Trying to improve the Makelangelo accuracy with v-groove bearings
Since I started the Makelangelo project I’ve had varying levels of precision, or repeatable accuracy. It is the end result of every other factor coming together just right – the software, the hardware, even the environment around the machine all play a part.
The original bobbins were friction-fit onto the shafts. Often the motor would turn and the bobbin would slip a little bit. Pictures had noticable “cracks” in them and the pen never finished a drawing where it started.
Then I 3D printed bobbins. They were better but they were still a bit off. The surfaces were uneven and the hole through the center was slightly larger than the motor shaft, which left room for mistakes.
Since Makelangelo v2 I’ve been using milled bobbins that are as close to perfection as modern machining can make. They are a gorgeous fit and exactly the right size. That’s a problem I consider solved.
Now that the bobbins are working correctly I’ve tried changing a few other things. Since the v2 I’ve been putting all the electronics in a nice box to keep things tidy and better looking. to the uninitiated, loose wires = scary. I hoped that less intimidating would mean more sales. Unfortunately, this has created two new problems.
Firstly, making parts: the box is made of laser cut parts in two materials. I could go on for a long time about manufacturing problems. Let’s just say they’re very real and they suck.
Secondly, the strings to the pen holder are much longer and they go around corners. When the bobbins turn the pen holder does not immediately react. For lack of a better word, I’m going to call this string give or just give for short.
The give appears to be the same in either direction when a bobbin turns – the left motor bobbin has almost 1/4 turn of give and the right has about 1/6th. I’ve only eyeballed it so far. I’m going to measure it step by step and confirm my observation, then modify the firmware to compensate.
The amount of give is not the same in each bobbin. I suspect the difference has to do with how much string is on each bobbin, as well as the length of the string. This means any time the strings are replaced – either from wear or breakage – the best practice will be to immediately recalibrate and check that the give has not changed. I wonder what kind of string has the least give?
Anyways, back to the picture at the top of the post: I wanted to eliminate friction as a source of give. I thought maybe a small amount of rubbing of the string on the bobbins was causing the give.
VHS member James saw my post for v-groove bearings on the mailing list and lent me a set to try instead of the eye bolts. In the picture above you can see I replaced them all. So far it looks like the bearings make no difference on friction. Worse, to use them would mean getting new screws, new washers, and increasing the total number of parts in every kit.
So now I ask you, Makelangelo users: Are you seeing similar results? Have you found a solution? Let me know in the forums where we can all join in the conversation. If we can find a solution that works I can move on to more interesting parts of the project – like finally making a pen holder for spray paint.