If you look into the spectrograph with the imaging camera removed, you can see the diagonal mirror which sends the light towards the grating through the lens. It is important that the return beam misses this mirror when it returns to the camera. This is achieved by rotating the mirror. The mirror should be adjusted so the spectrum is approximately 1/3 the way up from the bottom of the camera field when the spectrum is horizontal and orientated with red on the right.
The fact that the focus of the spectrum (The width of the spectrum line) is varying so much as you adjust the mirror. (80 pixels means the spectrum is way our of focus). I do not recall seeing this before This suggests to me something may be out of alignment. We may have to revisit this.