Arbitrary Waveform Generators for Transmon Qubit Control

This article provides a direct technical explanation of how arbitrary waveform microwave controllers generate shaped nanosecond pulses to manipulate superconducting transmon qubits. It covers the complete signal generation chain within quantum control electronics, starting from digital waveform synthesis in FPGAs, passing through high-speed digital-to-analog conversion, and utilizing IQ mixing for microwave upconversion. Additionally, it explains the necessity of pulse shaping techniques like DRAG to prevent quantum leakage and achieve high-fidelity quantum logic gates.

Digital Pulse Synthesis via FPGAs

The generation of a shaped nanosecond pulse begins in the digital domain using Field-Programmable Gate Arrays (FPGAs). When a quantum algorithm requires a specific gate operation—such as a Pauli-X or Hadamard gate—the control software translates this instruction into a digital envelope definition.

The FPGA acts as a sequencing engine. It stores predefined waveform envelopes in its internal memory—typically Gaussian, cosine, or derivative shapes—and dynamically scales their amplitude, phase, and duration in real time. Because transmon qubit manipulation requires deterministic timing, the FPGA operates on a strict clock cycle, ensuring that pulse sequences are triggered with picosecond-level precision and ultra-low latency.

High-Speed Digital-to-Analog Conversion

Once the FPGA calculates the digital representation of the pulse envelope, it streams the data to high-speed Digital-to-Analog Converters (DACs) within the Arbitrary Waveform Generator (AWG).

To shape pulses that last only 10 to 50 nanoseconds, the DACs must operate at sampling rates ranging from 1 to 10 Gigasamples per second (GSa/s) with a vertical resolution of 12 to 16 bits. The DAC translates the binary numbers from the FPGA into continuous, smooth analog voltage signals. These signals represent the intermediate frequency (IF) baseband envelopes, typically oscillating at a few hundred megahertz.

Microwave Upconversion and IQ Mixing

Superconducting transmon qubits have transition frequencies that typically reside between 4 GHz and 8 GHz. Standard high-resolution AWGs often generate baseband signals at lower frequencies and use an analog upconversion stage to reach the microwave regime.

This upconversion is achieved using an In-phase and Quadrature (IQ) mixer:

  1. The Local Oscillator (LO): A continuous-wave microwave source generates a stable carrier frequency near the qubit's resonant frequency.
  2. I and Q Channels: Two separate DAC channels output the In-phase (I) and Quadrature (Q) components of the shaped nanosecond pulse.
  3. Mixing: The IQ mixer combines the I and Q analog envelopes with the LO carrier wave. By precisely varying the relative amplitudes and phases of the I and Q signals, the controller can manipulate the final microwave pulse's amplitude, frequency, and phase. This precise phase control allows the electronics to rotate the qubit state around any axis on the Bloch sphere.

Modern Direct Digital Synthesis (DDS) architectures are increasingly bypassing analog IQ mixers by utilizing ultra-high-speed RF-DACs capable of directly synthesizing the shaped 4–8 GHz pulses, which eliminates mixer calibration issues like LO leakage.

Pulse Shaping and the DRAG Technique

If a controller sends a simple, rectangular microwave pulse to a transmon qubit, the sharp turn-on and turn-off edges create a broad frequency spectrum. Transmon qubits are weakly anharmonic oscillators; the energy gap between the ground state (\(|0\rangle\)) and first excited state (\(|1\rangle\)) is very close to the gap between the first (\(|1\rangle\)) and second (\(|2\rangle\)) excited states. A broad frequency spectrum will accidentally excite the qubit into the non-computational \(|2\rangle\) state, a critical error known as state leakage.

To prevent this, arbitrary waveform controllers generate shaped envelopes. A Gaussian envelope smooths the rising and falling edges of the pulse, narrowing its frequency bandwidth so it only addresses the \(|0\rangle \leftrightarrow |1\rangle\) transition.

For extremely short pulses (e.g., 10 to 20 nanoseconds), even a Gaussian shape is too broad in the frequency domain. Controllers solve this using Derivative Removal by Adiabatic Gate (DRAG) pulsing. Using the IQ channels, the AWG applies the standard Gaussian pulse to the I-channel while simultaneously applying the mathematical first derivative of that Gaussian pulse to the Q-channel (90 degrees out of phase). This creates destructive interference at the \(|1\rangle \leftrightarrow |2\rangle\) transition frequency, effectively canceling out state leakage and phase errors.

Cryogenic Transmission and Attenuation

Once the shaped, upconverted nanosecond microwave pulse leaves the room-temperature arbitrary waveform controller, it travels via coaxial cables into a dilution refrigerator housing the quantum processor at approximately 10 to 20 millikelvin.

As the pulse travels down the cryostat, it passes through a series of inline attenuators and infrared filters at various temperature stages. This step reduces the thermal noise coming from the room-temperature electronics down to the single-photon level. The pristine, shaped microwave pulse finally reaches the superconducting circuit, coupling capacitively to the transmon qubit and driving the exact Rabi oscillation required for the quantum gate operation.