Extended Data Fig. 6: Pulse sequences and additional data for coherence and switchability measurements.

a, Altered Ramsey pulse sequence used for Fig. 4c in which during the wait time Twait a bias ΔVJ is applied to J. b, Coherent Ramsey oscillations for the sequence in a, and processed to produce Fig. 4c. A virtual phase is added to the second X/2 pulse to aid data processing and fitting. c, Altered Hahn echo pulse sequence with offset wait times (as in a) used for Fig. 4c. d, Measured data for the pulse sequence in c, and processed to produce the data in Fig. 4c. e, Coherent switching pulse sequence. The qubit is switched Ncycles times between two bias points Δ1 and Δ2, being close to and far from the degeneracy point respectively. A virtual Z-phase is added before the final X/2 pulse. In addition to this, a variable time Twait is added in order to keep the total time between the two X/2 pulses the same for all values of Ncycles. The wait time is only added in the off-state, Δ2. Each switching cycle uses 20 ns on/off ramps, with a 20 ns pulse settling time at each level. f, Measured data using the pulse sequence in e, with Twait = 30 μs. This data is used to calculate the return amplitudes for sequence e in h, also shown in Fig. 4d. g, Coherence switching pulse sequence in which the amount of time spent in each of the states Δ1 and Δ2 is kept constant in order to cancel out the effect of fast decoherence near the degeneracy from the switching fidelity calculation. h, Measurement of the return amplitudes for varying Ncycles for sequences e with Twait = 30 μs (also shown in Fig. 4d), and g with Twait = 4, 8 μs. We see that for sequence g the baseline return amplitude is reduced, due to the short coherence time at the Δ1 point, however there is no observable coherence decay associated with increasing Ncycles.