There are several tests you can perform on your CNC machine.  In this post I will concentrate on a positional accuracy test. 

 

Accuracy can best be described as the ability to place your machine at a particular position. I recommend using this test when first constructing a new CNC or mill and later after tuning.  This will give you a set of benchmarks for trouble shooting problems later.

You will need a the following:

 

 

Digital Dial Indicators

This one was purchased from Harbor Freight, but you can find them on amazon.

I highly recommend this digital indicator

Update

I started using this dial indicator and got much better results.  I shaved off .001" inaccuracy in most readings. I suspect the reason was the Harbor Freight unit sticks a little when you try to use the full 1" of range.

I purchased this unit (Accuremote) off Amazon when I was doing my 3D printer series.  I used it to level the bed on the Rostock Max 3D printer and it worked perfectly.

It works even better in this application.

Test Indicator Base

While any base will work I prefer the test indicator bases, like the one shown here.

 

Scrap Steel

You need a heavy piece of steel to hold your indicator. It needs to be heavy enough to stay in position as the machine takes it through its range.

 

Test Procedure

The test procedure is simple. Back the machine up against the indicator fully depressing it then back it out about 1/32", Zero the meter then run it out a full 1". Write down the reading the push the meter against the machine until it reads zero (or very close) then run machine out another inch". Write the reading and repeat.

Here is a quick video showing the results on one of my machines.

 

Readings

Your readings should be better than the mechanical resolution of your machine.  The mechanical resolution is your Steps per Inch divided by the micro stepping. 

For example, this machine has 1909 steps per inch and is using a G540 with 1/10th micro stepping. So in this case we don't want a reading of more than 1 mechanical step or .0052".

 

Results using first Meter

Reading of .997 

Subtract from 1 and you get .003

Reading of 1

Subtract from 1 and you get 0

Reading of.9975

Subtract from 1 and you get .0025

Results with Accuremote Meter

Reading of 1

Subtract from 1 and you get 0

Reading of .999

Subtract from 1 and you get .001

Reading of.9995

Subtract from 1 and you get .0005

 

You will want to take readings across your complete axis on both X and Y.  If you find some that are more than .005", repeat to make sure. If it is still high, its possible you have a worn pinion or rack. It could even be buildup on the rack or pinion.  In an extreme case it could be a damaged rail. 

 

Keep in Mind

This test does not measure backlash, or runout. Both of which can bring the accuracy of your machine down.