While the spectrum looks pretty good, the DC level which typically appears DC coupled, the offset from zero is >4000 counts.  It would appear from trends that this has been the case for months but since we don't use these sensors for anything we've decided to ignore this.  And if the spectrum is good, maybe the sensor is still usable

Attached is spectra from ~3am this morning and the IPS weirdness is evident, the issue with the H1 L4C is not as obvious.  Historically looking at past data (back to Dec 2013,) the L4C variance is noted but the IPS looks worse now.