View Full Version : Axis resolution advice

22-09-2010, 03:40 PM
I am wondering what you guys do condsidering resolution when building a laser cutting or similar router type machine. I wish to cut normal modelling parts, wing ribs etc. but certianly don't want to scrimp on the accuracy. I was going to gear the steppers to give me 0.02mm movement per microstep. Is this ball park ?


Robin Hewitt
22-09-2010, 05:15 PM
A microstep being what?

Stepper accuracy under load goes a bit pie in the sky when you go beyond half step.

OTOH stepper smoothness goes way up as you increase resolution by microstepping.

OTOOH what are your screws rated at backlash-wise?

22-09-2010, 05:45 PM
I'm using Gecko drivers, so a microstep is 1/10th of a normal step.

I am going to use toothed belts, so my design choices are belt pitch, reduction before main drive belt, and main drive belt pulley. Steppers are 1.8 degree per step.

So you think I should base the resolution on half steps ? So lets say I want accuracy down to 0.05mm I just make each half step equal this ?

22-09-2010, 10:39 PM
if you are using belt drive, you need to take torque requirements into account as these will dictate the main drive pulley sizing and reduction ratios. if you work to 1/2step resolution than you wont go wrong... you can still use microstepping to smooth the motion...

Robin Hewitt
22-09-2010, 11:30 PM
So lets say I want accuracy down to 0.05mm I just make each half step equal this ?

50 microns probably requires an iron machine rather than aluminium and certainly not MDF :eek:

Why not find someone who has built a similar machine to the one you have planned and ask them what tolerance they can get out of it?

I don't have a laser or a router, I have a mill. FWIW I quarter step to 5 microns and hope for 10 :smile:

23-09-2010, 12:15 AM
True Robin, I think Cam is confusing resolution with accuracy. To be accurate to 0.1mm you need to resolve to something better, 0.05mm being an obvious one, ideally better than that. For a router cutting wood, 0.2mm is probably accuracy enough, needing 0.1 or better resolution. You'll never be as accurate as the lowest resolution you can maintain. So if you want 0.05mm accuracy then you need a 1/2step of 0.025mm or better. However that is just the accuracy of the drive train. Once you allow for backlash, flex of the supporting environment (even without cutting forces the mechanics will flex as they move), etc. your accuracy will suffer...

For a model wing rib, 0.25mm accuracy should be good enough, so 1/2 step resolution to 0.05mm plus some flex allowance should get you close...

23-09-2010, 12:49 AM
I think Cam is confusing resolution with accuracy. To be accurate to 0.1mm you need to resolve to something better

This is exactly what I was driving at, I just missed the terminology by a mile. To re-phrase I wanted to know what sort of ratio you give between resolution and desired accuracy. From what you are saying double is OK a little more is better (5x), but a lot more (x10?) will be lost in the machine.

When I posted I was thinking an order or magnitude, i.e. 10 fold between the two but back then I was also considering microstepping. That said although my understanding has improved very considerably my gear ratio is going to be very similar, rather good luck ! This because I have reduced to considering half stepping as the smallest finite usable resolution, but also lowered my resolution to accuracy ratio.

I have looked in to other peoples machines which is probably why I ended up at more or less the right ratio even if it was for the wrong reasons. I just wanted to check they had done there homework correctly, no point copying a bad design :)

For the record the machine will be ally, hopefully quite rigid, no unsupported rail, a router was only and after thought it will primarily be a laser cutter so no cutting loads.

Thanks for all the help, very useful.

Robin Hewitt
23-09-2010, 04:38 PM
One last tip, avoid tiny pulleys when microstepping.
Big fat pulleys are a lot more forgiving with regard to belt tension.

23-09-2010, 06:10 PM
Thanks again guys, hopefully on the right track now.