[BanduraMaker]

I've been suspecting that my steps per inch settings might be a little bit off. My machine has 10mm pitch ballscrews all around and with a calculated setting of 5080 steps / inch, I seem to get .999" instead of 1. I can't detect any backlash as when I command it back to zero, it goes to zero.

I'm finding that with a setting of 5083.333 it seems to be right at 1". I've checked the X & Y axis and it seems to be true for both. I checked the X in two spots along its travel as well.

Has anyone else seen something like this?

Thing is though that the dial indicator is a $10 one from enco or something like that and 1" is my longest travel in an indicator. I eyeball set it up to be orthogonal to the axis I'm measuring but it is just an eyeball setup.

Any recommendations on whether I should just go with 5080 or should I go with 5083.333? Is there some better way of measuring over a longer distance? Losing .001" / Inch is kind of a lot!

TIA

[wendtmk]

If you have a metric ballscrew, shouldn't you have your settings be steps per millimeter or something like that? Conversion inaccuracies are probably what's causing your problems.

Mark

[Torsten]

Some components like rolled Ballscrews often have a cyclic runout meaning that they may be off by a certain amount over a certain distance and return back to zero at the beginning of the next cycle.
To test this you could check the runout at 1. inch intervals over the entire Travel of the Axes. Then it will be aperent wich Step setting will give the best overall acuracy.
I have a Rack and Pinion that cycles by about .004 inch every 3.5 inches.

[ger21]

Your screws probably have a manufacturing tolerance of ±.003"/ft.
It's pretty common to have to play with the steps/unit setting to get within .001".
It's hard to measure, but it's also possible that linearity changes along the length of the screw.

If you'r really losing .001" per inch, then cutting a 30" part would be about 1/32" short. You should be able to check that with a tape measure.
If a long part is pretty accurate, then I'd probably leave it. But I'd try both 5080 and 5080.33333333 on long parts and see which is more accurate. Subject to your ability to measure, of course.

Be sure to use at least 6 or 8 decimal places on your settings, though, to minimize accumulated rounding errors.

[JerryBurks]

I checked the calibration/linearity of my current setup with a precision 24" etched steel ruler, a cheap USB microscope camera attached to the Z-plate and a (removable) hairline on the computer screen. The hairline allows to step jog to the center of the fine etch markings with one or two mil accuracy which is decent for a 24" distance. This also helped me find the limits of acceleration when the machine started losing steps.

I do not have a calibration cert for the ruler but I believe it is pretty good.

[robe_uk]

I checked the calibration/linearity of my current setup with a precision 24" etched steel ruler, a cheap USB microscope camera attached to the Z-plate and a (removable) hairline on the computer screen. The hairline allows to step jog to the center of the fine etch markings with one or two mil accuracy which is decent for a 24" distance. This also helped me find the limits of acceleration when the machine started losing steps.

I do not have a calibration cert for the ruler but I believe it is pretty good.

Thats a cool solution, I used a laser cross hair on a standard steel tape over a 1200mm distance and got it 'good enough'. The longer distance you test over the more errors will show allowing you to correct easier.

[RomanLini]

Yeah that is clever!

I just cut plastic pieces 50mm, 100mm, 200mm, 300mm etc with toolpath moving only one way (to negate backlash) and measured their length with a good quality vernier caliper.

My NSK ballscrews are just rolled type, but they calibrated at almost perfectly 8mm/turn at 25'C in real life, which is what they were supposed to be.

[BanduraMaker]

Thanks for the idea guys. I really like the microscope idea. I have an old 24" "Draftsman" brand rule (like 50 years old+) That's marked in 100ths on one side. Hard to read due to the patina but it's still there. To be safe, I'd probably zero at 1" and not read all the way to the end as they're a little banked up.

I think that right away, I'll use my plotter pen to make a bunch of hash marks on a piece of paper every inch along the travel of the machine just to ball park it. and then move forward depending on how that comes out.

Any experience with these DRO's? [ame=http://www.amazon.com/iGaging-DigiMAG-Magnetic-Digital-Readout/dp/B003JULE4Y/ref=wl_it_dp_o?ie=UTF8&coliid=I3N7CWJ3Y31WKO&colid=2ZV0X7SPUJZV1]Amazon.com: Quint iGaging DigiMAG 24" Magnetic Remote Digital Readout: Home Improvement[/ame]

That's a pretty decent price for 24" - not much more than a USB microscope, but if it's not accurate, kind of wasted money!

[tracyranson]

I have been lurking on the forum for a bit and every so often I put my two cents in. I remember reading somewhere about some guy making an attachment that worked on the same principles of a Faro arm to check the accuracy of his home built CNC machine. I cannot remember where I read it, but that might be something to look into. :idea:

[BanduraMaker]

So I had a few minutes today so I made a quick program to cut some slots 5" apart (I have a 6" dial & digital caliper). Used a 1/2" bit to minimize flex but all I had around was a piece of wood.

With steps/ set to 5080, I read 4.9995 with the dial and 4.9990 or 4.9985 with the digital depending on how hard I squeezed them. I checked the calipers out against each other and they were dead nuts on*. I would definitely trust these calipers more than I'd trust the dial indicator.

Perhaps I'm seeing this cyclic error. Anyway, it seems that it's probably not worth worrying too much about. Later, I'll try a longer swing, probably 22" and see how it works. I actually had this on this program but the work piece came loose on the far end (bad clamping job) so I had to abort.

* interestingly, they're both mitutoyo, one probably from the 50s or 60s and the other from the 80's. I usually used the dial because the 80's digital one sucks the batteries down fast.

[BanduraMaker]

Pics aren't visible. Can you re-try?