “Category” and “measure” using the
corresponding menu buttons.

Categories are provided for related groups of parameter
measurements. The “Statistics” category is provided for
selection of histogram parameters. After s election of a category,
a parameter can be selected from the “measure” menu.
Selection of parameters is done using the menu buttons or
knobs.

²

Figure 1.2

. Now, up to five parameters

:3

The parameter display line is selected from the “On line” menu.
In

Figure 1.2

À The freq measure param eter from the “Cyclic” category for

Trace 1, which had earlier been selected, is displayed on
Line 1 as freq() (

À The avg m easure parameter from the “ Statistics” category

for Trace A is selected for display on Line 2. The avg
parameter provides the mean value of the underlying
measurements for the Trace A histogram section within the
parameter cursors (

Annotation

(

À No parameters have been selected for Lines 3–5.

:

➌

).

Annotation

Annotation

➊

) .

➋

), shown as “avg($)”,

2

1

3

Figure 1.2

²

3DUDPHWHU9DOXH
&DOFXODWLRQDQG'LVSOD\

,QWURGXFWLRQ

If a parameter has additional settings that must be supplied in
order to perform measurements, the “MORE ‘xxxx’ SETUP”
menu appears. But if no additional settings are required the
“DELETE ALL PARAMETERS” menu appears , as shown in the
figure above, and pressing the associated menu button res ults in
all five lines of parameters being cleared.

not

When Persistence is
channels shows the captured waveform of a single sweep.

For non-segmented waveforms, the display is identical to a single
acquisition. But with segmented waveform s, the res ult of a single
acquisition for all segments is displayed.

The value displayed for a chosen param eter depends on whether
“statistics” is “On”. And on whether the waveform is segm ented.
These two factors and the param eter chosen determ ine whether
results are provided for a single acquisition (trigger) or multiple
acquisitions. In any case, only the waveform sec tion between the
parameter cursors is used.

If the waveform sourc e is a memory (“M1”, “M2”, “M3” or “M4”)
then loading a new waveform into mem ory acts as a trigger and
sweep. This is also the case when the waveform source is a
zoom of an input channel, and when a new segment or the “All
Segments” menu is selected.

being used, the display for input

When “statistics” is “Off”, the parameter results for the last
acquisition are displayed. This corres ponds to results for the last
segment for segmented waveform s with all segments displayed.
For zoom traces of segmented waveforms, selection of an
individual segment gives the parameter value for the displayed
portion of the segment between the parameter cur sors . Selection
of “All Segments” provides the parameter results f rom the last
segment in the trace.

not

When “On”, and where the parameter does
waveforms in calc ulating a res ult (∆dly, ∆t@lv), results are shown
for all acquisitions since the CLEAR SWEEPS button was last
pressed. If the parameter uses two waveforms, the result of
comparing only the last segment per s weep for eac h waveform
contributes to the statistics.

The statistics for the selected segm ent are displayed for zoom
traces of segm ented waveforms. Selection of a new segment or

²

use two

:3

“All Segments” acts as a new sweep and the parameter
calculations for the new segment(s) contribute to the statistics.

Depending on the parameter, single or multiple c alculations can
be performed for each acquisition. For example, the period
parameter calculates a per iod value for each of up to the first 50
cycles in an acquisition. When multiple calculations are
performed, with “ statistics” “Off” the parameter result shows the
average value of these calculations. W hereas “On” displays the
average, low, high and sigma values of all the calculations.

([DPSOH In

signal. The initial impression given the viewer is of some
frequency drift in the signal source. The lower trace shows a
histogram of the frequency as measured by the oscilloscope.

Figure 1.3

, the upper trace shows the persistence display of a

²

Figure 1.3

,QWURGXFWLRQ

This histogram indicates two frequency distributions with
dominant frequencies separated by 4000 Hz. There are two
distinct and normal looking distributions, without wide variation,
within each of the two. We can conclude that there are two
dominant frequencies. If the problem were related to frequency
drift, the distribution would have a tendency to be broader, non-

not

normal in appearance, and normally there would
distinct distributions.

After a brief visual analysis, the measurement cursors and
statistical parameters can be used to determine additional
characteristics of distribution, including the most common
frequency in each distribution and the spread of each distribution.

Figure 1.4

Annotation

(
bin of the distribution. The value of the bin, ins ide the Displayed
Trace Field (

Annotation

by

, below, shows the use of the measurement cursor

➊

), to determine the frequency represented by one

see Chapter 2 for a detailed description

➋

.

be two

) is indicated

²

12

Figure 1.4

:3

Figure 1.5

Annotations

(
the distribution located between the cursors. The average value
of the measurem ents in the right-hand dis tribution is indicated by

Annotation

3

, below, shows the use of the parameter cursors

➊

and ➋) in determining the average frequenc y of

➌

.

2

1

²

Figure 1.5

,QWURGXFWLRQ

Finally,
(
between a bin in the center of each distribution. The value in
k Hz, in the Displayed Trace Field, is indicated by

Figure 1.6

Annotations

3

1

shows the use of the measurement cursors

➊

and ➋) in determining the differ ence in frequenc y

Annotation

2

➌

.

²

Figure 1.6

2

Theory of Operation

A statistical understanding of variations in parameter
values is of great interest for many waveform parameter
measurements. Knowledge of the average, minimum,
maximum and standard deviation of the parameter may
often be enough for the user, but in many other instances a
more detailed understanding of the distribution of a
parameter’s values is desired.

Histograms provide the ability to see how a parameter’s values
are distributed over many measurements, enabling this detailed
analysis. They divide a range of parameter values into sub­ranges called bins. Maintained for each bin is a count of the
number of parameter values calculated — events — that fall
within its sub-range.

While the range can be infinite, for practical purposes it need
only be defined as large enough to include any realistically
possible parameter value. For example, in measuring TTL high­voltage values a range of ± 50 V is unnecessarily large, whereas
one of 4 V ± 2.5 V is more reasonable. It is this 5 V range that is
subdivided into bins. And if the number of bins used were 50,
each would have a sub-range of 5 V/50 bins or 0.1 V/bin. Events
falling into the first bin would then be between 1.5 V and 1.6 V.
While the next bin would capture all events between 1.6 V and

1.7 V. And so on.

WP03: Histograms

After a process of several thousand events, the graph of the
count for each bin — its histogram — provides a good
understanding of the distribution of values. Histograms generally
use the ‘x’ axis to show a bin’s sub-range value, and the ‘y’ axis
for the count of parameter values within each bin. The leftmost
bin with a non-zero count shows the lowest parameter value
measurement(s). The vertically highest bin shows the greatest
number of events falling within its sub-range.

The number of events in a bin, peak or a histogram is referred to
its population. Figure 2.1 shows a histogram’s highest population
bin as the one with a sub-range of 4.3–4.4 V — to be expected
of a TTL signal. The lowest value bin with events is that with a

2––1

WP03

sub-range of 3.0–3.1 V. As TTL high v oltages need to be greater
than 2.5 V, the lowest bin is within the allowable tolerance.
However, because of its proximity to this tolerance and the
degree of the bin’s separation from all other values, additional
investigation may be desirable.

LeCroy DSO Process LeCroy digital oscilloscopes generate histograms of the

parameter values of input waveforms. But first, the following
must be defined:

¾ The parameter to be histogrammed.
¾ The trace on which the histogram will be displayed.
¾ The maximum number of parameter measurement values to

be used in creating the histogram.

¾ The measurement range of the histogram.
¾ The number of bins to be used.

Once these are defined, the oscilloscope is ready to make the
histogram.

Count

40

30

20

10

1.5
2

3

3.15

Range

4.35

4

5

6

Volts

Figure 2.1

2––2

Histograms

The sequence for acquiring histogram data is:

1. trigger

2. waveform acquisition

3. parameter calculation(s)

4. histogram update

5. trigger re-arm.

If the timebase is set in non-segmented mode, a single
acquisition occurs prior to parameter calculations. However, in
Sequence mode an acquisition for each segment occurs prior to
parameter calculations. If the source of histogram data is a
memory, storing new data to memory effectively acts as a
trigger and acquisition. Because updating the screen can take
significant processing time, it occurs only once a second,
minimizing trigger dead-time (under remote control the display
can be turned off to maximize measurement speed).

Parameter Buffer The oscilloscope maintains a circular parameter buffer of the

last
20 000 measurements made, including values that fall outside
the set histogram range. If the maximum number of events to be
used in a histogram is a number ‘N’ less than 20 000, the
histogram will be continuously updated with the last ‘N’ events
as new acquisitions occur. If the maximum number is greater
than 20 000, the histogram will be updated until the number of
events is equal to ‘N’. Then, if the number of bins or the
histogram range is modified, the scope will use the parameter
buffer values to redraw the histogram with either the last ‘N’ or
20 000 values acquired — whichever is the lesser. The
parameter buffer thereby allows histograms to be redisplayed
using an acquired set of values and settings that produce a
distribution shape with the most useful information.

2––3

WP03

In many cases the optimal range is not readily apparent. So the
scope has a powerful range-finding function. If required it will
examine the values in the parameter buffer to calculate an
optimal range and redisplay the histogram using it. The
instrument will also give a running count of the number of
parameter values that fall within, below and above the range. If
any fall below or above the range, the range-finder can then
recalculate to include these parameter values, as long as they
are still within the buffer.

Parameter Events Capture The number of events captured per waveform acquisition or

display sweep depends on the parameter type. Acquisitions are
initiated by the occurrence of a trigger event. Sweeps are
equivalent to the waveform captured and displayed on an input
channel (1, 2, 3 or 4). For non-segmented waveforms an
acquisition is identical to a sweep. Whereas for segmented
waveforms an acquisition occurs for each segment and a sweep
is equivalent to acquisitions for all segments. Only the section of
a waveform between the parameter cursors is used in the
calculation of parameter values and corresponding histogram
events.

The following table provides, for each parameter and for a
waveform section between the parameter cursors, a summary of
the number of histogram events captured per acquisition or
sweep.

2––4

Histograms

Parameters

(plus others, depending on options)

data All data values in the region analyzed.

duty, freq, period, width, Up to 49 events per acquisition.

ampl, area, base, cmean, cmedian, crms,
csdev, cycles, delay, dur, first, last, maximum,
mean, median, minimum, nbph, nbpw, over+,
over–, phase, pkpk, points, rms, sdev, ∆dly,
∆t@lv

f@level, f80–20%, fall, r@level, r20–80%, rise Up to 49 events per acquisition.

Histogram Parameters Once a histogram is defined and generated, measurements can

be performed on the histogram itself. Typical of these are the
histogram’s:

Number of Events Captured

One event per acquisition.

¾ Average value, standard deviation

¾ Most common value (parameter value of highest count bin)

¾ Leftmost bin position (representing the lowest measured

waveform parameter value)

¾ Rightmost bin (representing the highest measured waveform

parameter value).

Histogram parameters are provided to enable these
measurements. Available through selecting “Statistics”from the
“Category” menu, they are calculated for the selected section
between the parameter cursors (for a full description of each
parameter, see Chapter 3):

2––5

WP03

All Segments

avg average of data values in histogram
fwhm full width (of largest peak) at half the maximum bin
fwxx full width (of largest peak) at xx% the maximum bin
hampl histogram amplitude between two largest peaks
hbase histogram base or leftmost of two largest peaks
high highest data value in histogram
hmedian median data value of histogram
hrms rms value of data in histogram
htop histogram top or rightmost of two largest peaks
low lowest data value in histogram
maxp population of most populated bin in histogram
mode data value of most populated bin in histogram
pctl data value in histogram for which specified ‘x’% of

population is smaller

pks number of peaks in histogram
range difference between highest and lowest data values
sigma standard deviation of the data values in histogram
totp total population in histogram
xapk x-axis position of specified largest peak.

Zoom Traces and
Segmented Waveforms

Histogram Peaks Because the shape of histogram distributions is particularly

Example

Histograms of zoom traces display all events for the displayed
portion of a waveform between the parameter cursors. When
dealing with segmented waveforms, and when a single
segment is selected, the histogram will be recalculated for all
events in the displayed portion of this segment between the
parameter cursors. But if “
histogram for all segments will be displayed.

interesting, additional parameter measurements are available
for analyzing these distributions. They are generally centered
around one of several peak value bins, known — together with
its associated bins — as a histogram peak.

In Figure 2.2, a histogram of the voltage value of a five-volt
amplitude square wave is centered around two peak value bins:
0 V and 5 V. The adjacent bins signify variation due to noise.
The graph of the centered bins shows both as peaks.

2––6

” is selected, the

Histograms

Volts

0

Figure 2.2

Determining such peaks is very useful, as they indicate
dominant values of a signal.

However, signal noise and the use of a high number of bins
relative to the number of parameter values acquired, can give a
jagged and spiky histogram, making meaningful peaks hard to
distinguish. The scope analyzes histogram data to identify peaks
from background noise and histogram definition artifacts such as
small gaps, which are due to very narrow bins.

5

Binning and
Measurement
Accuracy

For a detailed description on how the scope determines peaks see
the pks parameter description, Chapter 3.

Histogram bins represent a sub-range of waveform parameter
values, or events. The events represented by a bin may have a
value anywhere within its sub-range. However, parameter
measurements of the histogram itself, such as average, assume
that all events in a bin have a single value. The scope uses the
center value of each bin’s sub-range in all its calculations. The
greater the number of bins used to subdivide a histogram’s
range, the less the potential deviation between actual event
values and those values assumed in histogram parameter
calculations.

Nevertheless, using more bins may require performance of a
greater number of waveform parameter measurements, in order
to populate the bins sufficiently for the identification of a

2––7

WP03

characteristic histogram distribution.

In addition, very fine-grained binning will result in gaps between
populated bins that may make determination of peaks difficult.

Figure 2.3 shows a histogram display of 3672 parameter
measurements divided into 2000 bins. The standard deviation of
the histogram sigma (Annotation ) is 81.17 mV. Note the

histogram’s jagged appearance.

1

Figure 2.3

The oscilloscope’s 20 000-parameter buffer is very effective for
determining the optimal number of bins to be used. An optimal
bin number is one where the change in parameter values is
insignificant, and the histogram distribution does not have a
jagged appearance. W ith this buffer, a histogram can be
dynamically redisplayed as the number of bins is modified by
the user. In addition, depending on the number of bins selected,
the change in waveform parameter values can be seen.

2––8

Histograms

In Figure 2.4 the histogram shown in the previous figure has
been recalculated with 100 bins. Note how it has become far
less jagged, while the real peaks are more apparent. Also, the
change in sigma is minimal (81.17 mV vs 81 mV).

2––9

Figure 2.4