Generating frequency agile and custom waveforms

Keysight Technologies Australia Pty Ltd
By Roger L Jungerman*
Saturday, 11 February, 2006


There is considerable interest in generating complex waveforms to test radar and communications systems. An arbitrary waveform generator (AWG) is one of the most flexible methods of signal simulation. Any signal that can be mathematically described can be output from the AWG.

The waveform description can be performed using common mathematical tools (such as Matlab), or by using more customised software, such as the Agilent N7509A waveform generation toolbox used for wideband signal simulation.

The AWG output signal waveform can be applied to evaluate portions of a system transmitter.

For example, a dual-channel generator can be used to drive the I and Q baseband inputs of a microwave I/Q modular up-converter. Alternatively, the baseband or up-converted output of the waveform generator can be used to evaluate the performance of complex radar or communications receivers.

To be of the most use in these test applications, the arbitrary waveform generator should have three major characteristics:

  • Accurate analog output performance;
  • Flexible sequencer to store and extend mathematical descriptions of the waveform;
  • Powerful and easy-to-use waveform generation software.

The signal quality of a waveform produced by an AWG is only as good as the intrinsic performance of the internal digital-to-analog converter (DAC). Nyquist sampling limitations together with the need to provide realisable, high-performance reconstruction filters (to eliminate out-of-band aliased signal content) limit the upper frequency of an AWG.

Typically, the maximum modulation frequency is ~40% of the sample rate. With ever increasing modulation bandwidth requirements in today's systems, higher DAC sample rates are required.

At higher sample rates, the dynamic range - also expressed as the number of effective bits - typically decreases.

Several factors can lead to degradations in the spurious free dynamic range (SFDR) including: harmonic distortion and clock or other unrelated spurious signal content. In addition, the noise performance, often expressed as phase noise, usually degrades at higher sample rates. There are exceptions.

At a 1.25 GS/s sample rate, the Agilent N6030A provides 15 bit resolution with -65 dBc SFDR, which corresponds to ~11 effective bits. This is adequate dynamic range to accurately render most complex radar simulation waveforms.

Even with carefully designed hardware, there are basic physical limitations in waveform generation use a DAC. These include the sin x/x roll-off due to zero-order hold in the DAC and deviations from linear phase in the reconstruction filters.

Correction software can be used to remove the effects of these and other residual linear impairments in the AWG output. A correction algorithm is applied to the waveform on download, based on factory measurements of the overall linear system response of the AWG. Without this correction software, a mathematical description of a waveform will differ significantly from the observed analog output. For example, on the N6030A, without corrections the overall roll-off versus frequency is ~6 dB, up to the maximum modulation frequency of 500 MHz.

Figure 1 shows a multi-tone waveform spanning 500 MHz of baseband frequency, with corrections applied on the N6030A Multi-tone signals are often used for noise power ratio (NPR) measurements in satellite receiver testing.

The correction software results in an amplitude flatness of better than 0.5 dB, as shown. The phase response is not observable in this measurement, but is also corrected to ~2°.

Secure radios often use frequency hopping to reduce the probability of intercept. An AWG can be used to create this time varying signal. In addition, other signals such as chirped and conventional radar pulses have distinct characteristics in the time domain.

Signals such as these can be created on an AWG in several ways. The desired waveform can be derived mathematically and written once to the waveform memory. One limitation of this approach is that while the AWG memory is fairly large (16 MSamples, in the case of the N6030A) the sample rate is also very high, 1.25 GS/s.

This results in only 12.8 ms of unique playtime. To 'compress' the signal waveform data and extend the playtime, a sequencer is used. Sections of the waveform memory can be repeated and looped to create long scenarios with rich content.

In addition, multiple AWG channels can be synchronised together to create phased array signal test suites and external triggering can be used to control the evolution of the sequence scenario.

In considering the waveform sequencer it is informative to describe some of the limitations and requirements to successfully employ it:

  • Waveforms are read from memory in fixed granularity segments. In the case of the N6030A, this is eight samples per segment. If jumps in the waveform being played are requested by the sequencer, they occur on the segment boundary;
  • When create CW tones in the sequencer by repeating a fixed number of cycles, phase continuity is required at the repetition boundary or spurious energy will be produced at the loop repetition rate;
  • Considerable memory can be saved in producing pulsed radar scenarios by looping the 'off' time between pulses and conserving valuable waveform memory.

In Figure 2, a multi-tone scenario is shown. This can be produced (as shown) as a single waveform. Alternatively, it can be constructed with the sequencer. A limited number of cycles at each frequency can be looped and the different frequency segments cascaded in time to produce the same scenario with significantly less waveform memory.

In summary, the application of an arbitrary waveform generator to produce extremely flexible testing scenarios has been shown. With high-performance hardware, realistic signal simulations can be modelled and played with minimal distortion.

These test suites can be used to characterise a wide variety of transmit-ters and receivers, both at baseband and with up-conversion to RF and microwave frequencies.

* Roger is an R&D design engineer with Agilent Technologies' Advanced Products Operation in Santa Rosa, California, and is the hardware architect of the N6030A arbitrary waveform generator.

Related Articles

Using AI to build your next-gen wireless system

The central challenge engineers face when designing wireless systems and networks is their...

Ensuring 5G network performance

To be successful in business-critical use cases, the wireless networks need to be as reliable as...

Ubiquitous connectivity is the future of wireless

Systems capable of seamlessly using satellite, cellular and local area networks are nowadays a...


  • All content Copyright © 2024 Westwick-Farrow Pty Ltd