Monday, 8 August 2016

CNC Cutting — a learning experience

My friend Andrew has a small CNC router/cutter that he built as a kit. It's by no means a heavy-duty machine, and it's not without its issues, but we've been playing with it over the weekend and are starting to learn how to make the most of it.

Our first attempt was cutting some 50mm hexes for my wargames terrain project in 3mm MDF. The machine runs off an SVG pattern, which I put together in CorelDraw.

It differs from a pattern for laser cutting in that you have to make more allowance for the thickness of the cutting bit. In this case we were using a 3mm bit, so each hex had to be over-size — 3mm over-size, in fact.

On this pattern, it got through the first four rows and then the bit broke. I suspect that that might have been because it was quite blunt, so it was being put under more lateral strain than it should be as it was being tracked through the MDF.

Ragged edges
You can see here how furry the cuts are on the first lot of hexes. That's because of the blunt bit.

This was the first purely MDF cutting project Andrew had tried, but he habitually uses MDF as a sacrificial support, and MDF is hell on steel tools, so the tip of the old bit will have been quite worn. You really want to be using tungsten-carbide to get long life out of any tools cutting MDF.

The remaining hexes, after we changed the bit, were cut much more cleanly, but we ran into another issue:

The bed of the machine doesn't appear to be perfectly flat and square to the gantry, so the hexes in the bottom-right of the pattern weren't cut all the way through.

Some of the parts for a paint stand
The following day, we tried something else: a pattern for a multi-tier rotating stand for my Vallejo paints.

We set the cutting depth slightly deeper to account for the distortion of the bed, and we used Instructables' Easel software to automate the placement of the tabs holding all the pieces in place. Though when I say "automate", I really mean "automatically put in far too many tabs and in all the wrong places". They're manually adjustable, but the amount of work involved ended up not being much less than it would have been doing it manually when I made the pattern to start with.

The cutting, once again, did not go without incident.

After partially completing the large circle (bottom-left of the photo), the machine suffered a brain-fart and started cutting again in completely the wrong place. Fortunately, Andrew had thought ahead and marked the original zero-point, so we could just stop it, delete the parts that had already been cut from the pattern, and start again.

Everything went quite smoothly until we hit another snag: because we'd increased the cutting depth to accommodate the bed inaccuracy, the auto-generated tabs were cut too deep, and therefore too weak. When it tried to cut out one of the edge-tabs (top-centre of the photo), the tabs broke as the bit was cutting an indent, flinging the piece about and breaking another bit.

We started again, deleting all those problematic elements, and things seemed to go fine until the smallest circle, which again broke free. It didn't break anything this time, but it did move around which meant that the tab-slots weren't cut accurately. Hey-ho.

So, what did we learn?

  • It's better to split up a large, complex cutting job into smaller, more manageable chunks. That way, when something goes wrong, only part of the job will need to be re-done.
  • Good, firm attachment tabs are vital to keep cutting accurate, and to protect the bit. The cut-out circles were all left without tabs, but they were all small, light pieces, and without any indents that could catch on the bit and over-torsion it.
  • Very light material like 3mm MDF should probably be secured with double-sided tape rather than edge-clamps, which tend to bow it slightly.
  • I suspect that the issue with bed flatness might be ameliorated if the whole shebang was placed on a firm, unyielding surface of guaranteed flatness. The material of the gantry and frame is insufficiently rigid to provide its own support.

In summary

It has the potential to be an amazingly useful machine, just as long as its limitations are taken into account. It's not as fast or powerful as the big commercial machines, but on the other hand, it didn't cost tens of thousands of dollars either. It would be of limited value in furniture making because of its small size and lack of power, but it could make making small things like boxes very easy, and I could see it being exceptionally useful for doing inlay work.