Very interesting that the error is constant and not cumulative and seems to occur beyond some fixed point on the Y axis. At least this is what I assume from the previous tsunami of posts.
So either the printer is going short on Y or the plotter is going long on Y at this point on Y.
If the printer were going short then not only would the object spacing be short but the the registration marks on that side would be short as well. If that were the case then the cut would be cumulatively short on Y starting at the origin. This is not the case.
Ergo your plotter or it's software is going short after a certain Y value. If your setup was working properly, as you reported, then it is reasonable to rule out software. Software neither breaks nor wears out. Unless, of course, there's some particular value calculated or encountered that starts a harmonic, for want of a better word. A really obscure notion that, in over 40 years of dealing with software, I have seen exactly once. That was an ill thought out piece of micro-code in an IBM 360-20 back in 1970. During the era of 'face down, 9 edge first'.
So most likely it's something going on in the plotter. It's the nature of electro-mechanical devices that 99% of problems are mechanical. In this case, I have no idea if Graphtec infers position or actually senses position of the Y axis. Inferring by keeping a running accumulation of Y motion or sensing via some mechanism like an encoder strip or something.
If Y position is inferred, look for something really silly, like a tooth missing or clogged on a drive belt, a gear loose on a shaft, etc. Something in the physical drive that would cause it to take a giant step at a given location.
If Y position is sensed, look to that mechanism, whatever it might be.
:
