Backlash in geared extruder

So, backlash is the enemy of precision regarding linear motion, but I've been thinking about whether or not absolutely trying to remove backlash from the extruder is as vital as the backlash in the linear motion axes (regarding precision prints). Obviously to have an accurate print, you need precise control of the extruder and how much flow is coming through, but my thought process is the following: since retraction settings are so important (especially with a Bowden-type extruder), and because of the fact that extrusion settings can vary even from color to color of the same filiment from the same manufacturer (only difference being color), one has to dial in their retraction and extrusion settings for each specific filament for precision prints anyways. In dialing in the retraction settings and calibrating for a specific filament, one is effectively negating the backlash effects of the extruder setup (assuming it's not a direct drive), correct?

I'm not sure you are negating anything. I honestly think that backlash is overshadowed by the fact that the filament isn't perfect diametrically or concentrically throughout a spool. The volume output of the filament is going to be less consistent than the amount of backlash at any given moment making any attempt to fix the small amount of backlash we have pointless. Also, theoretically speaking, direct drive will also have backlash, just a lot less. We are cutting into the filament to drive/force it into the hotend, essentially cutting a gear into the filament itself. Another note, we will never eliminate all backlash in any situation requiring movement (whether it's the threaded rods, the linear motion, or the extruder), because a certain amount of clearance is always required for smooth movement.

Also, we aren't accounting for the backlash, even theoretically (assuming we had perfect filament), via software. We are changing retraction settings and such, but if we were to account for backlash, we would need the software to recognize every change in direction of the stepper and add an additional amount and then we would hope that would remain consistent throughout a print (no filament finding it's way into hobbed gear teeth, etc) in order to truly account for the small amount of backlash.

I think we're doing pretty good as is considering all of the variables.

Just my thoughts, I could be (likely am) way off in my thought process.

Edited 1 time(s). Last edit at 03/02/2013 02:59PM by Bobbitay.

In an extruder I'm pretty sure there is no real advantage of removing backlash. Any backlash is accounted for by your retraction distance.

How is backlash accounted for by your retraction distance? It doesn't make a bigger move when going forward again (extruding) no matter what the retraction value is - it will always have the same clearance (backlash) between the gear teeth. I don't think it matters, though.

Bobbitay Wrote:
-------------------------------------------------------
> How is backlash accounted for by your retraction
> distance? It doesn't make a bigger move when going
> forward again (extruding) no matter what the
> retraction value is - it will always have the same
> clearance (backlash) between the gear teeth. I
> don't think it matters, though.


If you keep going in one direction then backlash has no effect, it only has an effect when you reverse direction, but since the extruder retracts the same amount that it de-retracts the net effect is zero. For example:

Say you are extruding, so the gears are meshing, and then command a 1mm retract. The first say 0.2mm of the retract takes up the backlash, then the remaining 0.8mm actually move the filament. When the retract is undone, 0.2mm is used to take up the backlash again and the filament moves back by 0.8mm. If you actually wanted to move the filament 1mm you would have to set your retract to 1.2mm.

You start and finish in exactly the same place - backlash is irrelevant apart from the slight delay between retraction starting and the filament moving. The retract value you find by experimentation includes required movement of the filament, plus backlash.

Thanks for the explanation. I understand what the belief is now.

To me, I still highly doubt the extruder gears are very accurate gear tooth profiles and that would negate the small amount of backlash retraction tuning we did. I still hold that the filament is too inconsistent to achieve any real effective backlash tuning especially considering that our gears are also, as a whole, not very precise. There are also gaps between the filament and the walls of the cold-end/hot-end which causes the filament to flex in different ways (similar to problems with bowden configurations, just much less)...

Seems like the odds are against us to me, but we still do well. I think that says that most of the factors that we theoretically consider are pointless for the level of precision we are seeking. I don't think it's a very pragmatic thought process to try and consider all of these factors. Retraction works, but if you look at it under a microscope I'm sure it looks "horrible" in almost everyone's cases. Our method of tuning e-steps is very imperfect. I have very high-end measuring tools and there is no way that marking our filament with a sharpie and measuring the distance it moves off of a non flat surface with a non-rigid filament that doesn't stay perpendicular to our non-flat surface makes for a very accurate decision. Then everything else is based off of appearance for "fine-tuning". I think it's an engineers mindset to be concerned with the non-pragmatic portions of our printing. Maybe we should just admit that it works and we probably don't know WHY it works as well as it does.