The figure below shows one possible cause of fringes. In each detector assembly there is a bowed black polyurethane filter B and a bandpass mesh filter F (which is wedged and anti-reflection-coated), and an aperture A. An off-axis ray R may pass through the filters but bounce in the lower cavity before entering the detector cone leading to the detector D. While bouncing in the cavity it may experience interference. The gap of the lower cavity is about 1.54 mm.

Were this happening, it might explain why extended sources in particular exhibit fringing. Also, small difference in the gap between F and A from detector to detector might cause the fringe function to vary slightly in its period, although discontinuities in its waveform across the band. Period differences from each detector's passband are being sought at this time.

        R
       .
|  ___.___B |
| /  .    \ |
|/  .      \|
|__._____F__|
| . .       |
|. . .      |
|_.__ . __A_|
    \  ./
     \ /
      D