parts with arcs grow

Having some issues with my DIY water jet that really doesn’t make sense and I am curious if you guys can give me any feedback or anything I can try. Currently the steps per in are near perfect to within a few thou. If I cut a square the part dimensions are perfect. However I am cutting a bracket for a customer that is roughly 12.75" by about 3" at its widest and 2" wide at its narrowest. The part has large radius on every edge and bolt holes in the middle. Also the part grows about .125" on the long side and a little on the short sides. I found a post from Les about going from incremental to ABS arcs by changing 2 lines of the post and that didn’t make any difference. I haven’t tried cutting in line segments versus arcs. I have a test part that is correct dimensions that was cut on a pro water jet. if I run the gantry again the end stop and move it the exact length of the part I can put the part in between the carriage and the end-stop and its perfect.
Things I have also tried:
Switching to metric units
Ive used both mach3 and uccnc (both sae and metric on both)
abs vs incremental arcs ( I think i did it properly)
MP1000 THC post (same on I use on my plasma without any issues)
replaced gear rack and gears with different brand in case there was issues

Gantry runs very smoothly, 46v, mx4660, uc400eth to a laptop. Same laptop make and model I use for the plasma without issues


My first guess is that the machine is losing steps. Move the torch to 0,0 and drop it down to touch the work. Draw a circle around the torch. Now cut your part then move mack to 0,0 and drop the torch. Does it line up perfectly with the circle? If not you are losing/gaining steps. In that case try reducing the acceleration and maximum speed in Mach3/uccnc.

I can rapid the machine at 2000IPM with accel at 150 all around the machine darting back and forth and it will still come right back to the end stops consistently. Current I have it turned down to 400IPM and i think 30ish accel. The parts only seem to grow if there arcs, squares cut just fine.

Did you try the test I suggested? With stepper systems you can get resonance problems at particular speeds that are far lower than your maximum speed.

Yes, I have tested in ability to return to the same spot at a range of speeds and feeds including the speeds I cut at.

Was that with one of your problem jobs?

Yes. are there any posts for mach that will do line segments as opposed to arcs? I know ill lose cut quality most likely but its worth a shot.

Go to Options->machine->post processor and click the ‘Edit post’ button.
At the very end of the post add this code:

function OnArc()

That will break arcs into short line segments.

Any other ideas as to what could be going on? Doesnt make sense to me at all.

I am assuming that (.01) is a variable? I am currently running in inches, do I need to make that number smaller?

The post always works in metric internally so you don’t need to change the number. The 0.1 controls how many line segments are generated. Decreasing the value increases the number of lines.

Tried the segments and it made no difference at all. Confirmed again that it was going back to the same spot after cutting the file and it is. I am sort of at a loss here. I manually jogged the machine to the top of the part using the gcode visualization window on mach3 and then put the template for the part that I am trying to make between the endstop and the carriage and it seemed to fit perfectly.

Looking at your fault description again it could possibly be a scaling issue. Use the measure tool in SheetCam (Right-click->measure) to confirm your drawing dimensions are correct. If they are, check your machine. Put a tape measure on the machine parallel with the X axis. Line the torch up with a convenient point on the tape and zero the X axis. Move the axis a decent distance and check your tape and X axis agree. Repeat for the Y. I know you have checked the machine movement with your template but sometimes it is best to go back to basics. One company I worked for had a machine that was producing parts that didn’t quite fit. After doing the tape measure test I found it was out by about 2mm per metre on the X axis only. The machine manufacturer declared that was impossible and refused to change it. They were so adamant that I went back and checked again with a couple of different tapes. Each tape read slightly differently (It’s surprising how much tapes can vary) but the error was still there. In the end I wrote a plugin for my employer’s cad package that automatically scaled the drawing before exporting it to the machine.

Are you any good at calculating steps per in off the gear rack specs? I was also thinking about upping the microstepping which if I’m thinking correctly would reduce the error percent of the calculation. Currently running at 1/10. Was thinking I could at least go 1/16. My gear rack specs:
16 tooth pulley

I just madr a fixture to measure the template and cut part and used that to scale my steps and came in about 999.xxxx from 1011.xxxxx. Going to cut another part tomorrow and test the new numbers.

Do you have some sort of reduction between the motor and gear? Your steps/unit does not seem right for a 16 tooth 1 module gear and direct drive with 10x microstepping. If you have reduction, how many teeth on each pulley?
Did you do the tape measure test? It will be more accurate than measuring your parts.

Its direct drive. No reduction

OK, a 1 module rack will move 3.14159mm per tooth. Convert to inches 3.14159 /25.4 = 0.12368". Multiply by 16 teeth 16 * 0.12368 = 1.97896" per rev. Most steppers are 200 steps/rev so you should get 200/1.97896 = 101.06338 steps per inch. If you are using 10 microsteps your steps/unit should be 1010.6338

Changing from 10 microsteps to 16 will not make a significant difference to accuracy. It will just change your steps/inch.

Did you do the tape measure test? As I said before measuring your part is not ideal as it can introduce other errors.

Thanks, So I was very close for the steps per in and even then the part grew by over .125" I ran it with a tape measure over 40 inches and it was within .0625 which is likely due to the end clasp being loose. I can send you the sheetcam save file to take a look at if you think that will offer any insight. I confirmed that after the job was cut that it returned to the same start point.

If you give me a copy of the job file (File->save job) I can check it over for you.

I just sent you a copy of the job file as well as 2 of the gcode files I was using. I really appreciate the help.