Are Your Parametric Two-Qubit Gates Limited by the Driving AWG?

September 13, 2019 by Jan Benhelm

If you are using the Zurich Instruments HDAWG, the answer is no. Let's see why.

Parametrically modulated superconducting qubit gates (parametric gates) are usually accomplished by RF-modulation – typically several hundred MHz – of the flux through a mediating coupler element or qubit. Usually, many dBm of signal power are required to achieve sufficiently fast gate operation; hence, an AWG with a large output range and bandwidth is desirable.

As the AWG driving the gate is directly coupled to the qubits, the question arises to what extent the output noise of the AWG limits the fidelity of the gate operation. This was studied and discussed in a recent paper [1] where the authors concluded that for a noise power spectral density (PSD) or "noise floor" below -135 dBm hardly any contribution to gate infidelity is expected (see Figure 6 in the paper). They show that for sufficiently large qubit coherence times T1 and T2, gate infidelities as small as 0.085% at a PSD of -147 dBm/Hz are to be expected.  As shown in Figure 1, this is exactly what the Zurich Instruments HDAWG offers at an output range of ±2.5 V. Both the Rigetti home-build AWG and the Tektronix 5200 AWG mentioned in [1] exhibit higher PSD at lower maximum output ranges.

Noise PSD of Zurich Instruments HDAWG

Figure 1: The noise power spectral density (PSD) of the HDAWG with the output range set to ±5 V and measured with the UHFLI. The noise floor of the UHFLI at an input range of 10 mV and 50 Ω input impedance is significantly below the HDAWG measurement data and hence not causing systematic measurement errors.

The HDAWG is thus an excellent choice when it comes to implementing fast parametric two-qubit gates with high fidelities on transmon qubits.

References

1. Schuyler Fried, E. et al, https://arxiv.org/pdf/1908.11370.pdf.