Hello Guest it is January 24, 2022, 09:58:00 PM

Author Topic: Perfect circle--I wish  (Read 7387 times)

0 Members and 1 Guest are viewing this topic.

Perfect circle--I wish
« on: July 10, 2008, 02:57:25 PM »
My machine is not cutting perfect circles. I don't think it is backlash. The steps per inch check out prettyclose, I have not use a mike to check  the distance, checking a inch of movement on X & Y with a ruler, it's on.  What else do I check?????????????

Re: Perfect circle--I wish
« Reply #1 on: July 10, 2008, 05:00:00 PM »
You need to fully eliminate the possibility of either ball screw backlash or mechanical slop first.

Set up a dial indicator in (or on) the spindle and indicate against a 1-2-3 block in your vice or a similar setup.

For the X axis, write a little g code program to approach the same position from 0.5" inch away from both sides, ie first approach from the left, then from the right.

Zero the indicator after the first approach, and step thru the program. Watch the difference in the indicator when approaching from either side.

This will approximate the ball screw backlash. Now this time do it again, but after it approaches from the left, push hard on the table from the right (the same direction the cutting tool will be pushing back against the table). As it approaches from the right, push hard on the table from the left.

Now the difference in the indicator reading will show the total amount of slop, including screw backlash, thrust bearing slop and any slop in your bearing yoke.

Do this same test also for the y axis. On a reasonably tight machine you should see less than 0.001" to 0.002" total slop on each axis.

If you have more than that you need to isolate where its coming from. Typically (and in order of most likely) its coming from worn or not properly preloaded thrust bearings, worn ball screws or a loose yoke.

To test the thrust bearings, put your dial indicator base on the end of the table and set up the dial indicator on the end of the ball screw. If the end is not flat, you can grab a small bearing ball, put a dab of grease on it and stick in in the center hole in the end of the screw an indicate on the ball.

Now step through the program above and watch the indictor at the direction reversals, and do the pressure on the table as described above. The total run out  you observe will be the thrust bearing slop. It should be less than 0.001". If its more either the bearings are worn or the preload is wrong.

Good luck-

Paul T.
« Last Edit: July 10, 2008, 05:04:57 PM by titchener »

Offline RICH

  • *
  •  7,422 7,422
    • View Profile
Re: Perfect circle--I wish
« Reply #2 on: July 11, 2008, 10:34:17 PM »
A CNC mill can do some very accurate work but they are not as accurate as a lathe. A CNC mill with good components can approach a lathe. Now if you have backlash, you will cut an egged shaped / ovaled circle. If you use backlash compensation you can minimize the oval within the tolerances of your machines capability. With experience and knowing your machine you can get a bearing to just drop into or press into a hole done on a CNC mill even on an old converted mill. I say the above so you have a practical perspective of a perfect hole or circle. That said, you need to get some of the settings right as already noted.

Prettyclose just doesn't cut it, but then, a lot of the things made in my shop certainly fall into that category.

Offline jimpinder

  •  1,232 1,232
  • Wakefield, West Yorks, UK
    • View Profile
Re: Perfect circle--I wish
« Reply #3 on: July 12, 2008, 03:32:57 AM »
I am on my soapbox again - The steps per inch checkout pretty close. What does this mean

Either they are correct - to the number you have calculated , or they are wrong. Please yourself. Your steps per inch will be a simple round number e.g. mine is 60,000 - not 60023 or 59395. Claculate your steps per unit and put them in.

Motor - probably 200 steps, microsteps - depending on drive - 4,8,10 or 16. gearing down motor to leadscrew 1 (straight drive), 2,3, etc and finally leadscrew pitch - mine is 10 to 1 inch - multiply together and get a round number. Do not measure, jog up and down or anything else. The calculated number is your steps per unit and cannot be altered.

If you are to do "perfect" circles then you will need backlash compensation engaged. You may have none - lucky you - but there is some there, otherwise your machine would seize up. In describing a circle, all axis change direction once and probably twice - depending where on the circle you start, and if you have not compensated for backlash, however small, you will not get a "perfect" circle.

Check your machine over by making sure there is no obvious slop and then simply zero the DRO and type in the MDI line  GOX1 to get rid of backlash.Make sure backlash compensation is not on Set up a DTI or a pair of digital calipers - I use digital calipers - take a reading and zero them. In the MDI line type G0X2 then G0X1. The axis should move an inch and move back. Measure with the digital caliper again - this should be zero, but it will not be - and the figure shown is the backlash - that amount of the inch the axis did not move, whilst the gears and everything were swapping round to push the other way.
You can repeat it if you wish and take an average, but I normally put the backlash in the computer and turn it on and repeat what I have done, and see if I get to zero.

Now the backlash figure is one you can mess about with until you get it right. My lathe is cheap, and the backlash is quite large (I shall fit ball screws when the pocket and inclination decide) but Mach 3 deals with it very well and I get acceptable results. I must admit I have not done what Rich says and tried to machine a hole into which I press a bearing - I might try that with the milling head.

Perfection - Ah Well !!!

Not me driving the engine - I'm better looking.

Offline bowber

  •  216 216
  • Kirkby Stephen,Cumbria, UK
    • View Profile
Re: Perfect circle--I wish
« Reply #4 on: July 12, 2008, 08:37:41 AM »
Got to agree with jim, calculate the steps per unit, don't measure it, if you measure the movement over 1" and your machine has 12" travel then any error will be 12X greater, just 1 thou could end up as 12 thou error. I suppose that if you have the equipment to measure the distance at the full machine travel then it's worth doing but I'd still use the calculated steps needed.

I'm lucky in that my machine was made in the early 80's as a medium size training CNC mill and has good quality oversize screws etc so my back lash is very small but there is a tiny amount, just noticable on a DTI.


Offline RICH

  • *
  •  7,422 7,422
    • View Profile
Re: Perfect circle--I wish
« Reply #5 on: July 12, 2008, 09:39:26 AM »
Cutting a circle is a good test of what the machine can do.  Play around by milling a circle a number of different ways to see what you get. This is done done after MACH is set up. Having a flavor of what your machine will do saves you from not wasting time on something that may never happen. Use a piece of round stock as this is easier to measure. if you can't achieve some tolerance on the outside you probably won't get it on an inside cut.
Start from the center of the stock and make a single circular light cut. Measure diameter in 90 degree increments  noting the plus or minus away from what was wanted.
Now try milling a circle using same cut, multiple cuts or finish cuts, start the actual cut from different quadrants, etc. and see what you get out of them all.
You can mimic rotating the piece, offset the center starting point, change the axis backlash deliberately wrong or the steps wrong, etc.. 
Long story short, your just trying to set up and compensate for what your machine couldn't do in the first place such that a circle comes witin some tolerance.  Please note, that what I am posting, is after MACH is set up, and again are
just machining "tricks" which you can fool around with later on.