Possible glitch caused by ramping filter

Sheila Dwyer and Camilla Compton reported a ripple in an H1 DARM output when one of the filters on the input was turned off in a ramping mode.

The ramp was set to 10 seconds. The ripple starts at or very near 10 seconds after the filter is switched off, so near the completion of the 10 second ramp.

This is from the h1omc model.

Output channel is H1:LSC-DARM_OUT_DQ. The filter that was turned off was the first filter "FM1" of H1:LSC-DARM1. DARM1 the output of DARM1 is added to other values before it reaches H1:LSC-DARM_OUT (a plain test point), so the output is not a pure witness of DARM1.

When the filter module is turned off, SWSTAT changes from 40521 to 40520. The change in the first bit is correct. No other changes were made to DARM1 from that point prior to the glitch.

The switch happens at GPS 1417200583.063 according to the SWSTAT record, which is only saved at 16 Hz.

The ripple starts at GPS 1417200593.11, has some KHz frequency components and lasts for several milliseconds.

The glitch zoomed in.

The span from the filter module switch to the glitch.