Well… to be honest, I’m not sure why I’ve decided to build this thing. The original idea came about because I build quite a few electronics projects and even with today’s surface mount technology, there’s still plenty of holes to drill in a PCB to make a completed prototype board. I hate prototyping on ‘veroboard’ or with wirewrap or worse still ‘christmas tree’ type of assemblies. I’ve learned the hard way, that half the time the prototypes don’t work, not because the design is flawed, but because the hurried prototype construction has introduced a litany of faults (often intermittent – the most insidious to debug!). So I build PCBs by the kitchen sink (first reference to patient wife) and i get really bored drilling them. So, I thought there has to be a better way…
"I need a better way to drill those holes"
It turns out that most free PCB drafting software (I use Eagle – I cannot say enough good things about it) produces drill files that can be used to automate the drill process. They tell the drill head where to drill the holes. Perfect. So I figured that all I’d need would be a small robotically controlled drill that could be moved in three planes – hence the CNC project.
Now, I’m impatient, and I’ve not done any metalwork since school so i was skeptical that this could work. I was torn between wanting to do a good job and building the thing quickly to actually see if it would ‘fly’. So, 4 years ago, i ended up with a revision 1 CNC machine. It was sort of ‘designed on the fly’, was a mixture of wood and Aluminium sections and ran off some very old underpowered motors by an even more underpowered power supply. At least the Stepper Motor design was built on a PCB (which was possibly the only professional looking part of the whole thing).
I think that I probably spent more time setting it up then using it. The surprising thing about it, for all its faults (and there were many) was that it essentially worked! With some fiddling around i got pretty good accuracy and pretty good repeatability. The problem was that the accuracy and repeatability was not anywhere near good enough for reliable PCB drilling (200 microns) and the thing wasn’t really robust enough to do anything else useful (although i did engrave some brass and surprised myself with the results – in a good way)
Now, I’m glad that I didn’t set out to design the worlds best CNC machine from the get-go because i realize that you learn a lot by trying to design you first and that helps you when you come to building your second. Once you’re done with your second, your accuracy should be so good that you can use it to build your super accurate third machine. I like the idea of a machine that can replicate itself…
Of course, i could just go out and buy the darn thing – but a reasonable system with electronics and all could easily set you back a minimum of 2000 euro – and to be honest, where’s the fun in just buying one – you won’t learn anything. So here we are half way through my 2nd CNC build which I have spent much time on… (Probably more than 2000 euros worth of time, applying standard babysitting rates!)
Before I set off with design number 2 I spent a lot of time looking at the earlier design and trying to figure out how it could be improved. This lead to a long list of potential improvements and some ‘guiding principles’ which I list here:
My Guiding Principles for CNC Design
Let me just state that I am by no means expert in CNC design. At an average social gathering of non-mechanical or Industrial engineers, I’d probably come across as a tenth dan black belt with my invigorating discussions on Anti Backlash Systems, Bearing Blocks and Moments of Rotation. But, in the rarefied world of home CNC builders I really am a nobody. Nonetheless, here are my guidelines:
- Design for accuracy – Wherever possible, design the key sections of the machine to be as accurate as the capability allows: Every element of an axis drive system introduces some inaccuracy; play in the motor couplings, screw-thread pitch accuracy, drive nut backlash, bearing slop and axis co-alignment are additive errors. Some folks say that it’s pointless to try to correct backlash unless you’re using the best quality threaded rod. I say let’s remove as much inaccuracy from the drive system as we can and addresses other elements at a later stage when more time/finance allows.
- Watch the pennies – Keep a budget in mind and stick to it: It is very easy to get carried away by buying more and more specialist tools and equipment to finish the job. Before you know it, with what you’ve spent you could have bought an off the shelf system! Usually you can circumvent the need for ‘that special tool’ with a little ingenuity.
- Perfection takes time – If the end result of your task is not quite good enough then do it again: If two sections are not quite square, if a hole is drilled in the wrong place, if you’ve cut a piece of stock undersize/oversize then redo it. There may be a tendency to convince yourself that it’s ok, that ‘it will do’, but often that inaccuracy will come back to haunt you when the system is difficult or impossible to calibrate accurately enough or when that troublesome part keeps failing and requiring constant repair.
- Plan – Design everything on paper (or PC screen) first, before you pickup any tools. This will save you mountains of time and give you a better end result. With so many free quality design packages (www.doublecad.com, http://sketchup.google.com/) you’ve got no excuse.
- Research, research, research – Chances are that if you’re facing some design challenge you’re not the first and only person to do so. Don’t be the 2nd person to invent the wheel or the 200th person to discover that drill bits lose their edge really quickly without lubricant. Stand on the shoulders of giants and learn from their experiences. The web has lots of invaluable user forums with some real experts available just an email away.
That’s it. I tried to restrict it to five rules, because, let’s face it, nobody is going to remember more than 5 rules!
Building Stuff 