Laser cutter accuracy?

[Ian Dickinson]

Hi,

Has anyone experienced any issues with the laser cutter cutting parts a little bigger than expected? 

I was trying to cut some perspex today, for the first time since my induction, that needed to be exactly 11 1/4 in, but it ended up being 11 3/8 in.

i.e. 1/8 in (or approx 3mm) too big.

I'm pretty sure my design is correct, as a printout from Inkscape is spot on.

Is this normal, and just needs to be compensated for now I know? Or, could there be there a scaling factor somewhere I am missing?

Ian

[Martin Raynsford]

I found that laser cutters never really cut to size straight out of the box, you need to do some fiddly tweaking with the settings to make sure it's exactly right and newly draw is a bit of a pain for that sort of stuff.
Once you get to know a machine though it's much easier just to tweak the input drawings, in this case scaling them down a little. You know how big you wanted it and how big it actually cut so you should be able to figure out what the adjustment factor is.

[Dominic Morrow]

Hello Ian 

Jon at Leeds had a similar issue, I suspect it can be tweaked out. I've copied Jon in, maybe you could email him and ask.

Dominic 

Sent from my iPhone

--
 
 

[Adam Froggatt]

What program have you been using to produce the .dxf files? I've Inkscape gives me slightly rough measurements, whereas Visio (Microsoft being useful, the end is nigh) has been very very accurate.

[Gazz]

Not sure if it applies here, but when i first got my craft cutter (a vinyl
cutter but with almost a kilo of cutting force to cut upto 3mm card, emboss,
scratch engrave etc) i had a hell of a time using inkscape to lay out the
work, then transfering it to the cutting proggy,

Turns out inkscape exports certian things at a different dpi setting than
most other proggies expect, but this was making big differences to the cut
work, like 1/3 larger cuts.

In my case it was something like inkscape exporting at 96dpi, and the
cutting software expecting 75dpi, or the other way round.... and possibly
other numbers, but could be something to check out, as when you mentioned
inkscape it rang a bell in my head.

unfortunately i havent used inkscape for a few years, so can't help anymore.

--------------------------------------------------
From: "Martin Raynsford" <msray...@googlemail.com>
Sent: Sunday, January 27, 2013 7:31 PM
To: <notti...@googlegroups.com>
Subject: [Nottinghack] Re: Laser cutter accuracy?

> --
>
>
>

[Ian Dickinson]

Thanks All.

Martin - Is the accuracy of cutting paper, likely to be the same as cutting perspex? I'm just thinking, whilst I can calculate a scaling factor now I know, it may be better to test getting the size right with paper rather than waste too much more perspex. 

Adam - The DXF was from Inkscape, and the second time I cut the part I checked the overall dimensions in Newly Draw were correct, just in case I had accidentally resized the part on the first attempt. 

[Martin Raynsford]

Hi Ian,

In all my testing I have found the accuracy to remain consistent (as long as someone else isn't tweaking with newly draw for you)
My machine had a slight skew and scaling issue and I quickly learned to live with it. A slight oversize meant that 3mm material would sit snuggly within a 3mm hole even with material loss from the laser kerf.

Martin

[Chunky]

Hi Ian,
What I have found with the laser is that it has an inaccuracy of approx 1% in cuts from the top left had corner.
For example - I cut a slit that was meant to be 250mm long. It turned out to be 252.5mm long. I then cut a slit 5mm long and it was 5.05mm.
So basically with small things you get less apparent inaccuracy (my designs can live with 0.05mm tolerance), but with larger things you start to get problems.
Obviously everything can be scaled to take this into account, but its a bit annoying...
But this is just from my observed operation and measurements,

Matt

[Adam Froggatt]

I have noticed that when I compare my designs to the cut, where missing material from cuts should be (eg; the plot line was dead on 3mm, missing material should have shrunk it a little), that the size is right despite the cut suggesting some material shouldn't be there.

At a guess, the tolerance may be dialled into it to somehow prevent it from making cuts too small due to cut away material.

Maybe.

[Adrian Godwin]

I'm interested to know what causes these inaccuracies.

We're all using relatively cheap cutters from similar sources. Are
these inaccuracies inherent in hte process, or just a result of
corner-cutting by the manufacturers ?

I'd expect (and I'm only working from limited experience of this type
of machine) there to be 3 sources of error :

Slop and backlash, causing offset changes related to direction of movement
Width of cut (kerf)
Scaling, caused by design or manufacturing errors

Slop can be partly corrected by careful maintenance. Design changes
might also improve it.

Kerf can only really be addressed by changes to the design of the item
you're cutting, though software can help you do this more easily.

Scaling should be a consistent error, as it appears to be from this
thread. At the level we're seeing, it's probably due to slight errors
in the diameter of the pulleys driving the toothed belts. It ought to
be easy to correct in software (if the steppers have enough
resolution) but apparently Newlydraw doesn't offer that. If that
settings's not merely hidden, it could be corrected by machining a
custom pulley. There are also efforts to replace the closed-source
control boards and operating software with something that can be more
easily corrected.

Reason for asking is to emulate the work that's been done on cheap
machine tools : there are many people who have bought a small mill and
used it as a kit of parts to build something better, replacing or
remachining poor components where it helps. The result is a better
machine than is usually available at that price, at the cost of plenty
of your own time. Perhaps the same process is possible with cheap
laser cutters.

-adrian

> --
> You received this message because you are subscribed to the Google Groups
> "Nottingham Hackspace - Nottinghack" group.
> To unsubscribe from this group, send email to
> nottinghack...@googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>

[Martin Raynsford]

The problem is largely a set up problem. 
Newly draw only really has one setting to control these scaling errors, buried in the software somewhere you can set the X and the Y dpi resolutions. Obviously this setting doesn't directly correspond to steps per mm or any other easy to calculate value. I spent the best part of a week tweaking this value up and down with no significant effects and I ultimately ended up returning them where they started. The problem could be how newly draw converts that value into step sizes, we don't know what rounding it does. Finally because I noticed that the laser had a minor skew (and my new one does too), newlydraw offers no settings to compensate for that so if I was going to tweak my files before cutting I might as well solve both problems. 

Obviously it depends on the work, 90% of my items don't care about 1% larger and 1% skewed

Martin

To unsubscribe from this group and stop receiving emails from it, send an email to nottinghack...@googlegroups.com.

[Ian Dickinson]

Thanks again to everyone for helping with this.

The advice from Jon at Leeds, is similar...

"It's perfectly normal - you'll find a DPI setting in NewlyDraw in the advanced part of the properties page (where you set the cut speed). The default seems to be 1016, which is a bit large, ours is now on 1001 and is far more accurate (I suspect it's probably supposed to be 1000, but 1001 is where my calculations came out when I was trying to calibrate it, so that's where it's set). "

For now I'll just compensate in my design file, and I agree with Matt (Chunky) it's about 1% working left to right.

[David Clarke]

I don't think that 1016 is necessarily 'too high' - it is a recognisable metric / inch conversion (from my past work with photo-typesetters)

25.4 mm = 1 inch
1016 DPI = 40 DPmm

So, assuming that the leadscrews are a standard metric pitch this factor would be used in mapping a bitmap specified in dots per inch to fractional turns of the stepper motor.