Centroid Community CNC Support Forum

I discovered a rather odd behaviour with my probe setup today. I have a KP-1 in a dedicated toolholder (T10) and have it running true and have done the touch probe calibration cycle, giving an effective probe dia of 5.0287mm which seems reasonable. I'd only previously probed ID's and surfaces, and it probes the ID calibration ring within a few tenths of a thou anywhere on the machine bed, so all seemed good. Today I wanted to probe a 55.60mm outside diameter to find centre for a BHC but when probed using the boss probe cycle it kept coming out at around 55.02mm consistently. To check I actually mounted the 25mm ring on top of the 55.6mm round and probed that so the centres were in the same place to ensure no stick-slip etc. During this time I didn't remove the KP-1 from the spindle and because of the spindle brake it did not rotate.

Report is attached and the probe cycle extract below. <--- are the notes I added into this post

2025-May-06 10:59:52
Center of Boss
-44.892640
-66.066327
55.030701 <---- this should be around 55.60mm
New Probe Cycle
Enter Description Here
2025-May-06 11:02:37
Center of Boss
-44.889536
-66.055770
55.018937 <---- this should be around 55.60mm
New Probe Cycle
Enter Description Here
2025-May-06 11:03:57
Center of Bore
-44.217264
-66.509100
25.027377 <------------- KP-1 calibration ring on same point of machine bed X-Y which is 25.01mm
New Probe Cycle
Enter Description Here
2025-May-06 11:05:28
Center of Boss
-44.889536
-66.069432
55.004039 <---- this should be around 55.60mm

Any idea why I can probe ID's accurately but the boss probing is way off?

I think your mm/rev are not ok:

1 X 2.4830000000000001 2000 0.0500 2 2 0 1 Y N
2 Y 2.4840000000000000 2000 0.0300 2 2 0 1 N N

This is a lot of deviation from 2.5mm, this also results in a calibrated probe diameter 5.0287mm

010 -129.2479 5.0287 H010 D010 OFF OFF 0 5mm ball probe

Usualy the calibrated tip diameter is smaller than the real size.

Uwe

I had calculated/measured the overall turns ratio and backlash using the method in Marty's youtube videos, so using a digital micron level dial indicator mounted on the machine bed and touching the spindle in X and then Y, and using the jog function to step 1mm at a time over 10mm total distance. Both axes are within 0.01mm over 10mm in both directions and regardless of if moving onto the DTI, zeroing, then travelling 10mm, or reversing motion whilst on it. I also checked that a couple of days ago and it was still okay.

Next time I'm in the shed I'll try over 200mm or a much longer distance and see what that yields. My stepper drives are not indicating position errors of commanded vs actual of any more than a step. At my 4000 steps per 5mm that is a max error of a micron so not significant. Anyway, thanks for the intel on expected ball probe dia values after calibration, I'd have thought it would show up as being larger, but thinking about it more sure it would be smaller than 5mm when using an ID to calibrate it.

What kind of deviation would be expected/normal for a 5mm pitch ballscrew on a 40 year old machine? 0.1%, 1%?

Out of interest are mm/rev values usually under or over the nominal value? The people from Centroid will see loads of reports and machine configs, I only see mine!

Cheap C7 screws are max +- 0,05mm / 300mm, so with 2:1 reduction on 5mm pitch

mm/rev must be 2.50mm +- 0.00042 mm

I bet setting straight to 2.5mm will give you a better result in probing ID and OD.

Uwe

suntravel wrote: ↑Tue May 06, 2025 7:49 am Cheap C7 screws are max +- 0,05mm / 300mm, so with 2:1 reduction on 5mm pitch

mm/rev must be 2.50mm +- 0.00042 mm

I bet setting straight to 2.5mm will give you a better result in probing ID and OD.

Uwe

Yup, but I bet it also knocks out the accuracy on Marty's 0-10mm test. I spent hours getting that within a few microns My hauser jig borer gets round that by having dynamic changes of the scales to account for leadscrew pitch error, no such luck with Bridgeport/Centroid.

10mm is too short for measuring pitch deviation, use at least 100mm, better more.

With Acorn Six you can also have full closed loop with scales.

Uwe

The "pitch deviation" on a 40yo machine won't be much different from a modern machine, certainly nothing like 0.1% or even 1% but it depends on the quality of the machine, not its age. As Uwe says, use the calculated value for steps/mm and take it from there. To test the accuracy, make 2 cuts as a long distance apart as you can accommodate in your machine (and measure) in the same direction, and measure what you get. Backlash and compliance (springiness) in your system will be significant factors that will be almost impossible to eliminate.

Ballscrew compensation is going to be impossible to get right unless you have a means of measuring that is much better than your machine. But even then, what are you doing that could require such extremes of accuracy? As for backlash compensation, this only works for light cuts. Where deeper cuts are taken, the load on the ballscrew can reverse which can actually double the backlash error, possibly with catastrophic results for the cutter. You are better to eliminate it as much as possible to begin with.

Muzzer wrote: ↑Tue May 06, 2025 8:41 am Ballscrew compensation is going to be impossible to get right unless you have a means of measuring that is much better than your machine. But even then, what are you doing that could require such extremes of accuracy? As for backlash compensation, this only works for light cuts. Where deeper cuts are taken, the load on the ballscrew can reverse which can actually double the backlash error, possibly with catastrophic results for the cutter. You are better to eliminate it as much as possible to begin with.

Gearbox end covers/plates that locate the bearings for the layshafts and mainshafts, a thou or two difference is a problem in the tolerance stack. Looks like the jig borer is still the daddy for getting those right. SIP/Hauser make pretty good compensated ballscrews

So I went back to basics, zeroed the backlash and 2.5 mm/rev base setting. I then set up the X, Y, Z and W with my two longest slips, so 175mm, and used a finger DTI at Marty's recommended 300mm/min feed:



The result, more sensible values. Note this machine has metric ballscrews on X, Y, Z and an imperial one on W:



After that I redid backlash and that seemed to be consistent, then I recalibrated the KP-1 probe and it came out at 4.9056, is that around what other people see?

probe styli pre travel calibration needs to be done with the proper overall turns ratio, that is a requirement!