CNC Real World Accuracy

What are realistic practical targets for CNC machining precision and consistency? September 27, 2009

For those of you with a CNC, what is your design thickness of a tongue, or width of groove, and what is the actual measured number and variation of the piece that comes off the machine. For example: designed at 10.4mm actual 10.2-10.6mm etc. Also machine and bit type would be nice but not necessary. I have also found that the dimension of a groove at the top of the groove and bottom of the groove varies. What have you found?

Forum Responses
(CNC Forum)
From contributor S:
Accuracy and wood are diametrically opposed. I think it's good if you can machine wood/PBD/MDF to +/-.2mm (or more). Even IF it could be machined to perfect accuracy, it's going to expand and/or contract by that much in a few hours.

From contributor W:
CNCís are definitely are more accurate than the material. You can cut perfect parts in the morning and have more than a few thousand in movement by lunch. What kind of accuracy do you need?

Click here for higher quality, full size image

From the original questioner:
I guess one of the things I am looking for is perhaps not so much specific numbers but consistency. What sort of dimensional deviation should I expect within the tongue itself, regardless of what the design dimension is?

From contributor W:
It would depend on the quality and rigidity of the machine and the runout of the tooling/spindle. I think a mid to upper level closed loop servo driven machine would hold the =/- .001"-.002". Plenty for any type of wood joinery I am aware of.

From contributor M:
I generally use a working target of .1mm or 4 thousandths. My hardware is capable of accuracy and repeatability well in excess of that, at least with the spindle. The drill bank is just a little less accurate, but well within reason for shelf and assembly holes. Like contributor S says, you are doing pretty well if your real life results are .2mm or better.

This goal is not as hard to achieve as it sounds with care and consistency in a few areas. Better is possible but really requires some work. This work may occasionally be desirable if you are doing solids joinery or something really special, but really hardly ever that I am aware of.

In the real world the biggest limiter to accuracy in actual performance is accurate tool measurement. There are issues like getting an accurate measured length of less than .001mm requires care and decent measuring equipment. Even .1mm requires attention to detail and reasonably decent measuring tools.

For example, how do you get the length of your tools? Can you really measure the same tool 3 times and get the same number? Dust alone can be a big barrier to accuracy. How do you get diameters, especially for 3 flute tools? If you are running a spoilboard is the set of your spoilboard cutter knives accurate to less than .01mm? (Likely not unless you are jointing after setting). All of these things can be done by woodworkers, and are done by machinists all the time. We just have to actually think about them and get down to basics with it.

The next biggest limiter is positioning of the workpiece. If you nest this it is less of a problem because the bit cuts the perimeter as well as all other operations. If you are working off pins it is a big deal. If you are on pods, you have additional consideration in that small movement can occur during machining if the vacuum doesnít hold firm to the pod and the gasket wiggles, for example, or the workpiece flexes between pod positions slightly. All of this, and the factor that others have mentioned, that wood (and most sheet stock as well as plastic) moves with temperature, humidity or both.

From contributor J:
"How do you get diameters, especially for 3 flute tools?" You cut a test part and measure it. If it's incorrect, then you adjust the size of the tool in the controller parameters. Same goes for the length. Tool size is only a starting point.

From contributor M:
That's exactly what I am talking about; see the question was for discussion on accuracy, not that I or any other competent person does not know how. Cut that same groove in MDF, melamine, solid wood and acrylic and measure the grooves. Then measure all of those grooves again tomorrow morning. Tell us what you actually find and to what degree of accuracy. It isnít so straightforward all of a sudden I bet. Unless you are using a tape measure, those grooves aren't going to measure the same just after cutting never mind later.

For that matter, cut a groove in melamine, a strip 8 feet long. I bet (if you measure accurately) you will find that at the entry point or very near it itís one thing and due to a small amount of bit deflection as the speed increases you get a slightly different reading at the middle of the cut.

Some people trust the sharpener to mark their tools, and this works pretty well, to a point. Mine came in for quite some time marked in thousandths. (way more accurate than the "cut a groove" method). Now I use a calibrating tool that measures the diameter directly.

I have never known anyone that gauges length by cutting a dado. Most use a touch pad or a measuring stand. If you are happy with the results though, and it works well for you, have at it. Maybe you can get to .1mm that way, but I can't seem to get consistent results with that method. I attribute it mostly due to a variation on pull on the spoilboard depending on how thick it is and how well covered it is, and also the fact that the accuracy of the "cut a groove" method depends on the accuracy to which you have already gauged your flycutter.

You are absolutely right in saying that tool size is only a starting point. If you start with accurate tool sizes diagnoses of other issues is possible. If you are shotgunning or guessing at tool size then all else is futile.

From contributor T:
I apologize that this is a tad late, but MSC , Enco, Travers, etc. sell flute mics. These are micrometers designed to measure odd fluted tooling. Good points by all, even lower end machinery is inherently more accurate than most of the materials cut on a router!

Click here for higher quality, full size image