?

Log in

No account? Create an account

Previous Entry | Next Entry

House numbers

Bwahaha! I have mastered the mill's backlash! At least on the final digit 3 I did.

Tags:

Comments

( 14 comments — Leave a comment )
maskedretriever
Jul. 25th, 2009 12:34 am (UTC)
Wwwwow.

That is some epic tooling right there.
ideaphile
Jul. 25th, 2009 03:11 am (UTC)
Okay...
What was the problem?
revar
Jul. 26th, 2009 01:36 am (UTC)
Re: Okay...
Turned out to be a combination of replacing the Y-Axis lead-screw and nut to fix the stalls, and measuring my backlash more correctly.
cjthomas
Jul. 25th, 2009 05:06 am (UTC)
I've been thinking about backlash, as the Cupcake CNC will have more of it, as would the bodged-together screw-driven router I'd been thinking about building.

Tentative conclusion: springs applying constant pressure in one direction on each axis should reduce the problem. Haven't tested it yet. Haven't acquired either fabber yet, for that matter (still @envying yours };>).

How much backlash did you have, and how did you resolve the problem? I was under the impression that your mill was good to 1 mil or less even with backlash.

-Deuce
maskedretriever
Jul. 25th, 2009 07:42 am (UTC)
For the record, the Cupcake CNC has more trouble with non-backlash problems than with backlash. Well-tightened belts don't twist far before changing direction at all, and the z-stage has a solid 1g of acceleration on it at all times.

The problems with mine so far have more to do with software not having buffering, acceleration, etc, to keep the flow working well. Software upgrades to things like the so-called "5D GCodes" used in higher-end RepRap designs might be more important...
revar
Jul. 26th, 2009 01:41 am (UTC)
...

Err, I'm about 3/4th of the way through writing my own accelerating firmware, with lookahead.

Dangit. :)
revar
Jul. 26th, 2009 01:40 am (UTC)
I am currently correcting for 0.002 inches backlash on the Y-axis, and 0.005 on the X-axis. The milling software compensates. I can probably reduce my backlash more in the X-axis by tweaking the lead-screw nut adjustments.

For really small and tight backlash requirements, going to a ball-screw for the leadscrew and nut would be the way to go.
revar
Jul. 26th, 2009 01:58 am (UTC)
With the backlash compensation adjusted and turned on, the mill has an effective backlash of less than 0.00025.
maskedretriever
Jul. 26th, 2009 05:23 am (UTC)
...um, wow, yeah, the MakerBot probably has more backlash than that.

Of course, my prints don't look anything like Nophead's yet, either, so presumably I'm ~way~ behind all that. Might get buffering working tonight if I'm lucky though.

Which would do wonders for my awful build quality.
revar
Jul. 26th, 2009 07:45 pm (UTC)
The Makerbot honestly doesn't need that small a backlash, really, given that the basic voxel size is 0.5mm.
maskedretriever
Jul. 26th, 2009 07:51 pm (UTC)
Hm, there's a higher precision available to the CNC in terms of number of steps per mm, basically meaning the plastic can be positioned more accurately than its thickness by quite a lot.

For example, ignoring backlash and ooze, it's possible to lay two lines of plastic with a 10um gap between them...
revar
Jul. 26th, 2009 08:25 pm (UTC)
Hmm. The machinist across the parking lot from my shop showed me a part he'd milled that had holes drilled with a 0.009 inch bit. That'd be like 0.2286 mm. That'd make for a wicked hi-res extruder. Is Skeineforge adjustable for extrusion size?
maskedretriever
Jul. 26th, 2009 09:59 pm (UTC)
Yes, definitely! 0.5mm is a compromise figure to make it:

* easy enough to machine with a drill press and a steady hand
* possible to print fairly large things in a reasonable amount of time (maximum extrude speed is lower for finer strand)
* relatively hard to clog

I have my number wrong there-- it's 85um steps, but still way lower than 228.6um strand. A "wicked-high-res" extruder build would probably have a few new challenges, but also be of definite value to the MakerBot community :3
revar
Jul. 26th, 2009 08:38 pm (UTC)
My mill uses simple thread-screws. A high precision motion system will use ball-screws and ball-screw nuts. Backlash is improved (but never removed) by adjusting the lead-screw/ball-screw nuts.

Springs are a bad idea because of inconstant pressure that may make your steppers stall on one extreme of the axis.

If you must have something to provide side pressure to eliminate backlash, consider using a couple 1 Kg weights hanging via a couple pulleys to pull each axis to the side. It's a bit more constant a force, so long as you don't swing the weights around.

Edited at 2009-07-26 08:39 pm (UTC)
( 14 comments — Leave a comment )