Now the stability of the instrument can be checked with and without the temperature/convection control elements installed. The measurements in this slide were made with the interferometer mirror phase modulating at the same fringe of the centerburst. These plots represent 4000 data points accumulated over 20 minutes. Ideally, these lines would be flat, indicating stability of the detector signal. The 1.7% variation of the signal (top trace) was so serious that a test of the electronics was conducted with the use of a very stable HP 3336B signal generator as a false detector signal. The signal generator frequency could not be exactly matched to the phase modulation frequency, so much of the residual 0.1% variation is due to a beat frequency between the two. The short-term stability of the detector signal with the insulation installed (bottom trace) was improved by a factor of 15. There is still a slow drift in the intensity, but it does not go through a complete cycle during the interferogram acquisition time, and consequently would not appear as noise.