Instrumentation & Measurement Magazine 23-2 - 18

epoch (only the pulse visible on the screen), which makes the
frequency resolution very coarse, and the correction is not easily applicable.


sin (π fT )

⋅ 10 6  [dBμV/MHz],
S ( f ) = 20 log  2 A ⋅ k ATT ⋅ ΔT ⋅


π fT


(6)

Fig. 7. Measurement of pulse amplitude and duration, pulse-modulated RF
generator.

The pulse repetition rate should be stable, and the correction
factor should not change with CW frequency. An example
of the pulse modulated CW signal with frequency 5 MHz is
shown in Fig. 7 (note the nonzero rising and trailing edge of the
pulse due to the modulator non-idealities).
The method is also applicable for base-band pulse generators. The pulse shape must be very close to rectangular, which
is not always true for the measured IGUU2916 main generator
(especially band C/D). The pulse amplitude and duration are
measured using an oscilloscope. The pulse shape in the time
domain is generally distorted by the transfer function of the
cable + attenuator(s) and the oscilloscope transfer function.
To remove this effect correctly, one has to perform deconvolution of the transfer function in the frequency domain. The
pulse shape, however, was acquired using a short oscilloscope

where A is the sum of voltage samples in the pulse trace in μV;
ΔT is the time resolution (sampling time) in s; T is the total duration of the pulse in s; f is the frequency in Hz; and kATT is the
attenuation of the cable + attenuator between the generator
and oscilloscope. An example of the IGUU2916 pulse shape in
the C/D band is shown in Fig. 8 for changing the pulse repetition rate.
In Fig. 8, the pulse shape in band C/D is distorted; moreover, it contains more noise than in the case of the pulses in
bands A and B (not shown). The results are listed in Table 4.
They do not correspond very well with the results of the Fourier transform method or the IF pulse measurement method.
The measurement uncertainty mainly comprises the cable +
attenuator correction, the impedance mismatch and type A
uncertainty. It is obvious that this method is better suitable for
pulse-modulated RF generators and not the base-band pulse
generators like IGUU 2916.
The reason for the large deviation of some results compared with the results of other methods may be the presence
of noise and undershoots or overshoots in the pulse trace. It
can be shown that the contribution of these components to the
total surface in V·s is significant. The signal is noisy, and especially in band C/D, the error can be reduced by summing over
only those samples, which correspond to the pulse (the choice
of samples may be subjective).

Fig. 8. Pulse shape for the band C/D of the IGUU2916 Main generator. (a) With changing pulse repetition rate; (b) Detail of the pulse maximum on the right
(amplitude without correction for the cable and attenuators).
18

IEEE Instrumentation & Measurement Magazine

April 2020

Instrumentation & Measurement Magazine 23-2

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 23-2