Optical Table CNC Lathe
When building CNC macines that use any meaningful cutting force at all cutter rigidity is of utmost importance. Usually that means building the machine of as big, heavy sections as possible. For a variety of reasons cast Iron is a favorite for machine tools, but as a hobbyist it is not easy or cheap to get custom castings made. Granite slabs also make nice bases for CNC machines, and I have used them in the past:
But I move more frequently nowadays and I need my machines to be more portable than that.
So that brings me back to the consideration of stiffness vs weight.
Lets use beams as an example, keeping in mind that the issues are perfectly generalizable.
Say we take a simple solid circular beam, then it's second moment of inertia makes it more resistant to bending (stiffer) than a hollow circular tube:
But it is also far heavier. In fact, the very central axis does, theoretically, not resist the bending at all. For any part of that beam, the further the material is away from the central axis the more it contributes to the stiffness of the beam.
In reality, of course, if the walls are too thin they will tend to buckle. However, the important concept here is that for a given amount of material volume, hollow structures can be made far stiffer than solid ones.
It occurred to me that optical tables are made to be very stiff. Vibration and bending can be a big problem in optical experiments, so they are of primary concern in the construction of these tables. One solution is the afore mentioned granite slab, but another solution is this:
And, much like the I beam, the center section is largely hollow with the exception of a lightweight reinforcing honeycomb structure.
That seems to be a very good candidate material for a machine base when both stiffness and weight are a concern.
So I got one of these tables, and after a brief usage to work on the Laser Interferometer, I re-purposed it to make a CNC lathe.
One of the nice things about this type of table is that it comes with many predrilled and tapped mounting holes. That makes attaching things very quick. One of the very bad things about this type of table is that it comes with many predrilled and tapped mounting holes. Non-used holes have to be sealed, or cutting chips will fall in. I used set screws to do that, but many of them felt very loose, so that vibration was likely to shake them until they fall down inside of the table. So I used a bit of threadlocker to keep them in place.
At the bottom, I added some casters. They fulfil a dual purpose. First, the obvious is mobility, but secondly they help with vibration isolation. I do live in a high rise at the moment. I used the softest rubber casters I could find in this size. The CNC mill is on dual isolators for that reason too:
Pneumatic casters would have been even better...
I used Parker stages for both the X and Y axes. This was a very easy way to do it, because they already have the mounting holes in all the right places to bolt together perpendicularly. Furthermore, they are precision machined and have precision screws. Also, the mount points lined up perfectly with the hole size and spacing of the optical table.
Sort of like this unit:
I got the spindle from eBay. This unit is made by Setco, and it most likely a secondary spindle from a CNC lathe. I put a dial indicator and capacitative gauge on the spindle and both agree that lateral runout if 5 microns or less. Since this number is a combination of the bearing runout and manufacturing error of the spindle face plate the rotational error as to less than or equal to that number. Nice! Also the mounting holes on the spindle were oversized enough that I could use the mounting holes on the table.
As a tool post I used a quick change tool post (QHTP) of size AXA from shars:
I had a Kollmorgen ServoDisc DC motor lying around, so that one drives the spindle.This is nice, because a DC motor can be driven using a bench power supply...
The general process to get a prototype up and running is pretty simple:
- mount x and and y linear actuators
- mount spindle
- cut mounting hole for QCTP into MIC6 aluminum tooling plate
- mount tooling plate to xy actuator stack
- mount QCTP to tooling plate
- mount motor to spindle using aluminum bracket and old leather belt.
- drill approximate holes for chuck backplate into aluminum plate using a hand drill and use that to mount chuck to spindle (the spindle can then be used to make a proper back plate....)
And this is what that looks like:
This worked great, and I was able to make a proper backplate with that setup. However, the stiffness was not what I had hoped for. Aluminum cut well, but cutting in steel was almost impossible. I could see the cutting edge move in response to the force necessary to push into the steel to cut.
It turns out that the culprit was one of the linear tables. Parker makes them in two versions, one uses crossed roller bearings and the other uses ball bearings. I had one of each. The gap between the top slide and the bottom of the linear tables is pretty small, so it is easy to put your finger across both parts at once. When cutting, vibrations were easily felt to stop across the ball bearing. Closer examination of the ball bearing showed it to be a design that could not handle any load. Most linear ball bearings cup the ball on both sides, providing petty good rigidity:
But some designs simply roll the balls over another circular surface, which is much weaker:
The Parker linear actuator has a very weak variation of the second design. (At least one of them does. The other one uses extremely stiff cylindrical roller bearings.)