Pulse Spectrum Analysis
- details and method for using a spectrum analyser to measure pulsed signals and pulses..
Pulse signals are being used increasingly for a variety of radio and other RF applications.
As a result, spectrum analysis of these pulse signals is required.
Traditionally pulse spectrum analysis techniques and approaches are normally aimed at steady analogue RF signals. However pulse spectrum analysis demands a little understanding of the signals being analysed, and this can enable additional information to be gained.
Pulse signals take a variety of forms, but have a number of common traits. As a result it is possible to apply these to pulse spectrum analysis techniques. Accordingly the basics of pulse signals will be analysed.
The basic waveform of a pulse is modulated onto the RF carrier. A look at the nature and make-up of the pulse waveform will provide an understanding the spectrum of the modulated pulsed waveform.
The basic pulse waveform is shown in the diagram below. It has a repetition time of T and the pulse duration of t.
Basic pulse waveform
Using Fourier analysis it can be seen that this waveform is made up from a fundamental and harmonics. The basic waveform of a square wave can be made up from the fundamental sine wave with the same repetition rate as the square wave and then odd harmonics with the amplitudes of the harmonics inversely to their number.
A rectangular pulse is just an extension of this basic principle. The different waveform shape is obtained by changing the relative amplitudes and phases of harmonics, both odd and even.
Harmonics making up a pulse waveform
These base-band signals can then be plotted and the amplitudes and phases of the infinite number of harmonics, both odd and even result in the smooth envelope shown below.
Spectrum of a perfectly rectangular pulse
The envelope of this plot follows a function of the basic form:
This single can then be modulated onto an RF waveform to give a spectrum. As the harmonics of the baseband signal, extend out to infinity, so too do the sidebands of the modulated signal. In reality, however, the bandwidth will never be infinite and the harmonics, especially higher order ones are attenuated. Although this results in some distortion of the signal, the levels are generally acceptable.
Spectrum of a pulse waveform modulated onto an RF carrier with phase inversions
Pulse spectrum analysis
It has been possible to see how pulse signals are generated and the resulting spectra. While the phase of the sidebands is accommodated on the plots above, spectrum analysers are scalar test instruments and do not normally give an indication of the phase of a signal. Accordingly the plots from spectrum analysers are only shown "above the line.".
Spectrum of a pulse waveform modulated onto an RF carrier
There are a number of points can be noted for this:
- Spectra lines: The individual spectra lines shown on the graph of the modulated waveform are separated by a frequency equal to 1/T.
- Nulls in envelope: The nulls in the envelope or overall shape of the spectra occur at intervals of 1/t. Further nulls occur at n / t
- Envelope null distinctness: The nulls in the pulse spectrum shape are not always particularly distinct because of the finite rise and fall times in the modulating signals and the resulting asymmetries that exist.
Sometimes the issue of pulse desensitisation is referred to in terms of pulse spectrum analysis. The issue is that when the modulation is applied to the carrier, the peak level of the envelope is reduced, appearing that the signal has been reduced in overall power.
The apparent reduction in peak amplitude occurs because adding the pulse to the signal and modulating it with a square wave results in the power being distributed between the carrier and the sidebands. As the level of the modulation increases, so does the level of the sidebands. As there is only limited power available and each of the spectral components, i.e. carrier and sidebands, then contains only a fraction of the total power.
The overall effect as seen on a spectrum analyser is that the peak power reduces, but it is spread over a wider bandwidth.
It is possible to define a pulse desensitisation factor α. This can be described in the equation:
It should be noted that this relationship is only really valid for a true Fourier line spectrum. For this to be applicable the resolution bandwidth of the analyser should be < 0.3 PRF.
The average power of the signal is also dependent on the duty cycle as the power can only be radiated when the signal is in what may be loosely termed the "ON" condition. This can be defined by the equation below:
α = Pulse "desenitisation factor T = pulse repetition rate
PRF = Pulse Repetition Frequency (1 / T)
t = pulse length
teff = effective pulse length taking account of rise and fall times
Pavg = Average power over a pulse cycle
Ppeak = Peak power
Triangular and trapezoidal waveforms
While pulse spectrum analysis is normally applied to square or rectangular waveforms, similar principles also apply to triangular and trapezoidal waveforms.
The format of the waveform has many similar characteristics to those of a pulse waveform but with different levels of the different constituent signals and hence the sidebands.
It is therefore possible to analyse these waveforms in a similar way.
Pulse spectrum analysis measurement tips
When looking at a pulsed signal using a spectrum analyser it is necessary to employ techniques to ensure that the signal is displayed to reveal the aspects that are required.
Some of the chief aspects are:
- Measurement bandwidth less than line spacing: To resolve the individual spectral lines, the measurement bandwidth must be small relative to the offset of the lines, i.e. Bandwidth < 1 / T. If the measurement bandwidth is reduced further, them the spectral lines will retain their value (as expected) but the noise level will be reduced, although measurement time will be longer.
- Measurement bandwidth between line spacing and null spacing : The next stage occurs when the measurement bandwidth is greater than the spectral line spacing, but less than the null spacing. For this condition the spectral lines are not resolved and the amplitude height of the envelope depends upon the bandwidth. This is because a greater number of spectral lines, each with their own power contribution re contained within the measurement bandwidth. For this case 1 / t > B > 1 / T.
- Measurement bandwidth greater than null spacing: For this case where the measurement bandwidth is greater than the null spacings on the signal spectrum envelope, i.e. B > 1 / T, the amplitude distribution of the signal cannot be recognised.
With pulse transmission being widely used, pulse spectrum analysis is an important element of characterising and testing any equipment that is developed and the signals they produce.
By Ian Poole
Want more like this? Register for our newsletter