Wednesday, 26 June 2013

BuildBrighton HPC laser cutter

For what seems like months now, the laser cutter at BuildBrighton has been out of action. First the tube went, then the new tube cracked. Then a replacement tube went in but it wouldn't cut. Now it does cut, but only in certain parts of the cutting bed.

So a few months ago, a few of us clubbed together to pay for a second hand LS3020 laser cutter that was going cheap. It's always a risk, buying stuff like this second hand. But we were already familiar with the LS3020 (we've got one at Nerd Towers which works like a dream) so were confident it should be a half-decent machine.

Today, I finally located a Windows XP machine down at the hackspace and got NewlyDraw installed on it (NarlyDraw does install on Windows7 and I've had my laser cutter working with my Win7 notebook, but it's a pain getting the drivers working properly, so Windows XP is much more favourable). I don't know where the PCs at BuildBrighton go to - one day there's a glut of XP machines, the next, some bugger has put Linux on, or replaced them with Raspberry Pi's or something! But with a WinXP machine finally located, and NewlyDraw installed on it, there was no need to delay trying the machine out.

Amazingly, it cut, first time!
Not perfectly, but it did scribe a circle onto some acrylic (a fairly wide, melted shape instead of a circle, but it was definitely a good start).

At some point, someone has changed the optics in this machine, and put a different lens in it. The lens focal guide is much smaller than the one that usually comes with the machine (or at least, the one that comes with later LS3020 machines - this one was definitely an earlier model than the one we're familiar with, albeit very similar). This means that the cutting bed is much higher up, and the cutting head is very close to the material being cut. It also makes it very difficult to see the little laser dot, for lining  up cutting on a sheet which may already have shapes cut out of it.

One thing to note is that when the machine powers up, the cutting head goes to the top-right of the machine, not the top-left as one might expect. This is easily resolved using NarlyDraw - just move the head to -80,0 and when prompted, reply that you know this might make the head move out of the cutting area. The head should now move to the top left corner of the cutting bed:

We cut a few test circles, 20mm in size, in opposite corners of the cutting bed, to see if the alignment was good (if the alignment was out, we'd expect one corner to cut better than the other). The cutting was the same in both corners. We tried a few different settings - the power is controlled by a knob on the actual machine, the speed is controlled in software:

At 15mA and 12mm/sec, the laser almost cut through the entire sheet in one single pass:

The pieces could easily be pushed out, but a slight ragged edge remained in two points on the circumference of the circle. We needed either a slower speed or more power to get a perfect finish. In the end we settled on 20mA and 10mm/sec speed.

(pressing both laser buttons together fires a test beam which allows you to adjust the power setting and see the output on the dial above)

At 20mA and 10mm we got a perfectly circle (no ragged edges) which cut all the way through the 3mm acrylic in a single pass, and the circle fell away from the sheet when lifted out of the machine.
We repeated the cutting process in the opposite corner of the cutting bed, to check that the cutting action was the same (and thus confirm that the bed was straight and the mirror(s) correctly aligned). Both times, the circle cut exactly as we expected:

Placing one circle back into the hole it came from and rotating showed that the shape wasn't actually 100% circular. It started to bind in opposite corners of the hole (if the shape was totally circular, it show rotate freely with no binding at all)

The difference is in the regions of tens, if not hundreds of a millimetre - but it is possible to see. This suggests that either the mirrors are not exactly lined up, or one (or both) axis have a tiny little bit of backlash in them. It will take a bit of time of entering numbers, cutting and re-cutting shapes and measuring the results to reduce this entirely.

For now, we're happy that we've got a working laser cutter again down at the local hackspace. Someone else can take over now!