I measured the full width at half maximum (FWHM) of a large number of stars (several thousand) on each 40 minute subframe of each of my integrations. The charts below show the results as smoothed FWHM histograms, one for each integration, with combined FWHM measurements from all of its subframes. Each chart shows just four histograms, to make them more legible. Each integration is labeled by the month of its acquisition.
FWHM values are shown in units of pixels. My setup at the image scale of 4.2 arcsec per pixel is undersampled, so FWHMs can't be much smaller than about one pixel. But how close to one pixel can they be in good seeing?
For the months Jun 2012 through Jul 2013 sensor tilt and focus drift due to temperature change often resulted in FWHM measurements larger than one pixel. The histograms have long upper tails as a result.
In Jun 2013 I began experimenting with focus drift compensation. But the resulting FWHM measurements were large due to both sensor tilt and below average seeing conditions, the target was at low declination which contributed also. Although sensor tilt in my setup had been a problem, modifications to my setup's image train this month made the problem even worse.
By Aug 2013 focus drift compensation was working well, sensor tilt was partially addressed, but some tilt remained. Since Aug 2013 the histograms have shifted leftward and become more peaky. This of course means that FWHMs are getting smaller and more equal across the frame. This improvement is due to better focus and better collimation with reduced sensor tilt.
In Nov 2013 I nearly eliminated sensor tilt and focus drift compensation continued to work well. The upper tail of the histogram for this month is significantly shorter compared to prior months, and a significantly larger portion of the FWHM measurements lie at or below one pixel in size.