Making the right choice in your selection of a signal generator

Cindy Ong, National Instruments Corp.

July 29, 2007

Cindy Ong, National Instruments Corp. July 29, 2007

Signal generators come in many different forms. The most prevalent types of signal generators include arbitrary waveform, function and RF signal generators, as well as basic analog output modules. These types vary in their features and functionality, and are suitable for many different applications.

Arbitrary waveform generators (AWGs) generally provide deep memory, wide dynamic range and high bandwidth to meet the demands of applications such as communications and IC component and system tests. AWGs receive user-defined data from a PC and use this data to generate arbitrary waveforms. An AWG user can download a list of waveforms that they want to generate into the onboard memory of the AWG. Often, both the actual waveforms and the waveform sequencing instructions that play them are stored onboard.

To generate a waveform from an AWG, the arbitrary waveform itself must first be created. The generation sequence often begins with a TTL hardware trigger. Waveforms are built of individual samples, and the generation sample rate is determined by the onboard sample clock.

Figure 1: Waveform passes through memory to a DAC, which translates digital samples into the desired analog output.

There are modes for deriving the sample clock from the internal sample clock timebase (100MHz VCXO), including direct digital synthesis (DDS) and Div/N clocking, as well as modes to provide external clocks. There are several choices for providing the frequency reference for the onboard phase-locked loop.

The waveform passes through memory to a DAC (Figure 1 above), which translates digital samples into the desired analog output waveform. Before the DAC, samples are digitally filtered, after which the analog waveform is passed through an analog filter. These digital and analog filters improve signal quality by increasing the effective sample rate through interpolation and removing spurious signals through harmonics low-pass filters. Most often, these filters are software-programmable.

Waveform looping
AWGs allow you to specify waveform segments that the AWG can repeat to construct complex waveforms. Because AWGs store waveforms in on-board memory, the length of the waveform is limited. Waveform looping helps generate signals with subcomponents that repeat many times. Looping a waveform segment improves memory efficiency and increases the potential duration of the waveform.

AWGs can also specify waveform stages that each consist of a waveform segment and looping information. They generate each defined waveform stage sequentially. By combining sequencing and looping, you can construct highly complex waveforms using minimal memory. AWGs can specify different waveform segments for each stage, although the transitions are not necessarily phase-continuous.

Finally, many AWGs have an emulated function generator capability. In this case, when asked to output a standard function waveform, it will be created in software, downloaded to the AWG and played. This is different from a full DDS technology. Function generators create built-in waveforms such as sine, square or triangle waves at adjustable frequencies. They do not require continuous input from the computer or large memory buffers because the device dynamically generates waveforms.

Function generators can be either analog- or digital-based. Analog-based function generators use analog hardware to create simple functions and are often used when an application calls for a static sine or square wave at a specified frequency.

Digital-based function generators use DDS, a DAC, signal processing and a one-cycle memory buffer to dynamically create signals. DDS is a technique for deriving under digital control an analog frequency source from a single reference clock frequency. DDS produces high-frequency accuracy and resolution, temperature stability, wideband tuning, and raid, phase-continuous frequency switching.

Figure 2: In a DDS function generator, one complete cycle of the function waveform is stored in the memory LUT.

'Divide-by-N'
Many signal generators create clock signals by dividing an internal timebase by an integer factor. This is called the divide-by-N method. However, this gives a limited set of clock frequencies. AWGs, and even several clock frequency generators, can use DDS to generate clock signals at very specific update frequencies not available by divide-by-N clocking.

One complete cycle of the function waveform is stored in the memory lookup table (Figure 2 above). The phase accumulator keeps track of the current phase of the output function. To output a very slow frequency, the Delta phase, between samples would be very small. For example, a slow sine may have a delta phase of 10.

Sample 0 of the waveform would be the amplitude of the sine wave at 00, sample 1 of the waveform would be the amplitude of the sine wave at 10, and so on. All 3600 of the sinusoid, or exactly one cycle, would be output after 360 samples. A faster sine wave may have a ?phase of 100. Here, one cycle of a sine wave would be output in 36 samples. If the sample rate were constant, the slow sine wave would be 10 times slower in frequency than the fast sine wave.

Table 1: Signal generator types vary in their features and functionality, and are suitable for many different applications.

Furthermore, a constant Delta phase would entail a constant sine wave frequency output. However, DDS technology allows users to quickly change the Delta phase of the signal through a frequency list. Function generators can specify a frequency list containing stages that each consist of waveform frequency and duration information.

They generate each defined frequency stage sequentially. By creating a frequency list, you can construct complex frequency sweeps or frequency- hopping signals. DDS allows function generators to make phase-continuous transitions from one stage to the next.

Vector signal generators offer a highly flexible and powerful solution for scientific research, communications, consumer electronics, aerospace/defense and IC test applications as well as for emerging areas such as softwaredefined radio, RFID and wireless sensor networks.

Cindy Ong is Marketing Engineer at National Instruments Corp.To read a PDF version of this story, go to "Choose the right signal generator for your apps."

Loading comments...