Yep, it's not a trick question.

I just completed my X3 cnc conversion and have been making quite a few parts. I've adjusted all of the axis's to 0 backalsh at 0.5in and 5in. I can repeat this 10 times without fail. My only problem is that all of my parts are 0.01"-0.008" over the perspective measurement in the X and Y axis. I'm pulling my hair out on this.

Anyone have an idea on how to fix this?

just check ur tool dia again and also check for run out in tool holder....

just check ur tool dia again and also check for run out in tool holder....

I checked for run out in the tool holder: less that 0.001 at full rpm. What did you mean by check to tool diameter? Do you mean in Mach 3?

Have you checked for a burr in the tooling or dirt inside the collet?

Have you checked for a burr in the tooling or dirt inside the collet?

Check for dirt? Now that's something I haven't done yet. Would that really make that much of a difference?

Check the diameter of tool with Vernier calliper. If its right the other thing may be the STEP/UNIT to be corrected in Mach3 for both X and Y axis

Draw a circle of about 50mm diameter and check whether the circle is round or not...Check the diameter readings on 0-6Oclock and 3- 9 Oclock.

If you have small pieces of any type of debris in the collet or between the collet and tool or spindle it could cause additional runnout.

Since you have checked for runnout and you only have .001 then its not really bad however that could be .001 on all sides of what ever you are cutting.

Draw a circle say 50mm diameter and check the 3-9 oclock and 0-6 oclock readings...

I checked for run out in the tool holder: less that 0.001 at full rpm. What did you mean by check to tool diameter? Do you mean in Mach 3?

May be the case... if u gave the tool dia 3mm in ur CAM program and actually the tool u are using in ur machine is 3.2mm also with a little bit run outat the tip of ur tool.....u will be having the same trouble..

just my 12 x 70000/12 cents...

If you had to set your steps very far from 10160, I'd say something is loose and moving around some. Spanner nut pre-loading bearings? A block not fully tight?

When I got my kit, I put in 10160, and every move in the same direction checked dead on and when I changed direction there was .0007" backlash and every move in that direction was dead on.

After saying all of that,.......... Yea, I don't know either! :) sorry

Good luck,
Jack

P.S. If you had dirt or something in the collet, the endmill can get out-of-round. You can check that by putting an indicator on one of the flutes of the endmill and spin the spindle.

P.S. If you had dirt or something in the collet, the endmill can get out-of-round. You can check that by putting an indicator on one of the flutes of the endmill and spin the spindle.

Just spin the spindle by hand:)

The tool diameter was 0.498 and wasn't any dirt around the collet. :confused:

Now check from catalogue of the tool, what is the actual diameter???....I think this will be 0.5" tool.....

Now check what tool dia u have given in ur CAM program, if it is 0.5" then change it into 0.498 and generate new G-code and run the machine...ur problem will be solved...

Have you mapped or measured the screws? Pitch can vary from screw to screw by several thou per foot. TPI doesn't always mean EXACT TPI! You might have to adjust the step/inch in Mach to correct for pitch error.

The tool diameter was 0.498 and wasn't any dirt around the collet. :confused:

Did you get your measurement from the flutes or the shank?

Stepper Monkey,

Please post a link or directions for mapping the screw.

It might not be easy to do without directions.

Not attempting to being facetious by the way, many of us have no idea how to map the screw.

Jeff...

There is some good info on screw mapping on the Mach forums if I remember correctly. Mach even has built-in it as a feature. There is an even simpler way to do the same thing though that works for the relatively short screws on most mills.

In short, if you have a screw pitch that is off, just because you tell it to go 12 inches doesn't mean it actually travels that far. It could actually travel 12.010" or 11.992" or whatever. However, this is a fixed error. For instance if you tell it to go, say, 121,920 steps (or an ideal 12" in this example) in one direction, it may actually travel 12.010". Tell it to go the same number of steps in the other direction, (ignoring backlash) it will obviously travel the exact same distance back. Therefore you have perfect repeatability like he mentions, but still has a problem with the actual size of parts being oversize. Textbook screw pitch issue.
If you carefully measure actual distance traveled, then calculate the proper adjustment to the steps/inch value in Mach, then retest, you will quickly find the right number for each individual screw on your machine. Test several times and at several points in both directions to be sure.
You don't need to go really nuts though, as if a screw pitch varies, and they often do, it is thankfully almost always very consistent over the length of the screw unless you are talking a very long screw (several feet). Therefore mapping inch-by-inch pitch variation isn't usually necessary with smallish mills. If you simply measure how much travel you actually get over a long movement versus what you are supposed to, you can often just divide that error evenly and assume a fixed amount per inch of needed correction over the entire travel and be done with it.

Did you get your measurement from the flutes or the shank?

That was measured on the shank. I'm using 3 flute end mills. It's kind of hard to measure it on the flutes.

If the bit is 3-fluted, and center cutting, I wonder if you could do plunge-cut straight down into a piece of scrap metal, and measure the diameter of the hole.

[I can't find it now, but I have seen pages for measuring such things as bolt circles, there is probably a way to trig' the tip measurements out, from point to point approximating the diameter.]

I ran into the screw pitch issue after I installed a belt drive on my Z axis. The screws are preloaded ground ball screws. These were perfect before I installed the pulleys and belt. These are toothed pulleys. I think they were 10 and 20 groove pulleys. I expected to have to exactly double the step count, but after measuring, I had to add another 130 steps more than I had expected to come out near perfect. I scratched my head on that one for awhile, but it's now as close to perfect every time as I need. The step count just had to be adjusted in real time as opposed to what the lead and belt drive should have rendered.

3-flute endmills(EM) are hard to measure. I was asking because the (EM) shanks are a little smaller then the fluted area. I couldn't remember how much smaller, so I went out to check some of my (EM). On my .5 (EM), the shank measured .4995 and the flutes measured .5005

Even if you had programmed a cut for a end mill that was actually .002" smaller, that would only account for .004 of your .008-.010.

Yep, it's not a trick question.

I just completed my X3 cnc conversion and have been making quite a few parts. I've adjusted all of the axis's to 0 backalsh at 0.5in and 5in. I can repeat this 10 times without fail. My only problem is that all of my parts are 0.01"-0.008" over the perspective measurement in the X and Y axis. I'm pulling my hair out on this.

Anyone have an idea on how to fix this?

If it were cutter or collet runout , it would cut oversize on inside cuts and undersize on outside cuts. Regardless how many flutes it has, you can use a test indicator at the shank and mark the high spot. Then z up to place the flute end at the indicator stylus. There by rotating the spindle by hand in reverse to see where each flute is in relation to the shank high point and noted runout.

What type of material? Are you running a rough and finish pass? Good quality cutting edge with appropriatre speeds and feeds?

Light duty machine repeatability verses no load table motion, I'd expect to be very different than real world cutting pressures. Flex, gib adjustment and bearing slop could all contribute to your conditions.

In most circumstances, you may need to do a first article inspection and comp the cutter a wee bit to get it to cut at the size you expect verses what you get.

Many other good suggestions here to help hunt it down.

DC

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.

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

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.

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.

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

Well, that's not quite IT :D

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.

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

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.

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