Improving accuracy with 'turnaround steps'

In my early reprap prototype I controlled the stepper coils directly using the parallel port and hence wrote my own stepper control code for it.
One feature I added was something I called 'turnaround steps':

When a stepper motor is reversed on a less than perfect repstrap it takes several steps in the new direction before the table / head actually starts moving. During these steps the slack in the worm gears / threaded rods is being picked up. It does not cause too many problems when printing large shapes but it limits the fine detail on PCB plotting.

In my early prototype I found that by just adding on the extra steps when the direction changed improved the accuracy of my repstrap greatly. Now i'm using the official (PIC) electronics it could only be implemented there (with is has very full memory). I've attached a diagram that may explain it a little better.

Has anyone else had this problem with their repstrap, or is it just because mine is made from lego?

thats called 'software backlash compensation' and its a good idea on sloppy systems.

i doubt we can fit it onto the PIC stuff, but i imagine it would be fairly easy to add to the Arduino code.

Edited 1 time(s). Last edit at 04/28/2008 10:19AM by Zach Hoeken.

You can overcome that with anti-backlash systems built into the machine too, right?

---

Jay

Or, you can use this kind of stepper which already has the anti-backlash capability built in.

Forrest, you really like those linear steppers don't you?
Well I mean, I love them too. I think that this summer I will build a pcb mill using a set of those. Do you have a website where you are looking at them?

Edited 1 time(s). Last edit at 04/28/2008 02:59PM by Jay Swift.

---

Jay

Haydon, Portescap & Nippon make 'em.