[cgallery]

I have a prototype K2 CNC desktop machine. I'm driving it with Mach3.

When I cut a 4" circle, it may come out .02" (or so) over or undersized. My bits are spot-on, I problem is machine rigidity. I can put my hand on the router bit and push/pull it and see the router motor wiggle.

I've traced the rigidity to the design of the machine, it just is what it is.

What I'm hoping for are, tips for designing with this in mind? I'm sure others have machines that are less than perfectly ridged, do you have any tips? I can draw parts over/undersized but I'm pretty sure if I cut .5" circles and 8" circles, the error isn't the same .02". I'm not positive on that, I will need to test that, but...

I'm just wondering if there are other attacks like altering the bit size in CAM or any other useful techniques.

[ger21]

Take shallower cuts, at lower feedrates.

[cgallery]

I'm already running pretty slow.

Any tips on measuring my deviation? I think I need to cut a series of circles and squares to get a handle on this.

[ger21]

Do some testing making using both conventional and climb cutting. You'll most likely see different sizes depending on which direction your cutting. One might be better than the other.
The only thing I can really say is to go over the machine with a fine tooth comb, and make sure every screw, nut and bolt is tight. Check for backlash. Every little thing adds up. Unfortunately, there's no substitute for rigidity.

[wizard]

I'm with Gerry, the very first thing to do is to go over the machine with a fine tooth comb. You want everything as rigid as possible fit up has to be as precise as you can make it.

Things to consider:

1. If you are sometimes oversized and sometimes undersized and the machine otherwise holds dimensions and positions, you might want to look into periododic error from a lead screw or other drive mechanism.
2. If you cut two circles at the same position on the machine are they identical? This might tell you something especially if you get random displacement on one axis.
3. Ultimately you will need to rework the machine attacking the worst axis first.
4. Finding the worst axis will require some tools. A dial indicator and stand should be part of your tool box.
5. 20 thou can be pretty bad (depending upon the intended usage of the parts produced) but to be honest I really doubt that many of the DIY machines do much better. If your expectations are higher you will have to put more money into the machine.
6. Also rigidity and backlash aren't the same thing. Backlash relates to clearances in your mechanical parts that cause a machine not to respond to movement commands or allow unintended movements. Rigidity relates to deflection of machine components which is another issue.

The reason this distinction is important is that speed and feeds this loads from cutting create variable deflections when rigidity is an issue. In other words a lack of rigidity means the hard you push a tool the greater the machine will twist out of its intended path. The impact of backlash however is far less impacted by feed rates and tool loads.

So if you vary the tool loading of your machine and don't see significant changes in the errors you are getting then stiffness of the machine MIGHT NOT be the issue. Note the word "might" here. Item 6 here is all about not assuming anything.

[Devastator]

Cut two identical circles, lay them directly on top of each other and then rotate them and see if you have an elliptical error. That's directly related to rigidity and there is no software compensation for it as far as I know of. Just changing the sizing won't fix that.

If it is sizing that is the only problem, you can just use an offset in software, no need to actually draw them bigger or smaller.

[Devastator]

Do some testing making using both conventional and climb cutting. You'll most likely see different sizes depending on which direction your cutting. One might be better than the other.

That's a good point. I recently went through this and found that if I used climb on outside paths and conventional on inside paths my parts were much more accurate and wood parts had less fuzz to clean off.