Shapeoko 3 – Some Tests and Some Results!

Lots of stuff today. Mostly testing and such.

First thing I did was setup the dial indicator to measure the Z flex. I set it up and measured in the X and Y axis with the Z fully retracted and then fully extended.

I wish I had measured it before I made the upgrades so I had a comparison, but oh well. Maybe someone reading this who has a stock S3 can duplicate my measurements.

I ran these tests with the machine on and the motors energized so they would hold position.

In any case, here is checking the in the Y direction at full Z extension.


Here I am checking in the X direction at full Z retraction.


I used my trigger pull gauge to get as close to 10 lbs of force as I could. The gauge only goes to 12 lbs, so I figured 10 lbs was a reasonable number.

In the Y direction at full Z retraction I saw a deflection of 0.025″. In the Y direction at full Z extension I saw a deflection of 0.040″. Which seems like a lot to me.

In the X direction I saw the same deflection of 0.015″ both at full Z retraction and extension.

So it seems to me that there is a lot of flex in the Z carriage still. I can think of several things that could be done to stiffen it further.

I also checked the table for flatness. It’s actually pretty good – within about 0.005″ most places. But I discovered that the front was higher than the back so I double checked the rail heights and discovered that the back right corner was low so I shimmed it up another 0.010″ which makes the table flat enough for my purposes so far.


Then to get some real-world numbers I ran the Diamond-Circle-Square test again and I’m very happy with the results!


The square should measure 3.500″ and it measures 3.502″ in the X and 3.504″ in the Y.

The diamond should measure 2.250″ and it measures 2.255″ across the flats in both directions.

The circle should also measure 2.250″ and it measures between 2.252″ and 2.254″ all around. So it’s even pretty darned round. 😀

I am about 99.9% satisfied with these numbers. I could spend a lot more time fiddling around to get a very minor improvement, but I’m just going to go with this for now.

Now on to some test cuts.

I was interested to see how fast I could generate some circles. I wrote a perl script shortly after I got my Shapeoko 1 that generates arrays of holes (something that I make quite often) but the S1 wasn’t rigid enough to be faster than my drill press and the holes never turned out very round.

I was curious to see what the S3 would do. I have to admit up front that I’m pretty happy with the results. But there were a few entertaining things that happened getting there.

On my S1 I had a small vise mounted at the end of the travel. It stayed there unless I needed to mill something a bit long. It is bolted to a piece of MDF and has holes on 6″ centers to attach it to the table. It also had two dowels to locate it so all I had to do was place it and bolt it down.

I’m not ready to drill dowel holes in my S3 table yet, so I cut the dowels off. Then I bolted the vise to the table and indicated it square.


I was getting ready to start making some holes when I noticed a problem. See if you can find it.


Yup, the vise handle interfered with the carriage plate. That’s not going to work. But luckily all I had to do was move it over to the next set of t-slots and square it up again. Then there was plenty of clearance for the handle and the bit could still reach the whole work area.

I wanted to try various feed rates to see if it made any difference in hole quality. I used a 1/4″ two flute router bit and made a program to generate a 1/2″ hole. The milling took two passes – one to rough and one to finish. I ran them at feed rates starting at 40 ipm and increasing by 5 ipm up to 70 ipm in hard maple.

After I had all the holes finished I measured them. They were all around 0.495″ in diameter with no appreciable difference in roundness or quality. The last hole at 70 ipm took 13 seconds to do both passes.

I was hoping to be able to mill 50 holes in under 12 minutes and 13 seconds per hole is right in that ball park. So I set up one of my loading block blanks, generated the g-code at 70 ipm and let her rip!


12 minutes and 2 seconds later I had 50 1/2″ holes milled in maple. Now we’re talking!


The next test will be to do the same work but use a 3/8″ diameter router bit so the holes get generated in one pass.

I have to say that I am pretty darned happy with the rigidity and accuracy of my upgraded S3 so far. The only thing I might work on still is the stiffness of the Z carriage plate.

But I’m sure I’ll come up with more to modify soon. And there is that SuperPID coming. And I need to make a control box one of these days…

%d bloggers like this: