All items created are .015" too small

They do want you to revert. It’s been advised in the forums to everyone that updated and got it shipped to them. There are instructions on how to do it. They also are shipping 1.09 again. There’s actually a really good chance that your problem is being caused by your firmware. As soon as I connected to the web I got the prompt to upgrade so I did. worked really well until it didn’t. If you do revert, just get a new card, don’t reflash the old card, some people have had trouble with that.

You won’t like some aspects of the UI at first, there were nice improvements in 1.1.1, but they really are no big deal. AND BONUS, you’ll get 1 inch squares when you program 1 inch squares.

I’m curious did you revert to 1.09? did it solve the problem?

No I’m going to wait for 1.2

Please tell me someone found the solution to this. I thought I was crazy, both my x and y are exactly that off on pockets. I’m using Cut2d and have tried in the latest and 1.1.

I’m afraid not.

At this point I have:

  • Calibrated the X & Y travel several times
  • Went from 1.09 to 1.21 software
  • Purchased new bits from Amana and verified the dimensions using a Mitutoyo caliper
  • Made multiple “tweaks” in VCarve Pro, all to no avail.

If I say I want an 8" square, I will get 8" on one side and 7.985" or so on the other. One cut is the X axis, the next cut is the Y axis, or any combination of these two.

I have tried inside cuts, outside cuts, on the line, doesn’t matter the results are the same. The .015" is random since sometimes it is a little less and sometimes it is a little more.

During calibrations the bit is dead on and I have used my caliper to verify this with an engraving bit to better measure. To me this rules out “slop” in the worm drive or steppers or mounting of the spindle or spindle runout. (There is virtually no runout - undetectable). The calibrations are done “in the air” using manual commands.

Needless to say it really made my inlay cutting board take a very long time since I had to dial in every piece and test each one it first. When I did cutouts for sockets in Birch I had to dial in almost every size in order to make it actually work. Not a single hole matched what either of 2 different calipers said. They were all a tad too small.

I’m sorry. I wish I had the answer. I still use the CNC for everything that does not require such precision, but I’m not in a hurry again to do anything that requires that level of detail. And I check the Forum every day to see if something was discovered.

Hey Bill,

I would think of two possible causes: Deflection under load and possibly also backlash. Did you measure ball screw backlash?

When you calibrate, you use the machine with no load. Under load, the Onefinity CNC machine has the Z slider as its weak spot (see here, here, and here). That’s why I would think of deflection under load first.

What does you machine show regarding deflection when you make this test? I assume you don’t get the diameter of a red blood cell as deflection like Marco Reps gets on his DIY machine :frowning: However to perform this test, to get a meaningful result, you would better attach the force gauge to the end of the milling bit, as it’s there where the deflection finally does it bad influence :slight_smile:

There is no deflection under load that I can measure. Plus the load equals 1/8" deep (max) with a 1/4" bit so if there was deflection in that, that would be really sad… :frowning:

I looked before at the calibration links and I did calibrate without any load. Several times

I will look into the last test you recommend

Hey Bill,

how do you measure deflection under load? What I mean is ideally the overall deflection at the end of the bit in relation to the workpiece position, which would mean 1. bit deflection, 2. spindle shaft deflection (presumably to neglect), 3. Z slider deflection (presumably of big importance), 4. the Z assembly deflection (due to X carriage linear bearings play), and 5. the X gantry deflection (due to Y carriages linear bearings play) and 6. Y machine’s feet deflection on the table (assuming that the workpiece is attached at an absolute and undeflectable position to the machine bed, otherwise this would be point 7).

Or, which would be easier, at least the overall deflection of the spindle case in relation to the workpiece position (remove the two first points in the above list then).

I am not yet decided on how to attempt to measure this under realistic work conditions. At the moment, I think of measuring at least a small but relevant part of it, i.e. the deflection of Z slider and Z assembly on X carriage to the X gantry (or Y carriages), by attaching a laser to the spindle pointing upwards and an image sensor running on a separate axis parallel to X axis that catches the laser beam and would record the deflection of the spindle.

Another point: Did you take into account bit deflection. Bits get dull and the more they get dull, the more resistance the material offers them, and the more they can bend. In addition, the hotter they become, which contributes to the bending.

Do you use the shortest bit possible?

Did you click on the links mentioned above regarding Z slider?

What about ball screw backlash? You did not mention it.

PS: By the way, did you read the thread Hole Diameter is inconsistent in my Wasteboard?

Preface, I’m with PreciseBits so while I try to only post general information take everything I say with the understanding that I have a bias.

While I can’t give you a method to separate it to components, there is a simple test for total deflection under load. Cut 2 lines in each axis with a gap between them. Start the cuts from opposite directions going toward the gap between them. Repeat this in the other axis. This will change the climb and conventional sides of the slots. It doesn’t really matter how long they are or the gap size other than you want it long enough to get to speed. Here’s a simple image to illustrate the cuts.

You can then measure the difference between the middle of the lines. Half of that is your total deflection in that axis. Avoid using the the ends as you want to avoid the acceleration/deceleration sections.

This should at least tell you if you have an issue under load in an axis. If it is a deflection issue then it might not be the same number you are getting on your cuts as it will be dependent on the cutting forces (material, cutting depth, chipload, stepover). There is a version I have of this that will get you closer for non-slotting cuts but it’s a lot more complicated.

The other limitation of this is that it will only give you the deflection at the top of the cut. You could through cut your material to get the other side if you wanted. Again though, it should be enough to give you a decent idea of your deflection or if you have an issue in the X or Y.

Let me know if there’s something I can expand on or help with.

1 Like

Apparently I can only do one image at a time as a new user so here the image for the measurement.

1 Like

Hey John,

thank you! I think this can be very useful.

Welcome to the forum!

[OT] When you have Trust Level “New” (0) then at the moment you cannot

  • Send personal messages to other users
  • “Reply as new topic” via Link button (UI removed)
  • Flag posts
  • Post more than 1 image
  • Post any attachments
  • Post more than 2 hyperlinks in a post
  • Have actual links in the ‘about me’ field of your profile (will be silently and temporarily converted to plain text)
  • Mention more than 2 users in a post
  • Post more than 3 topics
  • Post more than 10 replies
  • Edit your own posts after more than 24 hours

But you can reach Trust Level “Basic” very easily by:

  • Entering at least 5 topics
  • Reading at least 30 posts
  • Spend a total of 10 minutes reading posts

If you reached “Basic” " this way, you can:

  • Use all core Discourse functions; all new user restrictions are removed
  • Send private messages
  • Upload images and attachments
  • Edit wiki posts
  • Flag posts
  • Mute other users

No problem.


Not my first forum so not unexpected, just wanted to state why it was in a separate post. I’ll be around if there’s something I can actually provide info for without crossing any lines (I try to never get close to marketing or product comparisons). So I’ll probably get basic at some point.

Let me know if there’s something I can help with.

I hate hearing that. Mine appears to be about .025 off over the 1" square on both axis. My dad was a tool and die guy, so I have every possible calibration tool at my disposal and have checked every suggestion and all check out. My z does have a little deflection, but the test pocket was .01 deep. When cutting the outside it is perfect, so deflection isn’t the issue in this case. As I type this though, maybe this afternoon I will try the pocket, then add a step to cut the inside line and see if it achieves the right dimension. If so, the pocket is definitely the culprit.

I write code for a living and if I had time I would pour over that pocket code and see if anything stood out, but I simply don’t have time.


You are correct and I did not measure, nor have an accurate way to measure deflection. The gauge in the video you linked to costs around $500 so that’s not going to happen. I admit that I was “assuming” that if I used a 1/4" Amana bit that protrudes 1" below the spindle and I cut .2" deep in Birch that there wouldn’t be any deflection since this is very little effort for any machine. However, I want to understand better John Torrez suggested to see if that is true

Morning John!

Thanks for the suggestions and I want to try it out. But you said to measure between the two lines yet in your 2nd drawing you show the lines as offset. Can you clarify a little more for this old guy so I can give it a shot? I have attached a drawing to see if I’m close. And I assume a caliper between the cuts

I believe he means the difference between the centre lines of each slot. With no deflection the centre lines would line up perfectly when drawn past the slot end boundaries.

If there is deflection, the centre lines when extended would be parallel but not line up - the spacing between these two parallel slot centre lines would indicate deflection.

My machine always cut nearly perfect (no longer have it). But one thing I would check would be the nuts at the ends of the ball screws. Make sure they are drawn up snug, but not too tight as to create undue axial load on the bearings. But enough to ensure there’s no axial slop.

Hi I’m pretty new to all this stuff but would be possible to check for deflection using a digital angle gauge? Bolt a steel plate on the front of the z slider and run your tool path and watch the gauge, is it possible the deflection is in the rails themselves. Like I said I’m new but its just a thought. good luck

Hey Bill @Machinist, hey Bill @bwcherry,

the manufacturer has this description:

But wouldn’t play here have been noticed in the test on ball screw backlash?

By the way @bwcherry, while you’re at it, have you checked that the ball screw coupler is not loose?

I did a sort of deflection test by milling an outside profile in 3/4" hardwood, milled the outside climb cutting, one 3" square was done with aggressive cuts and one done with less aggressive cuts and a final .010 pass, (spring pass kind of) no difference both measure exactly .005 big in both directions.
For what it is worth.