Not-so-quick Update

I've not posted in a bit, but I've been steadily working on the mill.

The stepper motors and the components for the drivers have arrived. I was worried that they wouldn't have enough torque to drive the stages but after test energising one phase my fears were allayed. I had to put a lot of effort into trying to turn the shaft with my fingers, whereas the threaded rod on the stages turns easily - no contest, at least in the static case. I've got a driver half wired up on breadboard, I need to finish that soon.

One motor is now mounted on the X axis. Another trip to the hardware store and I came out with short lengths of vinyl tubing, about 5mm ID, that are extremely tight on the threaded rod. Unfortunately it wasn't so tight on the motor shaft - negative air pressure did make it difficult to remove, and it seemed to turn fine, but I fully expect it would have slipped under load. A piece of heatshrink placed on the shaft first fixed all of that and made for a nice tight join; although I expect I'll add some glue to it before I start using it. The Y axis is going to be tricky; I need to somehow make a trench in the X stage for it's motor to sit in... :S



The Z axis is almost finished; all that's left is to mount the motors and threaded rod. Getting it together was tricky; my supply of aluminium U channel is almost completely used up (I don't want to buy another $10 length if I don't need to) and I mis-cut the three plates used for the Z bearings - by which I mean I cut them all to the same length. That meant the pivot bearing plate was too long, and the other plates were too short. D'oh! "Fun" was had extending the holes towards the end of the plates to gain barely enough clearance for the assembly to work.


During my work on the mill I've been evaluating what I would be using for the milling drill. Originally it was going to be the dremel, but it is way too loud for my liking. I feel uncomfortable even using it for a couple of minutes in the evening because it must be disturbing the neighbours - a prolonged period milling a piece would be far worse.

Instead I'm investigating a dodgy rotary tool I picked up from Bunnings for $10; it's insides are at the top of the above photo. It's supposed to be cordless, but the batteries are bad and don't last long, and the torque generated is pitiful. It is definitely quiet however, and the driveshaft has ridiculously low runout and end-play! Unfortunately I'm stuck with a bad collet, because the tool isn't *quite* dremel compatible. :(

The motor normally runs on either 3.7v from batteries or 6v from the plugpack; and those are definitely underdriving the motor because I can stall it without appreciably overheating it. I hope that with a PIC closed-loop PWM speed controller and a 12V or greater supply with a good current rating it could be a respectable spindle motor.

Coupling nuts...

A quick aside: After my previous post and all that mucking about with capa in the bearings, MsJaye tells me that wall plugs work as well... and she's dead right, they work splendidly. I'm redoing most of my bearings (not the ones on the plates, they're not worth the effort) to extract all that valuable capa for reuse. :D

The coupling nuts proved to be a significant challenge... namely, I couldn't find any of the little buggers here in Brisbane. Bunnings didn't have them, the local (smaller) Mitre 10's didn't have them, and after MsJaye mentioned that her Mitre 10 did have them... I discovered that the big Mitre 10's further away from my residence did not have them either. Le Sigh.

I would be taking up MsJaye's kind offer to send me a couple, except that I think I have a potentially superior solution.

A big problem with the coupling nut is backlash. The nut must be loose enough that it can be turned on the threaded rod, which unfortunately also means lower accuracy in positioning it. I was thinking about using a few nuts together to simulate a longer coupling nut, when I realised that it could significantly eliminate the backlash issue. Tightening one nut against another pushes each nut to engage a different side of the thread, which means lower backlash. By taking two nuts tightened against each other in that fashion and capturing them in a bracket, I would get a tunable anti-backlash nut.

Obviously the earlier tradeoff works in reverse here - the more backlash that is removed by tightening the nuts, the greater the friction is. Minimal friction is important, so I only tightened the nuts very lightly by hand, giving me something similar to this photo:


Note that the nuts do not necessarily line up when tuned; it depends on the nut and which way around it was put on the thread. In fact, it is probably an advantage to be offset; when the nuts are captured in the bracket the hex shape will ensure the nuts stay at the same position relative to each other.

The capture bracket is made with (drumroll)... capa! It took a few iterations. I roughly formed a bracket by hand the first time, but that actually used 4 nuts. I decided 2 nuts was sufficient and disassembled it. The second time I took ages forming a bracket mould using modelling clay... and the result was massively oversized, ugly and completely unusable. It took a long time to cut the nuts free to boot.

The third iteration saw me roughly moulding it by hand again. Pressing it up against a cold metal support beam in my kitchen gave a decent flat on the bottom side. And when I was done I had a couple of these:


I said capa acts like chewing gum; so it shouldn't be a surprise that many things made with it look just like they were made with that self-same gum. In any case, this was only the initial state. The capa which formed around the threads at the sides had a lot of friction, and... well, the ugly thing looks like it was made out of gum. I used a hobby knife to clean it up and remove the excess capa from the sides:


A couple of holes drilled slowly (and not terribly squarely) into the sides and I had my coupling nuts. They're very smooth moving on the threaded rod, and the backlash is imperceptible by fingers. I may find I need to remake these with tighter nuts once the mill is finished, but it's a pretty simple task. Additionally, one of the brackets allowed the threaded rod to wiggle a bit side to side; I used this for the long X axis, and I hope the extra play (which doesn't appear to affect the backlash) will help prevent the thread from binding when it's in use.

Bearings!

In the previous post I showed bearings mounted to the stages, but I didn't go into particular detail on how I did it. This is rather important because the bearings have a big hole in the middle (approx 8mm ID), much wider than any fastener I have to go through it. A 3/4" washer on each side is perfectly sized to squeeze the inside ring, but if the screw gets even slightly loose then the assembly fails.

Therefore I needed to pack the centre of the bearing with something, and I immediately reached for my stash of polycaprolactone. For those not familiar with capa, it's a low temperature thermoplastic variously known as Polymorph, Shapelock and Friendly Plastic. It's getting easier to obtain here in the antipodes, but a couple of years ago I only managed to get some from www.thethreadstudio.com, which I have subsequently hoarded like it was gold and I was some sort of troll or leprechaun. (although my Irish Foreign Birth Registration hasn't come through yet, grr)

At normal temperatures capa is like nylon - hard, smooth and slippery. But stick it in a pot of hot water on the stove and you get this:


Yes, it's goth chewing gum (what flavour would that be? Answers on a postcard and sent to someone else; I think I'm better off not knowing). I wouldn't suggest actually putting it in your mouth, but gum is exactly the consistency it obtains. It's pretty sticky in this state; it especially likes certain plastics and stainless steel, but of course it's not too hard to peel off once it's cool.

First I filled the centre of a bearing with a small amount of capa and let it solidify. This very quickly showed itself to be a bad idea; not only is capa touchy about being drilled (drill bit gets hot, capa melts, things jam up) but this particular bit of capa was in the middle of a bearing! When I tried to drill it the bearing span freely and I got nowhere. Oops!

Good thing about capa is that it's almost infinitely reusable. I popped the slug out of the bearing and put it back into my stash for attempt two.

Attempt two involved putting a screw into the bearing and packing capa around it. Or more accurately, putting capa around the screw thread and sliding the bearing over that while it was still soft. Then the capa could be squeezed into the gaps and the screw centred as close as possible with fingers. Once the capa went hard, a hobby knife/scalpel was used to cut the excess capa away. It's much fiddlier than it sounds; the capa had to be squeezed into a thin tube to wrap around the screw, which made it cool down much faster - if I took more than 5-10 seconds then the surface of the capa would stop sticking to itself and I'd have to start over. And I never did get a screw perfectly centred - but at least that turned out not to matter.

The previous post showed two of the three bearings mounted to each stage; here is what I'm using for the third:


It starts out the same as Tom's design here; I cut a couple of strips of aluminium from some spare U channel (which is then turned into L channel - good for when I come to mount the stepper motors later on). The bearing and the screw were self-tapped into 3mm holes at the end of the strip; here's the underside:


I've since cut off the end of the bearing screw with the dremel and ground it close to flush.

These bearings go on the other side of the stage's primary rail and hold it tight. Well, that's the theory, but as things stood even a small lateral push would twist the bearing loose. But I already had just the thing to fix it from my previous trip to Reverse Garbage; a spring! It's a tightly wound tension spring, about 0.9" in diameter and 15cm long, and exceedingly stiff.

I originally was going to use the spring on the Z axis like in Tom's design, to hold the dremel down and provide some protection to the dremel if it's jammed down into the stage by accident. I've since gotten an incredible deal on a clone flex shaft off Ebay ($20 including shipping!) which I will use on the Z stage instead and won't need a spring for.

The dremel made short work of cutting two 5cm springs, but bending a couple of turns of spring perpendicular at each end was a nightmare. However once that was done, and a small tab dremelled out of each aluminium strip above, I was in business:



Nearly all my screws were much too large to use for anchoring the spring; in the end I'm using a couple of very long screws that came from a surplus stepper motor. It's far too long though; it sits about 1cm proud of the surface, so I'll have to find a suitable replacement. I am a little worried about the spring pulling the screw out of the plastic, so it's been angled away from the spring line to try and secure the spring more firmly.

It seems to work splendidly - the stages simply don't budge when I try to twist them manually. I have to be careful over-extending the stage positions though - once a bearing gets to the end of a rail the spring flips the stage around pretty violently :) I'll be adding limit switches at the end of the rails later to prevent that from happening.

Next time: coupling nuts. This post just got a *bit* too long to include it here...

CNC So Far...

I've wanted to build a CNC mill for quite a few years now, but most of the DIY projects around were all a little too involved (read "required woodwork") for me to bother attempting (I'm really not a wood guy.) That is, until Tom McGuire published his Easy Mill on Instructables. Since most of the structure is standard 1 inch pipe screwed together by hand all the excusesreasons I had for postponing the project were gone.

A couple of weeks later, and I have something to show for my toil. First up is the main frame; it's as close to Tom's build as Bunnings would allow - that explains the brass piece at the top; it's the only fitting Bunnings had with screw holes. The rails are attached to the pipe with self-drilling, self-tapping screws; this saved me a lot of trouble with drilling the pipe as I only needed to get ~4mm pilot holes drilled with my hand drill; the screws and my stubby ratcheting screwdriver (which can generate prodigious amounts of torque) did the rest. The rails extend beyond the left edge to allow the full use of the X stage width.


The X and Y stages are cut from a large plastic offcut I got cheap from my local Reverse Garbage. The 60cm by 29cm by 1.5cm piece only cost me AUS$13, and whilst I don't know exactly what material it is, it's smooth, easy to work and hard wearing. I'm guessing it's actually nylon - it doesn't feel like acrylic, and a rough density measurement seems to match. In any case, I'm pretty lucky; the rails don't seem to damage it.

Below is the top of the X stage. I discovered that while most of this CNC design does not require rigorous attention to accuracy, there are one or two places where it's critical, and the primary rail (to the right) for the Y stage is one. It needs to be very square to the primary rail for the X stage below it, and this is a real pig to measure with the big block of nylon in the way. A piece of channel, a straight edge, a set square, a lot of faffing about and associated cursing and I think I have it close. If it needs trimming I made one of the mounting holes oversize to facilitate this later.


I was going to use more U channel under each stage as in Tom's build, but when I got to that point I changed my mind and used bearings. Here's the current underside of the X stage:


And here's the current underside of the Y stage (you can still see the price on it...):


When it's all together, you get:


(Ok, so I didn't quite align the top stage properly for the photo... :D )

Next time: Bearings and the elusive coupling nut.