CNC machine update #5: exploring mounting and bearing options, thinking about backlash and progress on machining and funding

Posted on March 7, 2012 in CNC

Over the past couple weeks, I have been checking out options for improving various aspects of my Y axis design with the goal of reducing permanent bonds while keeping costs low. In particular, I have been thinking a lot about the interconnections between the precision rods, drive system and their respective wood panels.

Mounting the precision rods

Match drilling and press-fitting = no extra cost
One of the areas that I have been a bit wary about from the beginning is the way that the precision rods are mounted to the wood panels. In the original Mantis 9.1 project, a CNC router is used to cut holes that are slightly smaller than the diameter of the precision rods, so that the rods can be press-fit into them snugly. This removes the need for any sort of mounting hardware or epoxy, but a part of me feels that there is some amount of risk involved here. With use, the holes in the wood panels may widen a tiny amount, causing a small amount of misalignment between the two precision rods, which could inhibit or alter movement along the axis.

Another technique that the Mantis 9.1 project uses is called match drilling, wherein holes are simulatenously drilled through two pieces of material at once, resulting in holes that are nearly perfectly aligned.

Self-aligning mounted bearings with collars = ~$36.00
One way to work around the risk of working with wood as a support material for the precision rods is to use self-aligning bearings with shaft collars. A self-aligning bearing uses two rows of balls inside of it, allowing the inner ring to rotate along the axis in addition to rotating about it. This will not only alleviate the problem of material wearing down with use, but will also compensate for up to 5 degrees of misalignment of the precision rods, which could be immensely helpful for me.

To prevent the precision rods from sliding through these bearings and falling out, cheap shaft collars can be ordered to stop the precision rods from moving too much. Although the cost of this approach does add another $36.00 to the machine, I have a feeling it’ll be well worth it for me.

  • McMaster-Carr part (bearings): 1434K25
  • McMaster-Carr part (collars): 9414T8

Mounting the bed to the precision rods

Bronze sintered bushings = ~$3.54 ($0.59 ea.)
The original Mantis 9.1 project used four bronze sintered bushings (two per rod) which were epoxied to the bed to allow it to travel along the axis. However, I am very hesitant to do the same for my machine for two reasons. Firstly, the bond is permanent, so if I mess up or do not get the same results, I have to basically cut up my bed and disassemble my axis to try again. Secondly, since my machine will be somewhat larger than the original Mantis 9.1, I can’t be sure that the bushings will work as well. However, the bushings are extremely cheap, and therefore attractive.

  • McMaster-Carr part: 2868T72

Mounted bushings = $20.20 ($5.05 ea.)
Another way to attach the bed to the precision rods would be to use mounted bushings, which can be screwed into the bed non-permanently. To me, the benefit of being able to remove both the bed and bushings more easily would be worth the extra cost. Coupled with the use of bearings on the end of each precision rod, these could be removed and replaced very easily without damaging the machine, so I will likely spring for these.

  • McMaster-Carr part: 3813T1

Reducing backlash in the drive system

In earlier posts, I’ve mentioned that I had decided to use leadscrews and leadscrew nuts for my drive system for this machine. This is just about as low as you can go in terms of cost and still have a usable CNC machine, and I am well aware that there are many, many other options. Even if we just narrow our focus to look at drive systems that are similar to leadscrews, there are still many things to consider. One could use leadscrews or ballscrews, with a variety of thread types for each (square, ACME or buttress, for example). Each have their own advantages and disadvantages, but the two most important factors for me are cost and modularity. While I am willing to sacrifice cost for modularity, the reverse is less true. However, if the cost of modularity is so much higher that it’s prohibitive, I will be more interested in hacks.

One of the most important reasons that there are so many choices of drive systems and mechanics is the common problem of backlash. Basically, when you thread a nut onto a bolt, there will be some amount of space between the threads of each part. When you reverse the direction that you are spinning the nut, you will have to turn a small amount before any motion occurs, as the threads shift to make contact. In fact, as the threads wear down with use, this backlash increases, so a good solution here (or an easily serviceable one) is ideal. There is a great deal more you can learn about this topic, and I personally find it all very interesting. Use the following resources to learn more:

ACME leadscrews and hex nuts = ~$15
The original Mantis 9.1 project used a single hex nut and leadscrew per axis to provide the translational motion, but I didn’t see much information about backlash reduction. Two things made me somewhat uneasy about that project: the leadscrew nuts were home-made and they were epoxied right onto the moving bed. They were only able to produce their home-made leadscrew nuts because they had a tap of the correct type to do so, an $80+ tool that is conveniently not included in their BOM. Furthermore, since my Y axis platform will be considerably larger than the Mantis’, I couldn’t be sure that a single leadscrew nut would work well, so the idea of epoxying it on did not seem smart.

Anti-backlash nuts = $18 – ~$100 or more
One way to reduce backlash in a leadscrew drive system is to use the aptly-named anti-backlash nut. There are a variety of ways these are made, but in general, they use springs to apply dynamic axial or radial force to the nut assembly so that it actively ‘hugs’ the leadscrew. One common way I’ve learned about so far is to use two leadscrew nuts separated by a spring, such that the nuts are constantly being pushed away from each other, provided tension against the threads of the leadscrew. The other involves cutting the nut into slivers, then binding the slivers in a sleeve wrapped with a spring. The spring squeezes the slivers so that they are constantly ‘hugging’ the threads.

Backlash compensation in CAM software = free
Another option to reduce backlash is to compensate for it in software, which has the benefit of being an option regardless of hardware used. However, I feel there is a delicate balance to be had between hardware cost, minimum acceptable backlash and software compensation. If there is too much backlash, no software in the world will help. But if there is a reasonably small amount of backlash, we can account for it for free. Realistically, I expect to experiment with backlash compensation in my CAM software regardless of the drive system I choose. Right now, I’m looking solely at open-source CAM solutions, including:

Price comparisons

Each of these options carries a price – not a metaphorical one, but a real one. I’ve learned that by exploring these various options, the overall material cost can fluctuate quite a bit. In fact, the differences are so great that I’ve decided to maintain two separate BOMs at this point; one for a “basic” version of the axis, and one “deluxe” version. To save you the gory details, here is how those BOMs compare:

Basic version = $71.10
The “basic” version will use all the hacks I know, following the Mantis 9.1 project very closely. Match-drilled holes, press-fit precision rods and only a basic leadscrew hex nut, all aimed at keeping costs low. The trade-off, of course, is that labor, maintenance, tweaking and risk all increase.

Deluxe version = $142.36
The “deluxe” version includes more expensive parts, but reduces the risk of doing a bunch of work and having the machine not work. The parts would be more forgiving of small misalignments and machining errors, and would also result in less backlash along the drive system. However, the cost is double that of the basic version. So the question is, is it worth it?

Machining and funding progress

Beyond my thoughts on the design of this machine, two practical concerns have been on my mind the last couple weeks. Firstly, how I am going to actually make this machine. Secondly, how I am going to pay for this project. If I had access to a CNC router, I could easily use it to produce my parts (complete with features) based on my existing CAD drawings. However, I live in a small enough town where I may not have easy access to one. I hear that there is at least one shop in town that may have the machinery I need, but as of right now, I’m not sure of the costs involved. A relative of one of my advisers is a professional tool and die maker in a nearby town and may be interested in helping out, but I don’t want to jump to any conclusions until I’ve met the person. Otherwise, a Physics professor has agreed to sponsor me and teach me about traditional machining techniques, if I can convince our Industrial Technology department to let me use their machinery. Both routes are being explored now, so I should be able to make something happen soon!

As far as funding goes, I plan on taking out some student loans to pay for the materials. If I think about my work as an academic pursuit, as well as a personal one, I can use student loans to pay for materials. Of course this means I will go further into debt, but not too badly. Considering that my tuition has been paid as part of my graduate assistantship this last year, the loans won’t be nearly as bad as they could be.

Next steps

Over the next couple of weeks, I will be working on a couple of things at the same time. Pending approval from the ITEC department, I hope to get access to some machinery and start learning about how to use it to make my materials. By the time I’m ready to machine some features, I should be able to decide upon and actually purchase the materials from my bill(s) of materials.

As soon as I’ve decided on which BOM I want to use, I can begin the tentative plans for the X and Z axes, which will hopefully go much faster than the Y axis design phase. More to come!