Ferrules and New Belts for the ShapeOko

Yes, more mods and accuracy calibration for my ShapeOko mill.

There was a discussion on the forum about how it was a bad idea to tin the ends of wires when they were going to be clamped in screw terminals and that you should use ferrules instead.

Since I didn’t own a ferrule crimper, and I always enjoy a chance to buy a new tool, I started looking around. Man, crimpers are expensive! There was a crimper posted in the forum thread for not a lot of money so I ordered one. It works great.

I also discovered that there is someone on Amazon selling the ferrules in packages of 100 for about $3 each, so I ordered several sizes.

So far I have done the motor wires where they attach to the terminal blocks, and the power leads to the Grbl shield. I have not done the motor wires where they attach to the Grbl shield yet because I will be re-working that wiring when I build my electronics enclosure. Soon I hope.

Anyway, here is what that looks like.

ferrules1Doesn’t that look much more professional? 🙂

The next mod was replacing the MXL style belts and pulleys with GT2 style. The GT2 belting is supposed to have less backlash. I also bought some extra length so I could change the way it wraps around the pulleys.

The belts now come in high, wrap around the geared pulley, then around one idler pulley and go out low. This means I get more than 180 degrees of belt contact around the geared pulley. I don’t think I’ll be skipping any teeth now.


Since the GT2 belting is a different pitch than the MXL, and the new pulleys had 16 teeth instead of 17, I had to re-do all the calculations and calibrations for the stepper motor driver values.

The calculator said to use 50.00 steps per mm for the value, so I started with that. For a crude calibration test I told my X axis to go 20 inches. Measured with a tape measure, it was about .125″ off, so I modified the value to suit and got it a lot closer. I repeated this with the Y axis.

But I wanted to get a lot more accurate than that. I wanted to be able to accurately measure a long travel distance on the X axis, so I could get a more accurate calibration.

I talked to my friend Paul, who owns Tried and True Tools (a used and consignment tool store) here in Minneapolis and he suggested bolting 1-2-3 blocks to the table a known distance apart. But how to measure that distance? Turns out he had a large internal micrometer that he would lend me, and that did the trick quite nicely.

But first I wanted to make sure that the X and Y axes were square to each other. I realized that since there are two motors and two belts, it would be possible that the ends of the Y axis could be offset if the belts were not aligned.

So I clamped a couple of parallels to the Y axis, and put my large engineers square on the X axis, and sure enough, it was off about .015″ over about 5″. So I adjusted the front belt to move it in relationship to the rear one and got that squared up.

square2I ran the carriage down to the other end and back, and checked it again. It stayed square. So I ran it down to the other end again and checked it down there – still square. So I’m pretty happy with that. I’d like to figure out a good way to measure the belt tension of the two belts. Plucking them like guitar strings to compare them doesn’t seem very accurate.

Anyway, back to calibrating the stepper configs. This is the setup for the X axis. I bolted down a 1-2-3 block and then clamped another one to the table, using the inside micrometer to set the distance to 16″.


I attached my dial test indicator to the motor mount (I have a piece of 1/4″ rod there just for this purpose) and set it to zero on the first block.

x-axis2Then I lifted up, moved over 10″, dropped down and moved 8″ more. The indicator read .0115″ over travel.


I repeated the measurement a couple of times to be sure, and got the same number each time. So I calculated the ratio between distance entered (18.000″) and distance traveled (18.011″) and multiplied the existing steps/mm value to come up with the new value to use.

I then repeated my measurement a few more times and verified that I was within .001″. Good enough!

Then I setup the Y axis the same way, but could only travel 6″.

y-axis1I got the Y all calibrated and then I figured I’d better check the Z axis. I setup a couple of 1-2-3 blocks up so I could travel 4″ (that’s about all I have for Z travel) and discovered that when I told it to go 4″ it actually went 4.047″. Wow, that’s a bit off. So I fixed that too.

z-axisThen I ran another diamond-circle-square test. It turned out the best yet. The size was spot on, and the circle was the closest to round I’ve ever had it.

I plan on re-building the Z axis so that it’s sturdier and straighter. It’s close to perpendicular to the table now, but it’s easy to move out of alignment. It’s also not as rigid as I’d like it to be.

But in any case, the ShapeOko is now pretty darned accurate, and I’m happy. The next thing to work on will be replacing the limit switches with mechanical ones – the opticals keep tripping due to dust getting in them.

And a torsion table sometime soon. Just leaning on the table can cause the test indicator to move .002″-.005″

%d bloggers like this: