Make an Easy Flexure for your Machine

I built a little 3-axis desktop mill about a year ago, which functions well, but I really needed to make some improvements to turn it into a workhorse machine. First off, I had built flexures into the frame to support one of the Y-axis rails. While the flexures functioned properly (not overconstraining the Y-axis table), they did not have much strength, and broke off when I disassembled the machine. 

So, to make some better flexures!

Step 1: What would I want a flexure for?

Each of the three carriages on my mill bear on two rails. The carriage can move in a total of 6 possible directions (6 degrees of freedom), but I only want the carriage to move in one direction. So I have to constrain it in exactly 5 directions. If I constrain it in more than 5, the carriage will likely jam if the rails are even slightly misaligned with respect to the other. The first rail is easy - it is fixed. This one rail constrains the carriage in 4 directions - translation in X and Z as well as rotation about X and Z. So I only need the second rail to constrain the carriage in rotation about Y without constraining it in any other direction. 

This is where the flexure helps - it is stiff in one direction (preventing rotation of the carriage about Y) but flexible in the orthogonal direction (allowing for misalignment between the two rails. 

Comments

[rimar2000]

Forgive my ignorance, but, would not it be better to ensure that the rails are perfectly parallel? That does not seem so difficult to achieve, it is easier than to add a flexible support.

[pfred2]

Anyone can build a CNC machine that operates perfectly when it is motionless. The trick is to make one that works well enough when you use it. Doing that is not nearly as easy as it may sound. Guaranteeing those rods are parallel while the machine is running is almost impossible to do. Just the heat generated by motion could bend the rods enough to bind the system up. If the machine misses a single step then everything is messed up from there on out and you have an unrecoverable error. You're better off with some looseness designed into the mechanism to avoid that happening. That is exactly what this does.

In order to fully understand this you have to know the difference between accuracy and precision. They are similar terms with slightly different outcomes. Here some precision is sacrificed for greater accuracy. In reality it is precision you'd never benefit from though if you lost accuracy, which you would if the machine bound up and you missed steps. What good is precision if you miss the target? I'd go as far as to say you'd never benefit from the precision even if you didn't lose accuracy.

Others might like to debate me on that and I'll concede that in some rare circumstances they'd be right. Homemade CNC isn't one of those circumstances though so I don't see how their argument applies.

[rimar2000]

Hello, Fred. Your explanations are a bit hard to understand with my "yes bwana" English.

This is my first encounter with the difference between accuracy and precision, I found a good article in Wikipedia about that.

I think the heat issue should not be a problem in this type of machinery, since the efforts and speeds are too small. And I think that to make a couple of parallel slidings is as simple as moving the sliding carriage from one end to the other BEFORE finally fixing the bars in place. Obviously, the fixing holes –only one of them– must have some tolerance to allow this. Once you have the rods in parallel, you have not the problem anymore.

Mind you, I am aware that I am speaking from the outside, without knowing intimately the problem. I've built some sliders, but never with the precision you need in a device like that of the instructable.

[danielhfourie]

Hi rimar
all questions welcome :)
My response is "just how parallel?"
The inside diameter of the bronze bushings that the carriage rides upon are slightly larger than the diameter of the rails (maybe by 0.001" for my machine). This sliding fit allows them to move. This means that you have to make the second rail parallel to the other within about 0.001" in order for the carriage not to jam. This turns out to be actually quite difficult to implement in practice, especially if you want to make the machine frame and alignment features with some 2D machining process with lax tolerances (I used a waterjet machine). Using a flexure frees you to not have to worry as much about the tolerances of your fabrication/assembly process. Also if you design them right, flexures are easier than you'd think to integrate into a design - check out the rail flexures on the MTM Snap

I suppose you could make some adjustment fixture that allows you to tweak the parallelness of the second rail, but that'd be finicky. I suppose you could also glue the rail in place when you knew the rails were within the parallelness you wanted, but then you can't go back and fix your machine without having to use a lot of destructive force.

Hi phred,
I don't think "looseness designed into the mechanism" is a good way to describe what is happening here. In fact, the carriage is exactly constrained by the two rails. It cannot move in any direction except along the rails (barring very high forces). Rather, the flexure is a way removing an extra (superfluous) constraint direction that would otherwise be imposed by the second rail.
I don't think accuracy vs. precision is quite at issue here.

[rimar2000]

Thanks for your explanation, Daniel. I envy the quality of your work.

[danielhfourie]

meant to be a link there...
MTM Snap http://www.flickr.com/photos/mellis/6262212548/

[pfred2]

Fixing the flexible guide rod rigidly is not going to increase precision but could lead to binding that would reduce accuracy. Open loop CNC systems once thrown off cannot compensate to recover their accuracy either. So you are just better off avoiding that happening.