Burst.png shows the fluctuations in the fast monitor and NPRO crystal temperature.
The bursts occur during periods described as "mis-behaving" in subsequent plots.

lfpn.png shows the power noise from the NPRO when the laser was "well behaved" and
with the noise eater on and off.  The noise was measured out to 1 kHz.  Notice that
with the noise eater on, there's a broad peak from ~100 Hz to ~350 Hz that is not
there with the noise eater off.  NE.png is the same plot out to 1 MHz.

RRO.png shows the power noise with the noise eater on and off, so clearly the noise
eater is doing its intended purpose of suppressing the relaxation oscillation peak.
Unfortunately it introduces noise at the low frequency end.

    The observed behaviour might be relieved by adjusting the noise eater circuit
parameters, for which we have no documentation.  Or perhaps by adjusting the temperature
of the NPRO pump diodes.