Spectra in figure 3 and 4 were measured using a yokogawa optical spectrum analyser at the cavity output. Key settings include: resolution of 0.02nm, sample of 25001 data points and a sensitivity setting of high 1.

Autocorrelation traces in figure 5 were recorded on an APE PulseCheck autocorrelator using both a 15ps time scale and a 150ps time scale. Key settings were an averaging of 16 points, smoothing was on. 

Threshold points given in "Modelocking range data.xlsx" for figure 6 were recorded by increasing pump power from 0 to 140mW and recording the pump power where pulse behaviour changed. This was repeated several times over multiple days and the variance in thresholds was incorporated into the error bars.

Threshold points given in "Peak power data.xlsx" for figure 7 were also recorded increasing pump power from 0 to 140mW but noting the average output pulse power. Using the measured pulse width and repetition rate the peak power was calculated. This was then used to estimate the peak power and nonlinear phase at each cavity stage using known component parameters. This is then summed up to give the total nonlinear phase shift per cavity round trip.