Realistic repeatability expectations?

Hello All,

Trying to gauge what a realistic expectation would be for positional accuracy of my Shapeoko 3 (Standard configuration) should be? I bought my machine to use for making form tooling to thermoform Kydex holster shells as well as to trim the shells after forming, I roughed out a tool yesterday, and came back today to do a finishing pass and was surprised at how much of an offset there was today.

I used the same reference point for my zero and verified the values from the machine origin before running the program. Should cycling the machine on and off cause an offset of .020" in my Y axis and .010" in my X axis? I understand this is not a VMC, and wont have .0001" repeatability, I just want to know how much slop I need to allow myself in order to make sure tooling will be the proper size after machining. I’m hoping that the tool I finished this evening will still work for forming, 1.5" thick HDPE blocks add up quickly in cost. I’m assuming the inaccuracy would be caused by the homing cycle when the machine is cycled on and off, as I have run whole parts in the past and not had this occur.

Any insight is appreciated.

which home switches does your machine have?
Mine has the mechanical switches and is very accurate, even after power off and resets and this comes from a guy who made a living off of accuracy and precision.

1 Like

That is an unusually large delta, in my experience.

It is possible that your machine didn’t zero very effectively for one of the two startups.

Have you checked your belt tensions, ensured the belt pulley set screws are tight and checked V Wheel tensions?

When I need to run paths on the same workpiece after a power cycle I prefer to check the X, Y zero before starting the job as this does sometimes happen. I leave the X, Y zero corner of the stock as late as I can in the machining process to allow me to check or reset the zero point.

Alternatively, if this is a roughing followed by finishing toolpath then you could just leave about 0.025" stock to leave on the roughing toolpath to ensure you have enough to trim in the finishing toolpaths.


Make sure your machine is mechanically sound as @LiamN says. After that, one trick I learned is that when turning your machine on, pull it all the way forward and hold it against the front plates. You do this because when the steppers are first powered, they will turn to the closest full step and that may not be in the same direction on the Y axis motors. This helps significantly with repeatability when power cycling the machine.

As a general rule though, I don’t power cycle my machine between tools/files if I am trying for high accuracy (less than 5 thou).

1 Like

I have the inductive homing switches.

To the best of my knowledge the tensions are still all good, I’ve only had the machine for twoish weeks, and have ran it maybe a total of 6 hours, but I will double check these this evening and see if anything jumps out.

The part is two sided, and is doweled to the waste board when flipped using 4 3/8 dowel pins, the centerpoint of the lower left dowel is treated as the origin for the second side, I may look at running an indicator around the pin and seeing if its repeating to that position every time, I’m referencing the machine origin to establish its location, so if homing is my issue then this would show it. This is the first time I have had an issue of this sort, so I am inclined to believe it is related to the homing cycle.

I typically leave .015-.020" radial stock for finishing, my finishing tool is a 1/16" long reach ball mill so I try not to have more than 25-30% of its diameter left for cleanup. I could absolutely leave more radial stock, but where this will become problematic is when I go to use the trim fixture to trim my shells after forming, If I have to pickup the fixture for every part I run then its going to be vastly inefficient. I need the holes to be relatively repeatable location wise so that when the two shells are put together the internal geometry matches up.

As with any machine of this nature I guess I’ll need to do some tweaking and hope for the best.

1 Like

@nwallace I’ll give that a shot, I didn’t love stopping mid job, but sometimes I don’t have much of an option due to time constraints. I’m hoping I can get this dialed in a bit better though or this could cause a lot of headaches going forward.


On the new bitzero (which seems to be sold out)

There is a hole for proper probing and that means that Carbide Motion now understands how to do a proper probe in a hole.

If your jig or part allows you to have a suitable size hole in metal you can use a gauge pin in the collet to probe with;

Which might be a good way for you to hit zero after a power cycle.

1 Like

@LiamN I’ll have to do some more reading on the posts you linked this afternoon, I’d have to look at putting some sort of bushing in to probe if that’s the case since the tools are all made from HDPE. Appreciate the responses and suggestions.

1 Like

Turning the machine on and off will result in the machine losing position a bit - rehome it again and it should be able to get within less than .005".

1 Like

You could just leave the machine on until you come back to it.

Is there a difference in the precision (in terms of determining the pinpoint home location is the exact same everytime) of the mechanical switches and the inductive switches?

I have had both and have not noticed a difference. That’s not very scientific but I do cut almost exclusively aluminum and I make several parts that would show anything more than about 5-7 thou of variance in repeatability.



Are you saying the mechanical ones are more precise?

And if so, is there any compatibility with the mechanical ones and CM v5? I still have mine, but I just installed the inductive ones that came with the Z-plus (installed 2 weeks ago but haven’t been able to use the machine much since then)…

The homing precision is pretty important to me as I do a few pieces that are 2-sided, and I don’t always do side 1 and then 2 consecutively (as side 2 customizes the piece for the application, saves time to have a bunch of side 1 ready to go ahead of time)…

No, I am saying that I have not noticed a difference between either setup. Regardless, if you want to maintain zero point accuracy while also power cycling the machine I would do a couple things.

  • Re-find your zero every time you power cycle the machine. Since these are hobby grade machines, this is definitely the best way to make sure you get multiple operations to line up.
  • Pull and hold the X axis all the way forward against the front plates when you start up for the reasons I stated above. This will make sure that your X axis is perpendicular to your machine (as long as your machine is square).

The inductive homing switches should have (slightly) better precision), but the error is going to be:

+/- one-half microstep (1/40th of a mm) + +/- one-half the detection range/trigger variation of the switches

The mechanical switches are:

and there’s a datasheet which ought to have numbers of interest.

1 Like

Sorry, for some reason I missed the “not” in your first reply… That is good to hear!

And I appreciate the detailed steps for high repeatability as well!

Great to see this is quantifiable, thanks for the info!!

I can not answer that, as I do not have the inductive switches. But I have adapted other machinery to use them.
I can say this about the shapeoko, the machine can repeatedly return to the same spot at the complete opposite corner. I have taken advantage of this a few times.
such as…
One project I did not like the stark looking straight up and down appearance of the walls. To remove the top of the corners, I fitted a ball nose 1/16th bit and set the machine to carve the top corners off by 0.003" …it positioned and cut equally all around the project (the project was approx 30"x 12") with a LOT of cuts.
I can do a vcarve project, then send the cutter out of the way, sand the top lightly and then send the v-bit back in to clean things up. It will exactly remove all dust and remaining fragments without actually cutting into the wood further at any dimension.
I had to change out brushes in the router during the above (LOT of cuts). Reinstalled the router to the mount, checked the x-y zero and re-set z-zero and continued cutting without deviation.
could go on and on about it…
Mechanical switches will trip at the same precise location each time unless physically damaged.
Inductive can be affected by surrounding material, electrical line voltage or other means.
Personally, I do not know what the new switches are…usually (to me) inductive means a Hall Effect?
But some here have mentioned that they may be magnetic trip switches ?
If these are Hall Effect, when a water heater, or air conditioner or washing machine kicks on or microwave is started, it causes a line voltage drop in the house. That drop will change the generated field of a hall effect switch. It won’t affect a mechanical switch.
I am sure some 'lectrical wizard will come on and tell you of the industrial standards of using a certain switch, but industrial standard has NEVER been good enough for me.
Besides, the accuracy of the shapeoko is based on luck. It has no feedback to tell the controller the true position. And either switch should be good enough for this.
If you would like to read from a different source, here is a good write up on endstop sensing…

I might also add, If you are needing to pull your x-beam to the face plates before powering on, maybe you should go back to squaring up the machine…

edit… That article also sheds light on the disconnect issues of the shapeoko.


@nwallace I would have to leave the machine on indefinitely in that case, which I’m not sure is a great option due to the nature of how stepper motors work.