May 1967 Electronics World
Wax nostalgic about and learn from the history of early electronics. See articles
Electronics World, published May 1959
- December 1971. All copyrights hereby acknowledged.
Some things never change, and the basic definition of a pulse
waveform is one of them. This article form the May 1967 edition
of Electronics World does a nice job of defining the fundamental
characteristics of a pulse, including rise and fall times, overshoot,
pulse width, etc. Ham shacks and company labs alike are still
stocked with the venerable
HP 215A and
HP 213B signal generators. If you're lucky, you can get
a good deal on them from someone on eBay.
Selecting and Using Pulse Generators
By John Lenk
Basically a laboratory version of the square-wave
generator, but with adjustable on-off times, this instrument
has many uses in developing digital circuitry, in checking diode
and transistor switching times, as a klystron modulator, and
for impulse testing.
The output of a pulse generator
is similar to that of a square-wave generator. The fundamental
difference between the two concerns the signal duty cycle. Square-wave
generators have equal "on" and "off" periods, equality being
retained as the repetition frequency is varied. On the other
hand, the duration of a pulse generator "on" period is independent
of pulse repetition rate. The duty cycle of a pulse generator
can be made quite low that the pulse generator is usually able
to supply more power during the "on" period than a conventional
Pulse generators with fast rise
times are widely used in the development of digital circuitry.
Teamed with suitably fast oscilloscope, these generators enable
evaluation of transistor and diode switching times. Pulse generators
can be employed as modulation sources for klystrons and other
r.f. sources to obtain high peak power while maintaining low
average power. Pulse generators are also used for impulse testing.
A very short pulse is rich in harmonics so that input testing
amounts to simultaneous frequency-response testing of components
describe the characteristics of a pulse generator, it is first
necessary to establish uniform terms for pulses. These terms
are illustrated in Fig. 1. When actual pulses are very irregular
(with excessive tilt, over-shoot, or rounding), the definitions
may become ambiguous, requiring a more complete description.
The following are typical characteristics of a laboratory
Fig. 1. Terms used in describing output pulse
Leading Edge Only:
Rise time (Tr): <1.0 nanosecond (ns) (10 to
Overshoot and ringing: overshoot <5% peak; ringing <±5%
of pulse amplitude.
Corner rounding: occurs no sooner than
95% of pulse amplitude.
Time to achieve flat top (Ta): <6
Trailing Edge Only:
(Tf): <1.0 ns (10 to 90%).
Rounding: occurs no sooner than 95% of fall.
settle within 2% of baseline (Tb): 10 to 25 ns, varies
Baseline shift: <
01 % under all conditions.
Perturbations on flat top: <2% of pulse amplitude.
Peak voltage: >10 volts into 50 ohms, >20 volts
into open circuit.
Polarity: positive or negative.
Pulse width (between 50% points): continuously adjustable,
zero to 100 ns (zero ns width occurs when 50% points meet, creating
an impulse of one-half the amplitude of wide pulses).
Repetition rate (internal): <100 Hz to >1 MHz in 4
The quality of the output pulse is of primary importance
in the selection of a pulse generator. If the displayed pulse
is degraded, a high-quality test pulse will insure that the
cause is in the test circuit alone. Rise and fall times should
be significantly faster than the circuits or systems to be measured.
Any overshoot, ringing, or sag in the test pulse should be known
so that these faults will not be confused with similar results
caused by the test circuit.
Hewlett-Packard 215A pulse generator is shown.
The range of pulse-width control should be wide enough to
fully explore the range of operation of a circuit. Narrow pulse
widths are useful in determining the minimum trigger energy
required in some circuits.
HP-213B produces pulse with under 0.1-nanosecond
Maximum pulse amplitude is of prime concern if appreciable input
power is required by the circuit under test, such as a magnetic
core memory. At the same time, the attenuation range should
be broad enough to prevent over-driving the test circuits as
well as to simulate actual circuit operating conditions.
The range of pulse repetition rates is important if the
tested circuits can operate only within a certain range of pulse
rates or if a variation in the rate is needed. In some systems,
methods of external triggering are also significant. In fast
pulse systems, the generator source impedance is an important
consideration because a generator which has a source impedance
that is matched to the connecting cable will absorb reflections
resulting from impedance mismatches in the external system that
1. Use proper types
of cables, terminations, attenuators, and impedance-matching
networks. Always match impedances unless the test circuit specifically
calls for a mismatch.
2. Keep ground-return paths short
and direct. Use heavy conductors to provide low impedance in
the ground return.
3. Make sure that all connections
are tight and that all connectors are securely assembled.
4. Shield measuring-equipment leads to prevent undesired
coupling to other parts of the circuit. Shielding is especially
required where pulse radiation is a problem and particularly
where high-impedance dividers or circuits are involved.
5. Consider the effects of secondary parameters in components,
such as inductance in resistors and in capacitor loads.
6. Consider the possible non-linear behavior of components
due to changes in either voltage or temperature.
7. Select components which function properly at the frequencies
and rise times expected to be encountered.
Fig. 2. Resistive impedance-matching network
and formulas. (upper)
Fig. 3. Three-way dividing pad circuit for
Z measurement. (lower)
Obviously, the accuracy of rise-time measurements can be
no greater than the rise time of the pulse generator. If a pulse
generator with a 20-nanosecond rise time is used to measure
the rise time of a 15-nanosecond oscilloscope, the measurements
would be hopelessly incorrect. Also, if the same pulse generator
and oscilloscope were used to measure the rise time of another
system, the fastest rise time for accurate measurement would
be something greater than 20 nanoseconds.
As a general rule, if the rise time of the test device is
at least ten times as long as the rise times of the generator,
oscilloscope, or cables, the error introduced will not be more
than 1%. If the rise time of the device under test is less than
ten times that of the test equipment, it will be necessary to
calculate the rise time. The most common method involves finding
the square of all rise times associated with the test, adding
these squares together, and then computing the square root of
this sum. For example, using the 20-ns pulse generator and the
15-ns oscilloscope, the calculation would be: 20 x 20 = 400;
15 x 15 = 225; 400 + 225 = 625. √625 = 25, so 25 nanoseconds
is the fastest possible rise time capable of measurement.
Another rule of thumb applying to rise times is that
if the equipment being measured has a rise time three times
slower than the test equipment, the error is only slightly less
If there are significantly long lengths of
coaxial cable in the signal path, the above method can be used
only as an approximation, since the "skin-effect" losses in
coaxial cables do not add properly with this method.
Connecting Pulse Generators
1. In most measurements
involving pulse generators, a complete d.c. return path must
be provided between the device under test and the pulse-generator
2. If the pulse is applied to a load
which has a d.c. potential across it, the actual amplitude of
the pulse is equal to the voltage set by the pulse-generator
amplitude control less one-half the d.c. voltage across the
For example, assume that the pulse-generator output
is connected to a load which has +10 volts across it and that
the pulse-generator amplitude control is set to +1 volt. The
actual amplitude is found by substituting these values as follows:
Va = Vs - (Vl/2) or +1 - (+10/2)
= -4 volts where Va is the actual pulse amplitude,
Vs is the voltage setting of the amplitude control,
and Vl is the d.c. voltage applied across the load.
3. If it is impossible to use an impedance-matching
network, one possible solution is to employ a long coaxial cable
between generator and load. This will delay the load's reflections
until after the time of interest.
output can be supplied with an impedance-matching network that
will produce a smooth transition of power (no reflections) with
a minimum attenuation. Such a network is shown in Fig. 2. To
match impedances with the illustrated network, the values of
R1 and R2 must be selected carefully.
For example, to
match a 50-ohm system to a 125-ohm system, Z1 = 50 ohms and
Z2 = 125 ohms. Therefore, R1 = √125(125 - 50) = 96.8 ohms, and
R2 = 50 √25/(125 - 50) = 64.6 ohms.
as seen from one end of the network does not equal that seen
from the other end. Using the equations shown in Fig. 2, it
will be noted that a signal applied from the lower impedance
source Z1 encounters a voltage attenuation A1. Also, a signal
applied from the higher impedance source Z2 will encounter a
greater voltage attenuation A2.
For example, with an
R1 of 96.8 ohms and an impedance Z2 of 125 ohms, A1 = (96.8/125)
+ 1 = 1.77.
With an R1 of 96.8 ohms, an R2. of
64.6 ohms. and an impedance Z1 of 50 ohms, A2 = (96.8/64.6)
+ (96.8/50) + 1 = 4.44.
A pulse generator can he used to determine impedance of an unknown
device by comparing the reflected pulse with the incident pulse
on an oscilloscope. This can be explained as follows.
As a signal travels down a transmission line, each time
it encounters a mismatch or different impedance, a reflection
is generated and sent back along the line to the source. The
amplitude and polarity of the reflection arc determined by the
value of the impedance encountered in relation to the characteristic
impedance of the cable. If the mismatch impedance is higher
than that of the line, the reflection will be of the same polarity
as the applied signal; if it is lower than that of the line,
the reflection will be of opposite polarity.
signal is added to or subtracted from the amplitude of the pulse
if it returns to the source before the pulse has ended. Thus,
for a cable with an open end (no termination), the impedance
is infinite and the pulse amplitude would be doubled. For a
cable with a shorted end, the impedance is zero and the pulse
would be canceled.
The following procedure provides
a practical method of determining impedance with a pulse generator
1. Connect the equipment as shown in
2.Observe the incident and reflected pulses
on the oscilloscope. Using Fig. 4 as a guide, determine the
values of V0 (incident) and Vx (reflected).
(This method is generally limited to the first reflections unless
the deviations are small, due to multiple reflections and reflection
4. Using the following equation, calculate the unknown impedance.
Z = 50/(2V0/Vx - 1) where Z is the unknown
impedance, V0 is the peak amplitude produced by the
50- ohm reference impedance, and Vx is the peak amplitude
at the time of reflection.
Fig. 4. Waveform obtained with 125-ohm
cable and 50-ohm system. (upper)
Fig. 5. Test connections using generator
with conventional scope. (lower)
Using Conventional Oscilloscopes
A pulse generator is often used with a sampling oscilloscope,
and generator and oscilloscope manuals describe the procedure.
However, a pulse generator can also be used with conventional
triggered oscilloscopes. Fig. 5 shows the test connections.
Internal triggering is convenient since no external
triggering connections are required. However, with external
triggering it is possible to observe the shaping and amplification
of a signal pulse in the circuits of a device under test without
resetting the oscilloscope triggering controls for each observation.
If the external triggering signal is derived from the waveform
at the input circuit of the device under test, the time relationship
and phase between the output and input waveforms may be seen
and compared on the oscilloscope screen.
If the signal
from the test device is fast-rise non-repetitive or has a low
duty cycle, the oscilloscope used in this setup must have an
internal delay line so that the leading edge of the single waveform
can be readily observed on the scope.
One of the drawbacks
to a conventional oscilloscope is that the frequency response
of the test device may fall outside the bandwidth limitations
of the vertical amplifier system of the oscilloscope. In some
cases, the output signal from a device under test can be observed
by direct connection through coupling capacitors to the vertical
deflection plates of a conventional oscilloscope. Thus, the
limited bandwidth of the oscilloscope vertical amplifier can
The following factors pertaining to the vertical deflection-plate
system must be considered for pulse measurement: d.c. operating
potential of the plates, lead inductance, deflection-plate capacitance,
transit-time limitations, delay lines, and deflection factor.
Tektronix Type 109 pulse generator has rise
time of under 0.25 nanoseconds.
A typical circuit for direct a.c. coupling to the vertical
plates is shown in Fig. 6. This circuit permits the internal
vertical amplifier of the oscilloscope to be bypassed but still
allows the normal d.c. operating and positioning voltages to
be applied to the deflection plates from the internal vertical
amplifier. However, when using this circuit, a high-quality
external delay line must be used. This will retard the pulse
sufficiently to get it on the scope screen.
The values of R1 and R2 are found by solving the equation
given in Fig. 6. The resonant frequency (F0) of the
leads and the capacitance of the deflection plates (CD)
for use in the equation may be determined by the following procedure:
1. Turn off the oscilloscope power.
the vertical amplifier leads from the CRT neck pins. (A convenient
method of connecting to the deflection-plate pins is to use
clips removed from a miniature tube socket.)
a wire loop equal in length to the total length of C1, C2, R1,
R2, R3, and R4.
4. Temporarily substitute the wire loop
for the components between the vertical deflection-plate pins.
5. Bring a grid-dip meter near the loop and measure
the resonant frequency (F0)
6. Remove the wire loop.
7. With a capacitance meter,
measure the total capacitance between the plates (CD)
at the deflection-plate neck pins. (Capacitance between the
plates can also be found by referring to the specifications
of the oscilloscope.)
Tektronix Type R116 pulse generator.
Since the deflection plates are located close to the path
of the electro beam, a small amount of current will flow in
the deflection-plate circuits The values of R3 and R4 must be
low enough so that this current will not produce a large voltage
drop at the deflection plates. If the resistors are too large,
distortion, defocusing, or positioning difficulties may be experienced.
Since the deflection-plate current varies non-linearly with
the position of the beam, the effects are most noticeable when
the beam is positioned near the top or bottom of the screen.
The approximate value of 100,000 ohms that is given for R3 and
R4 will probably be satisfactory in most cases.
C1 and C2 should be physically small to minimize lead inductance.
The values of C1 and C2 are selected on the basis of the required
low-frequency response and may be calculated from the equation
given in Fig. 6. (Fc is the low-frequency cut-off.)
For example, if R3 and R4 are 100,000 ohms and if the desired
Fc is about 1.6 kHz. C1 and C2 should be 0.001 µF.
The stub cable that connects to terminating resistor
R0 should be long enough so that if a double-transit
reflection appears, it can be easily identified and corrected
by adjustment of the termination.
For making vertical measurements with the test setup, the
deflection factor of the oscilloscope must be known. This can
be measured as follows:
1. While the leads from the
vertical amplifier are connected to the deflection-plate neck
pins, connect a d.c. voltmeter between the pins.
Measure the voltage change as the beam is positioned vertically
over the full height of the graticule.
this voltage excursion by the graticule height in divisions
to obtain the deflection factor in volts/division.
(Many of the diagrams and techniques described above are based
on information from Tektronix, Inc. and Hewlett-Packard. -Editor)
Fig. 6. Circuit for coupling to vertical