I wrote that with the different channels, because in mine, red is almost always perfect and blue is always worst. I'm now at a point where red and green is perfect and blue is still a bit out. But this fits the spot diagram of the edge 8. It may be different with your edge 9, which has a different spot diagram for each color.
Regarding longer exposures:
Well, (at least in my edge 8) the problem is that there is sometimes a bright center and some blue smear. And this smear is distributed elliptically. With short exposures, I won't notice that. But it can be seen in guided long exposure images.
But one must also say that I have an active optics which runs at 10hz and more. So I do not get mount errors and I can always suppress seiing to 0.54'' (the optical resolution of the scope).
If you do not have an active opticsm it may help to use a high gain value or guide with an off axis guider. In fact, a high gain value may be enough to capture the smear for each color.
At least you should make trials that are similar to a normal imaging situation.
Also, it would be interesting if you now make an image with the pure edge and compare the FHWM.
Does the pure edge also deliver a better image quality at your new distance?
And finally, if you ask me, this problem is simply annoying. One has to make a series of images, remove the cables, screw out out the imaging train, add an 1mm spacer, screw the imaging train on, add the cables, recalibrate phd guiding, and repeat with imaging...
It just takes long time...
Celestron should measure this before delivery, not us by trial and error.