CNC Conversion Woes!

[LacL]

I converted a CT129N (BF clone) to CNC and I'm having trouble with my first pieces. The part which is a square is 45 mils too small on both x and y once cut. At this point, being a CNC noob, I'm not sure if the problem is h/w or s/w related.

What I have checked so far which resulted in no change:
1) gcode looks fine to me but I'm no expert. It does compensate for end mill dia and looks right in MACH3.
2) calibrated all axes in MACH3, but I did have issues when checking axis movement using MDI. ex. I calibrated for 6" movement which work fine but asking MDI to move 6" would be off but always returned exactly to 0 so no lost steps. I had to manually tweak the steps/rev values myself in the motor tuning tab. Anyone else have these issue with MACH's calibration function?
3) tightened and checked that there was no slack on the thrust bearings.
4) checked for and compensated for backlash.
5) increased the freq in MACH3 from 25KHz to 100Khz. (desperately trying anything)
6) turned off all plugins since I did write one for my MPG, just in case.
7) decreased speed and acc to half.

After all these checks/changes, movement using the MDI tab works perfectly as always and accurately with no lost steps, but when I load a gcode file and machine the part, its off by 45 mil each and every time. I've only tried this one part so I don't know if the error scales up or down with the overall size of the part. I'm using 1/8" thick delrin so it should not be putting any strain on the motors.

So can this be a software (MACH3) configuration issue or something to do with my drivers/setup? I'm using the C11 board on LPT1 in Windows 7 with Gecko G203 drivers.

Leo

[hoss2006]

Can you post your program so we can try it to see if we get the same error?
Hoss

[LongRat]

Measure your tool to confirm it is the same diameter as that which your G-code is compensating for.

[hndswthtshdws]

I agree with the above.

Sometimes the best way to measure tool diameter is to cut a slot and measure it.

[RTTIM]

Agree with the above.

Also are you using a collet or a chuck to hold the end mill? To be sure it's centered, you might want to use a dial indicator to check the runout on the shank of the end mill.

[LacL]

Thank you all for the input.

I've attached the gcode. I checked the extremes in MACH on test runs and the DROs show the expected values. I will try to setup a jig with a dial indicator to see what the table is doing without wasting any more material, but it will probably just confirm what the cuts are showing that the table is not where its supposed to be.

I measured the tool dia and it is 1/8" as expected. I also measured the width of the left over material and subtracted the width of the cut part and the difference is twice the tool dia to within 1-2 thou. So the end mill is not eating away more material than expected.

I did find this in the MACH3 manual,

Consider backlash compensation as a “last resort” when the mechanical design of your machine cannot be improved. Using it will generally disable the “constant velocity” features at corners.

Not sure how "constant velocity" affects this but since this piece has 8 rounded corners, I will try with backlash disabled and see what difference it makes.

[SCzEngrgGroup]

I agree with the above.

Sometimes the best way to measure tool diameter is to cut a slot and measure it.

Cutting a lot will not tell you the tool diameter with any precision. A slot will ALWAYS be wider than the tool diameter, due to tool flex. That is why, if you want to cut a slot to a precise width, you always start with a tool that is smaller than the final width, and make a pass down the middle, then finish passes along the two sides.

2-flute and 4-flute tools can easily be measured with calipers or a micrometer.

Regards,
Ray L.

[ninefinger]

Leo,

Look at this part here again - this is pointing to your problem.

When you ask for 6" movement how much are your getting exactly? Perhaps you didn't save the calibration correctly? MDI is G code - just line by line so if you're not getting what you want with MDI then you won't get it when you run a program.

My bet is that it is a configuration issue in MACH3 but I use Linuxcnc so I can't help there...

Mike

What I have checked so far which resulted in no change:

2) calibrated all axes in MACH3, but I did have issues when checking axis movement using MDI. ex. I calibrated for 6" movement which work fine but asking MDI to move 6" would be off but always returned exactly to 0 so no lost steps. I had to manually tweak the steps/rev values myself in the motor tuning tab. Anyone else have these issue with MACH's calibration function?

Leo

[LacL]

When you ask for 6" movement how much are your getting exactly? Perhaps you didn't save the calibration correctly? MDI is G code - just line by line so if you're not getting what you want with MDI then you won't get it when you run a program.

My bet is that it is a configuration issue in MACH3 but I use Linuxcnc so I can't help there...

The MDI moves were anywhere from a few thou to tens of thou off. Yet the calibration moves were perfect. I had to manually tweak the steps per revs value to get MDI moves perfect. I think I've solved this now by lowering the acceleration further. The calibration now matches MDI moves.

So far I have the speed set to 100 in/min and acc to 5 in/sec sec

However I still have the same error on the part being cut. It must be loosing steps but then again why exactly the same amount on both the X and Y axis and for each piece I cut? Wouldn't it vary somewhat between pieces and axes?

This is maddeningly frustrating.

[Winnfield]

Have you checked your tool runout?

[LacL]

I'm using TTS holders and it is within 0.001" as per spec sheet.

[Flenser]

The corner rounding in your post relied a lot on accel to keep the x-y motion at a constant velocity. Are the accel values for Z and for X and Y significantly different? I have had trouble in the past doing that, as if Mach3 used the wrong accel value to compute some moves.