How do you achieve best accuracy?

[Menatep]

What's the best way to measure accuracy of a cnc router that uses steppers? Could someone give me some good ideas how to achieve best accuracy?

My brother and I have a cnc machine and today we tried to measure how accurate it is. We put a sharp router bit in the spindle, lowered the Z-axis until it made a small impression mark in the temporary mdf clamping table, raised the z-axis and moved it 8 inches in mach 3 along the x-axis, lowered the z-axis again to make the second impression and raised it back up again.

Took 3 measurements with a digital caliper and on average they were about 7.984 inches. We are off by 16 hundreds of an inch and our accuracy is about 99.8%.

I'm not really sure if the 16th hundreds of an inch is because of measuring error or positioning error.

Our machine specs are:

- Gantry weight without router = 90 LBS
- X-axis : Fully supported THK 35 rails and bearings (1 bearing per rail)
- Y-axis: Fully supported Dual Vee bearings with hardened steel rails (cam adjusters on bottom rail)
- Z-axis: Fully supported IKO 20 rails and bearings (2 bearings per rail)

Ball screws:
X & Y : Nook 5/8" - 5 tpi - with 2 Nook recirculating ball nuts
Z: Nook 5/8" - 5 tpi - with 1 Nook recirculating ball nut
ALL ball screws are supported by a bearing block on each end, the fixed end has two sealed dual angular ball ball bearings to take axial and radial loads, the other end has two sealed ball bearings.

Steppers are nema 34 640 OZ for X and Y axis. Z axis has 420 oz nema 23.

Electronics are: 3 G203V drives, Grex G100 Motion controller wired to a laptop through ethernet.

[lunchtrayrider]

you should try the measurement over different distances in different places to see if it is alignment or backlash

[zeeway]

Firstoff, the error is 16 thousandths of an inch (not hundredths)...hope that makes you feel better.. You would be more accurate in this range by using a dial indicator setup - your method of using a digital caliper will introduce error on its own. You can buy a dial indicator and a stand for about $25 at discount tool outlets - there's always another tool needed, right?

Is this error consistent in the plus and minus directions? You will need to check every component in the machine to make sure it is tight and square. Are you using Lovejoy-type couplings - these typically allow backlash with elastomeric elements that do not fit tightly. And you cannot rule a slight error in the ballscrews - for this you can calibrate the the movement stated by your software to your actual movement.

Good luck.

[ger21]

Isn't a line drawn with a Sharpie about .016? If it's the same in different places, adjust your steps/unit until you get the correct movement.

[Menatep]

Thank you for your helpful replies. we'll look into this over the next few days and see how things can be improved.

We are not using sharpie to draw any lines.. we just have a sharply pointed router bit that we plunge into the mdf surface gently to leave an impression mark. I guess that could also be part of the problem.

[ger21]

Sorry, I guess my eyes are going bad.

[Drassk]

Isn't a line drawn with a Sharpie about .016? If it's the same in different places, adjust your steps/unit until you get the correct movement.

A Bic pen line is somewhere around 0.014 while a Sharpie is in the 0.03" range when new...I needed some common frame of reference to tell a client how thinly a material could be cut once

[rowbare]

Check out this video and see how Hoss does the measurement. Basically zero a dial indicator against a stop, put a bar of known length against the stop and zero the indicator against it.

"http://www.youtube.com/watch?v=XkO5tc-jSxw"]YouTube- Improved Mach 3 Axis Calibration

bob

[RomanLini]

I created a simple "live" backlash test you can see in this thread;
http://www.cnczone.com/forums/showth...766#post747766

Which cuts a test shaped pattern that can be measured with a micrometer and produces double the backlash error to improve the measurement. The "live" measurement of backlash with the machine cutting a test piece may be different than the static backlash measured with a dial gauge, due to friction, tool forces etc. At least on my machine it was.

[Menatep]

What kind of dial indicator should i get?
Is digital better than non-digital?
What's the maximum measurement should i look for? I've seen some indicators for 0.5" and some for 1" but i think that won't give sufficient accuracy. The bigger the measurement range of the indicator the less margin of error should be, right?

[zeeway]

Non digital (analog) is fine and usually cheaper. One inch of travel is more than sufficient to check backlash. I don't know if you have Harbor Freight (HF) discount tools in the frozen north (Canada, eh), but HF has the dial indicators with magnetic base on sale for $25 or so every few weeks.

[Zygoat]

i just picked one up last week. checked the receipt (sitting right next to me) $22. i think the sale goes till the 23rd.