Firmware locks up 20 or so layers into a print

For some reason, my firmware locks up just after recieving a G1 move command somewhere in the middle of printing my part. It generally gets about 20 layers in then stops moving. The heaters will either get stuck in an on or off state. I do not see any more activity in the host software console either.

So far, here are the details I have been able to collect:
1) Printing the same file twice results in locking up at the same point in both builds
2) Changing the infill settings and re-slicing the same model moves the lockup point to a new location.
3) There are absolutely 0 error messages prior to the lockup.
4) Slicing the same stl with settings appropriate to my maker bot yields a file that will print fine on the maker bot, so I assume the problem is not with the STL.

Has anyone else seen this happen? I am at a loss for a cause, but I suspect that whatever it is has something to do with why my start up script needed redundant moves inserted before it would run all the way though. See this for a little more background on that issue.

Having a similar problem using ReplicatorG and the Makerbot firmware with Gen3 electronics. Prints stop in the middle after working perfectly to that point. Anyone?

So what does the gcode look like around the point it locks up?

With the help of some other members, I pinned it down to a speed/step resolution problem - at least, this was the issue in my case.

The short and sweet version...
If you use higher than normal micro step settings or you are gearing down the axis in a way that has the same effect, you have to increase your feed rate to maintain slightly-faster-than-a-slug speeds. You also increase the number of steps that are required to move the same distance. One poster mentioned that high step rates may crash the firmware, but I'm thinking there is more to it than speed alone. I could travel relatively fast over 10mm (in fact, the firmware is capable of stepping faster than the motor is able to keep up - which causes it to vibrate between two magnetic poles rather than rotate), but if I traveled at a slower speed over 100mm, it would lock up, so I think it has to do with speed and/or the number of steps you are about to take.

After changing my Z axis to use half stepping, I was able to run at much faster speeds. I could probably run even faster if I backed off microstepping on the X and Y too, but I like how much quieter it is this way.

I started this thread when I first encountered the problem and really didn't know where it was coming from. Then I started to think it was something specific to the firmware, so I started another one here. If you want to see more of the picture, you can read the full discussion there.