in theory you are right that it’s easier to update software than hardware, in
practice, we could not figure out how to make things more accurate in software
for those original brackets.
it turned out to be insanely sensitive to even small errors in the measurements,
and the calculations do not account for the difference between the length of the
chain and the effective length of the chain taking into account that it will sag
a bit (especially under low tension)
all I can say is try to be extremely careful in your measurements. the chain
attachment points need to be the point where the chains pivot, not the end of
the chain, and since that eats up chain length, you need to move the chain an
extra link or two so that the distance between the top center of the sprocket
and the pivot point on the sled is what the machine thinks it is (anything
beyond that pivot point does not count)
also, make sure that you don’t put the retaining pin through the chain between
the rollers on the very last link (where there is no pin through the chain
hole), we have seen chains come apart letting the sled swing freely in such
cases. either put the pin through that last hole, or put it through the middle
of the next link.
the shape of your sled doesn’t matter, round, square, octogon, doesn’t matter.
If you want, you can cut your own links to make a triangulation linkage kit, it
takes a little sideways thinking (you don’t just lay out all the parts and let
the machine cut them, you only cut them in the exact center of the machine and
you move the material sideways to cut the different parts), but even with an
inaccurate machine, you can cut the triangulation parts to make it accurate.