Problem- Opposite of Backlash?

[Robin Hewitt]

Are you cutting uphill or down hill?

Is the tool tip cutting or are you cutting through?

I'm thinking flex in the column?

Downhill pushes the tool tip away from the work, uphill makes it want to dig deeper.

[MRM RCModels]

Are you cutting uphill or down hill? Is the tool tip cutting or are you cutting through? I'm thinking flex in the column? Downhill pushes the tool tip away from the work, uphill makes it want to dig deeper.

I'm climb cutting and I think you're right about flex in the Z Axis column. I tightened every bolt on the assembly and still couldn't get better than 0.01" backlash. Rather than use the backlash adjustment, I increased the steps/rev count in Mach3. I'm not a fan of the jumpy movements of the backlash comp.

________________________________________________

First and foremost, thank you all for the great advise on how to fix this problem. I am happy to report that I cand cut squares and circles within 0.001" true on all axis's (see pics). This is how I fixed it:

First I set my step/rev in mach to 10161 for all axis's. Then, systematically tightened every screw, bolt and gib on the machine. Next, measured the backlash on every axis and adjusted the difference using the step/rev function in Mach3 to those listed below:

X Axis: 10191
Y Axis: 10161
Z Axis: 10781

I've come to the conclusion that the X3 is not a precision instrument. More like a really sharp axe. It's great for large jobs, but it has it's limitations .

Thanks Everyone,

Kelvin

[Stepper Monkey]

So it sounds like it was was just a screw pitch issue after all?

I'm glad it worked out to be something so easily correctable for you, even if it took a lot of trial and error to get there! At least with that specific problem you know once its fixed its fixed - it won't be coming back to haunt you again.

[MRM RCModels]

Originally Posted by Stepper Monkey So it sounds like it was was just a screw pitch issue after all? I'm glad it worked out to be something so easily correctable for you, even if it took a lot of trial and error to get there! At least with that specific problem you know once its fixed its fixed - it won't be coming back to haunt you again.

I think so. I really can't do anything else to improve precision, so that has to be it.

[Robin Hewitt]

Originally Posted by MRM RCModels I really can't do anything else to improve precision, so that has to be it.

Well, that's not quite IT

If you're cutting to one thou on a cheapo bench mill you are doing extremely well and will probably need to fight to maintain it.

[One of Many]

Originally Posted by MRM RCModels First and foremost, thank you all for the great advise on how to fix this problem. I am happy to report that I cand cut squares and circles within 0.001" true on all axis's (see pics). This is how I fixed it: First I set my step/rev in mach to 10161 for all axis's. Then, systematically tightened every screw, bolt and gib on the machine. Next, measured the backlash on every axis and adjusted the difference using the step/rev function in Mach3 to those listed below: X Axis: 10191 Y Axis: 10161 Z Axis: 10781 I've come to the conclusion that the X3 is not a precision instrument. More like a really sharp axe. It's great for large jobs, but it has it's limitations . Thanks Everyone, Kelvin

This may work for 1" in that portion of the table and screw in concert with those settings. You could verify them with a grid of 1" squares to be certain there is no degradation or accumulation over nominal accuracy within the machining envelope. You may find in mapping any grid deviations, settling on a happy average steps/rev is as good as it gets. If anything else, you will get a feel for the trusted work area with its full travel.

DC

[LongRat]

I agree, cutting to 0.001" (25 µm) accuracy on this type of machine is a very good achievement.

I have noticed a periodic variation in the pitch over one revolution of the screw (Nook 5/8" * 0.2 ballscrew). It is described in the Mach3 manual as 'thread drunkenness'. If you cut a diagonal line in x and y and put a straight edge up to the cut, you see a periodic wobble where it should be straight. At least, that is how my set up is. Apart from backlash, this is the largest contributor to error that I have (X2 machine). The only real way of getting over it would be to upgrade to better screws, i.e. ground.

[bbox]

Kevin:

I run an engineering company with about 25 cnc machines of various types. I enjoy making hobby cnc machines at home and have done four to date, not including my current Super X3 project. We have talked on your great build diary on another thread.

While you can achieve better accuracy than you are now experiencing, .001" over a few inches is pretty good performance. We can do better than that at my company, but it usually involves test pieces, fine tuning the toolpath, etc. Not to mention expensive machines! The advice you have been given by the guys here is good.

Measurement is a big variable, also. We have experienced many problems with out-of-calibration measuring tools even though we have a regular calibration program. Measuring less that .001" can be surprisingly hard to do accurately sometimes, especially over a few inches. Sounds hard to believe, but it's true.

In the commercial world, +/- .0005" for a total swing of .001" will add considerably to the cost of the part. Don't be too disappointed if that's the best you can do. I'm always amazed we hobbyists can do as well as we do for a tiny fraction of what a commercial machine costs!

Bill