Precession cutting

[Albert Nix]

Being rather new to the laser scene I am not sure what to expect from my MII (40w). Naturally the sales folks said I could be very accurate with the machine. I have been tring to perfect the art of cutting shapes to an exact size and seem to be cutting up a lot of scrap acrylic tring to get it right. As a simple test I tried cutting 1" squares out of 3/16 acrylic at different settings and hardly ever came up with the same measurments. Top to bottom may read .996 and sides .987. I know the beam width of my 2" lense should be .005 and there will be some burn back. If I could get a constant reading I could compensate for the rest. How consistant can you guys be with your machines. I may be expecting to much!

Thanks Al

[David Fairfield]

The variance you're measuring seems a lot. Any variation I get is negligible, and I do extreme miniature work.

The laser cuts at a slight angle, if you measured the bottom surface of one piece vs the top of the other, that might account for the difference?

Dave

[Joe Pelonio]

The variance you're measuring seems a lot. Any variation I get is negligible, and I do extreme miniature work.

The laser cuts at a slight angle, if you measured the bottom surface of one piece vs the top of the other, that might account for the difference?

Dave
Agreed, try turning it over an you'll get a different measurement. The beam is hour glass shaped and focused on the narrowest part. The thicker the material the more of an angle the cut is. If they need a perfect 90 degrees on 3/16" material a CNC will be more accurate than a laser.

[Albert Nix]

angle does not seem to be an issue. I measured at the top of the piece and the bottom and get the same. I will give you and example. The items that I am tring to make are templates for grinding profile blades for like molding. One end has to be index has has to be .787" from one end to the edge of the profile. I may have to cut two or three of them to get it indexed. You would think the knife profile would be the hard part but it's not that bad.

[David Fairfield]

Sounds like the machine isn't delivering the precision it should and you're definitely not expecting too much. A square should be square! The tech people should be able to sort it out for you. Keep us informed

Dave

[Jack Harper]

I'm not so sure your variance is to high. Your machine should have an official variance tolerance that is guaranteed by the manufacturer. For low cost CNC like these lasers, 3-4 thousands does not sound that abnormal. My $250K waterjet is considered to be the one of the most accurate manufactured and it has a 2 thousands variance. I would check with your manufacturer, not the salesman, and just see what the guarantee is for your model.

[Martin Reynolds]

angle does not seem to be an issue. I measured at the top of the piece and the bottom and get the same. I will give you and example. The items that I am tring to make are templates for grinding profile blades for like molding. One end has to be index has has to be .787" from one end to the edge of the profile. I may have to cut two or three of them to get it indexed. You would think the knife profile would be the hard part but it's not that bad.


Doesn't sound too bad in 3/16" material. Are you using air assist? if it comes in from the side, then it will affect horizontal and vertical cuts differently. Also, the thickness of the material can vary, which makes the beam intersection different.

We've made a lot of circular parts that have to be a push fit on a 1" acrylic tube. The fit is very sensitive to air, focus, and material thickness. But it has to be +/-.001 to work, and we can do that.

I am suspecting that your squares are all consistently off.

To rule out calibration issues, you might cut a large square. If the error follows the size, then it is calibration. If it stays +/- .005, then it is focus, material, air and other tolerances.

[Rodne Gold]

If you have a GCC mercury (mII?) there is a method of calibrating the X and Y axes so they are equal and to compensate for errors , check the driver and documentation.
4/1000th of an inch is perfectly acceptable in terms of tolerances , considering pex expands and contracts when heated and cooled and that there is some heat affected zone melting when cutting it (IE there is additional melting either side of the beam width)

[David Fairfield]

Agreed.

Albert's problem could have something to do with the heated material warping from heat expansion and rising up a little from the focal plane, where the beam is wider. Suggest running a test with thin material like cardstock.

If material warp/expansion is the problem, you might be able to compensate by cutting the plexiglass in stages. Instead of a solid line, divide the drawing up into 2 or more sets of evenly spaced dashed lines, allow material to cool inbetween burning sets. And /or cut power in half or a third and run the drawing 2 or three times. You can adjust focus for depth of the cuts.

I'm not sure I could do some of the things that I do if I were dealing with a .01" variance and Albert is reporting .009".

Dave



If you have a GCC mercury (mII?) there is a method of calibrating the X and Y axes so they are equal and to compensate for errors , check the driver and documentation.
4/1000th of an inch is perfectly acceptable in terms of tolerances , considering pex expands and contracts when heated and cooled and that there is some heat affected zone melting when cutting it (IE there is additional melting either side of the beam width)

[Rob Bosworth]

Are using CorelDRAW for your graphics? If so, you might be getting different sized parts partly due to the dimensional instability of a pixelated graphics platform like CorelDRAW.

If you need better dimensional stability, you might try generating your parts configuration in a higher quality CAD package.

Also, how fast are you trying to run your parts? These machines are basically rubber tired rollers and rubber bands. How many of you got a guarantee with your laser system for true tolerance of the machine cutting. I've been in this industry a long time, and I'm not sure I ever got someone to commit to a specific tolerance number from any of the manufactures.

[Albert Nix]

Thanks for all of the suggestions. To bring everyone up to speed after reading what you guys had to say on the matter I tried changing some more variables. I ended up with a 4" lens focused about .125 below the top of the 3/16 plexi, 90% power 1.5% speed 1517 PPI, added .012 to the width and .006 to the height. I was within .001 to .002 anywhere that I measured. Close enough for me!! At first I thought the 4" lens was not goning to work because I was getting to much taper on the sides, but the lower I dropped the focus point the more it went away.

Thanks: all

[Kevin L. Waldron]

A lot of lasers allow for other air sources besides compressor air. Have you considered nitrogen for a gas? It makes substantial difference for me in flame out, consistance, and heat; I regularly cut thick acrylic with 120 watts with nitrogen at about 25 psi and the gas seems to last forever.

Just a thought.

Kevin

[Lee DeRaud]

Are using CorelDRAW for your graphics? If so, you might be getting different sized parts partly due to the dimensional instability of a pixelated graphics platform like CorelDRAW.Explain please? Unless somebody changed the definitions, CorelDraw is a vector-based program and AFAIK it keeps its coordinates in floating-point format until they're converted to the laser's coordinate system by the driver.

[Richard Rumancik]

I . . . added .012 to the width and .006 to the height. I was within .001 to .002 anywhere that I measured.

This method will work for a square but will not work for a general shape, or anything with radiused corners or a hole in it. If you ever want to do other precision shapes this is my suggestion:

1. Get a few sheets of white poster paper and create a calibration file for the laser. Draw some short (.50") lines at say x=1" and x=23" (22" apart). Draw some short perpendicular lines say y = 1" and y= 17" (16" apart). The larger the better, but you need to be able to accurately measure what you plot.

2. Get a precision 24" steel machinist rule and an optical loupe to read the graduations.

3. Plot the lines on the cardstock with the laser to make a fine cut.

4. Measure the distance between them in x and y axis and see if your laser is plotting in scale in each axis. You can repeat the test by moving the paper around slightly. Determine the necessary correction factor as a percentage.

5. Using the required lens, material and settings, find out your kerf width using a feeler gage. It may be a bit different at the top and bottom. Use the location that is important to you.

6. In Corel, convert the shape to curves, combine, and offset the outline(s) by half of the measured kerf width. Delete the original shape.

7. Scale the shape in Corel using the scale factors determined by the calibration exercise. (eg the shape might need to be expanded 100.22% in x axis and 100.10% in y axis)


Yes, this will sound really complicated but it is not that bad once you have done it. You don't have to calibrate very often. The biggest problem is keeping track of the files as you don't want to destroy your original file. Once you have added kerf comp and calibration factor, the file is basically uneditable and you have to go back to the source.

Rodne suggested using the calibration option in the Mercury driver. I have an older Mercury and found it did not have enough resolution. It ignored every second setting (rounded off) and it would jump maybe .010" per step so I was never able to hit the target value. Eg with your square I could get say .995 or 1.005, but not hit 1.000". Maybe the newer Mercurys are better. Is yours a Mercury?

I asume the error is a result of tolerances in the drive system (toothed pulley sizes etc.) However, the amount you are out in the Y axis seems excessive to me. Maybe you have a bad pulley.

I really wish they would allow for a calibration factor in the firmware or else an accurate driver method so it could be set once and then left alone. That would save me from making "stretched" files.