Got the pictures here.
@LumberSmith I do not want to spam too much but I have to make two points:
1) The "engineer's square" you used to check is NOT an "engineers square" and 100% the gap you see is because the "engineer's" square is not square while the GRS is /within tollerance/.
How I know - No precise (engineer's) square has distance marks on itself like the square you used has - making these marks would create tension in the metal and affect the accuracy ..
For your reference, here is how a precise engineer's square looks. This is a DIN class 1 /which is about the precision the GRS is made to/:
https://shop.kinexmeasuring.com/en/precision-flat-square-kinex-300x200-mm-din-875-1-p8471c22c34/Notice the square has no distance marks and the material thickness is 8mm for a 200x300 pieces. That is required as a thinner one could not maintain its shape sufficiently for the precision.
/I am not familiar with US metrology classes, so those from US please feel free to give an equivalent US class./
2) the back of the rail may not be /absolutely/ parallel with the rail ALONG ITS ENTIRE LENGTH, remember, the reference surface is the one where the saw rides /not the back edge/ AND the rail is guaranteeed to be straight
along its entire length within a certain tollerance. NOT at its beginning. To check if this is not your problem, try using the square on the other side of the rail.
I have this issue with one of my rails as it starts a bit "off" - by about 0.05 mm (0.002") on the 150mm square interface distance which puts the cuts off by about 0.5 mm (0.02") over the 1.4 meters of the rail which can be too much at times when combined with other errors.
I found out the hard way and that is why I now recommend the Festool FS-WA above the GRS as someones first or only rail square. It has to be calibrated - so you need a proper engineer's square to go along, but it CAN be calibrated. Unlike a fixed square. So even non-absolutely-straght rail starts or bad technique causing systemic errors can be weeded out /at the cost of fiddling/.
Hope helps.
The first thing I would do - in your shoes - is to get a proper engineer's square so you have some known reference. You will want to have it either way. No matter the outcome with the GRS. You need to shoot for DIN class 1 or better (equivalent) for it to be useful here. - for a 200x300 sqaure that makes 0.025mm max error over the 300 mm /notice that is only 1/4 of the error I have which already causes me problems/. Yeah, the reference square really needs to be that precise, ideally even better. The GRS 16 generally is at DIN class 1 /from my checks of the Festool version, that is/.
Second, I would try making cuts when the square is attached, e.g. to the middle of the rail. And see where it gets you. So the GRS is attached at a different position on the rail than on the usual "start" one. Using a different section for the reference.
