Accuracy upgrade

So I see that the x carve can be accurate to 0.075mm. Is there any way to improve upon this? What part of the machine is stopping it from being more accurate? The arduino? Steppers? Would it just a be a simple upgrade to increase the resolution?

One user added reduction gearing between his motors and pulleys.

You’d also want to make your machine more rigid and reduce the deflection on the Makerslide by adding side supports for the Y -axis.

You’ll also want to calibrate for belt stretch, and some people, esp. w/ larger machines have used wider belts. Getting 0.9 degree (400 step) stepper motors will markedly improve positional accuracy, and 0.45 degree (800 step) are available. And/or you can replace the belts and threaded rod w/ Acme rod (or equivalent) on all axes.

1 Like

Very interesting. Thank you

FWIW, if one can avoid backlash and deflection caused by load, the machine is amazingly accurate — I drew up a ruler on my ShapeOko 1 (un upgraded save for 400 step motors) and when scanned and compared to a known good graphic arts ruler it was never off by more than a pixel or two, 1/600 or 1/300th of an inch.

What are the specs to look for when searching for a replacement NEMA 23 motor with a higher stepping? Other than the 1/4" shaft, what other parameters should I look for? ie. voltage, torque?


~125 oz. in. (9000 g cm) max holding torque
max. 1.7–2A would be a good current rating for use w/ a gshield

Apparently most NEMA23 motor shafts are turned down from a larger size (which is still present internally) — if you can get replacement pulleys which suit, I’d certainly not restrict the search by 6.35mm shaft size — and if getting larger motors, might as well upgrade to wider belts, esp. on any axis which has been upgraded in length.

I am looking around the forms and this is the closest thing i could find to what I am searching for. I am trying to find out what would be needed to change the machine from being belt driven to rod driven in order to improve accuracy.

Any insight?

One of the most straight-forward ways to convert from belt to screw drives is this arrangement.

SDP/SI stocks pulleys and belts:


How would one go about calibrating for belt stretch? Do you mean measuring the backlash caused by a stretching belt? For my machine this only happens when the direction changes. It seems that there would have to be an extra parameter in GRBL that would accomodate this. I am not sure how to include this in the steps/mm calibration and get a consistent result.

With GT2 belts, properly tensioned, backlash should be minimal.

Minimal ? What is minimal. 20 on a circle? Please be more specific.

I can actually see my belt stretch at the fixture point at each end of the X axis. And yes, the belt is secured properly. I guess this is what is causing my problem. The stretch distance just do happens to be exactly the amount of backlash that I measured. The belt does not stretch at all on the Y axis. I was shocked that I could see the belt stretch. If the belt isn’t stretching, then something else is making it appear that way.

You may have a defective belt. Contact customer service?