Arbitrary Waveform Generators (AWGs) are instruments that generate signals from numeric waveform descriptions stored in the AWG’s memory. Selected samples of the waveform are sent to the AWG’s Digital to Analog Converter (DAC) and then, with appropriate filtering and signal conditioning, are output as an analog waveform.
Modern AWGs have high sampling rate (50 GSa/s and above), high bandwidth (>15 GHz), and deep memory (>1 GSa). When a junior engineer encounters such an advanced AWG, it can seem like magic: Just calculate or even draw a waveform with all the desired properties – jitter, noise, crosstalk – and literally connect the dots, plug it in, and press play. But once the engineer encounters the devilish details, things can go sour; from magic to impossible in a few hours.
Example 1. Assume that we want to emulate a 7 GBd signal, for instance to test a receiver. Using an AWG with high sampling rate (65 GSa/s), decent amplitude resolution (8 bits), and deep memory (2 GSa), we generate a signal in a few simple steps and verify the eye diagram with an oscilloscope immediately at the AWG output (see Figure 1). When we see a signal like that shown in Figure 1, we may reasonably be disappointed. These AWGs are either not so good or not so easy-to-use!
Figure 1. Eye diagram for a 7 GHz signal improperly generated with an AWG. The eye is distorted (note the prominent ripples: outward at –57.1 ps and inward at about –30 ps) and closed by about 15% even before entering the channel.
The insufficient quality of the waveform in this specific case is caused by improper choice of sampling rate and transition times (we address these topics in Sections 3.1 and 4.2). To create a high-fidelity signal with an AWG, one must understand how signals are processed. We have been using deep memory, high bandwidth AWGs in automated testing for years now and would like to share what we have learned.
Our paper concentrates on the diverse abilities of AWGs to generate the signals required for compliance to standards such as HDMI 2.1, MIPI C-PHY, MIPI D-PHY, and MIPI A-PHY. We will share our frustrations and show how we overcame them: cases where AWGs provide the best, fastest, and cheapest solution, as well as cases where AWGs are simply not the right tool for the job.
The paper referenced here was presented at DesignCon 2022. To read the entire DesignCon 2022 paper, download the PDF.