Using approaches such as superconductor circuits, quantum dots, nanowire, graphene, diamond, and many others, researchers are able to implement limited capacity quantum computers. To perform experiments on these systems, researchers require the ability to define and repeatedly send very low-noise, low-jitter signals into the quantum computer and then evaluate the results.
This need has led researchers to one of the staples of electrical and RF design: the AWG (arbitrary waveform generator). The AWG is ideal because it can pre-compensate for analog effects in a measurement system such as low-frequency droop when using a bias T -- cable loss. There is an excellent example of the pre-compensation approach in this paper from Massachusetts Institute of Technology.
Researchers have been using the 14-bit, 1.2 GSample/s Tektronix AWG5000 series. Comperable AWGs will also do the job. For quantum research applications, a 1.2 GSample/s rate is typically sufficient. Because qubits are so unstable, more important than sample rate is the ability to generate lots of patterns, low noise, low latency, and multiple channels with multiple markers per channel.
In talking to quantum computing researchers, we've found they're betting on multiple horses when it comes to building a stable quantum computer, and subsequently have a range of different use models for AWGs. One group uses AWGs to produce time-domain square-like pulses to nudge electrons into different energy levels. Another group processes I/Q baseband signals that go to an I/Q modulator, the output of which is fed to a resonant cavity. One group produces bursted RF signals. Yet another group creates pulses of general mathematical shapes such as Gaussian, Gaussian derivative, or haversine that are spaced some distance apart and then adjusted during experiments.
Another approach is to send narrowband microwave signals into quantum superconductors. For this application, the AWG is used for I/Q modulation, with carrier frequencies in the 5 GHz to 12 GHz range, and a modulation bandwidth of about 200 MHz. The typical experiment involves outputting 10 sample pulses at 1 GSsample/s. The shapes are typically Gaussian, or Gaussian derivative, with hundreds of microseconds between pulses.
As the simplified block diagram of an AWG below shows, the AWG creates signals by reading digital waveform samples from memory locations and feeding them through a DAC (digital-to-analog converter). Because of this design, AWGs can create any waveform imaginable for a wide range of applications.
An arbitrary waveform generator takes digitized waveforms through a DAC to recreate them in analog form.
In addition to traditional electronics and microwave/RF applications, AWGs are also being used in many research applications to help define unexplained phenomena. The typical approach is to use the AWG to send a precise signal or data into an environment then take measurements on the response. Researchers often use equation editors or MATLAB to create waveforms to load, edit, and compile equation files.