Kit's Machine

[Kitwn]

Not so much a 'Build Log' as this is now a finished machine, but I've been asked to describe the machine I'm using so this seemed the obvious place to put the description.
First a video of it doing something useful:



This picture shows the pile of source material I managed to scavenge off the local tip over a period of a few months. The design of the machine was very much dependant on the material available.

[NB70]

Thanks Kit - great video. This link worked for me: https://vimeo.com/444172007

[Kitwn]

Thanks Kit - great video. This link worked for me: https://vimeo.com/444172007

Thanks, I've editted the original to match.

Kit

[Kitwn]

This machine has not so much been designed as it has evolved. I DO NOT RECOMMEND THIS METHOD!!!!

I started out making a plywood and MDF construction which served no useful purpose other than to prove that I could pull together all the various components, mechanical, electronic and software, to produce a working whole. The current incarnation, about number 5 I think, is made from steel box-section scavenged off the local tip. The design has been heavily influenced by the materials available from that source and partly by the collection of components I had gathered for the previous incarnations and other, since abandoned, projects.

The base frame is made from four 800mm long, 100 x 50mm pieces bolted on to two 40 x 40 x 1300mm stringers. I had no welder at the time so bolted construction was the only option. The plywood baseboard is an essential part of the machine and keeps the frame square.
This frame was never going to be very rigid so the whole design of the machine had to allow for adjustment after the fact rather than built-in perfection. This is achieved by sitting it on 4 adjustable feet. One of the current major weaknesses is that it sits on an old wooden table rather than a solid metal or concrete base to provide the rigid support it needs.

Previous experience with the wooden machines had shown up the difficulty of getting the rails and ballscrews suitably aligned all at the same time so I opted for a design based on two ‘linear actuators’ for the long axis. Each consists of a 1300mm long piece of 65mm square (I’d have preferred 75mm but the supplier didn’t have any at the time!) box section with a 1200mm SBR20 rail mounted on it and an 1100mm 1610 ballscrew plus direct drive stepper motor. The thinking was to get all these components aligned with each other as a single component and then move the whole structure to bring the complete machine into alignment without any readjustment of the rails/ballscrews relative positions. This has worked very well in practice but the overall rigidity of the structure is not as good as for other methods. The picture below shows an actuator with an earlier set of rails fitted.

The two actuators are fixed onto the base frame with a single bolt at each end. This allows lateral adjustment to keep the spacing between the rails identical along their length and shimming to ensure the rails are in the same plane. Fine adjustment is done with the feet on the base.

[Kitwn]

The original steel gantry was a bolted design and a complete disaster so something new was required. The quote for two lengths of aluminium profile, including delivery to the wilds of Western Australia had me reaching for the smelling salts, so it had to be steel again. I bought a cheap stick welder for about 1/3 the cost of the two pieces of profile and went for it. Fortunately I still had enough material left in stock.

This was my first attempt at welding and uses a technique I have described as ‘Bird Poo’. An angle grinder, plenty of car body filler, some sanding and a coat of paint hides a multitudinous multitude of sins.

To keep things simple I decided to put both rails on the front face and use epoxy to provide a guaranteed, perfect level surface… Ha, ha. Unfortunately this was before I’d heard of the recommended West Systems 105 resin/209 hardener combination so I bought some cheap stuff off eBay. The advert includes two options, Low Viscosity (LV) which they claim is self-levelling and High Viscosity (HV). All I can say is that if the LV is low viscosity, how are you supposed to get the HV stuff out of the tin?
To be fair this product is aimed at filling cracks in concrete so should not be criticised for its poor performance on a CNC machine.

I also chose to drill the holes for the rails before applying the epoxy (can’t remember why now) so there were greased screws in the way to prevent the gloop from flowing. The end result is probably less flat than the bare metal was but is glass-hard so I have shimmed the rails rather than chip off the epoxy.

By a stroke of luck, my choice of 16mm supported rails and a 1610 ballscrew with matching mounting blocks meant that with the ballcrew mounted directly on the same epoxy plane as the rails there's a sub-1mm gap between the ballnut and the Z axis backplate. A can of ice cold Guinness was consumed to provide a suitable shim.

[Kitwn]

I first built the Z axis assembly as a plywood mock-up as I didn’t want to waste any of the aluminium plate I didn’t have anyway. The minimum width I could fit all the pieces into was 160mm so I went with that to maximise Y axis travel. At the time I didn't know where I was going to source the final aluminium or what sized pieces I would be able to get. By chance 160mm turned out to be the largest common standard width for aluminium bar and was available in 12mm thickness. Finding someone who would supply me with less than a 6m length was impossible until I eventually took a trip to Perth and found a boat builder willing to sell me an off-cut that would fit in my car. It was something of a relief to know that I would not have to do any more hand cutting than hacksawing the standard width bar into the required lengths.

The bar isn’t perfectly flat, it bows slightly across the width but is acceptably flat along its length and not twisted as far as I can see. This has to be good enough, like it or not.
Three lengths of the bar are used to make the assembly. One has the horizontal Y axis bearings on it. The second has the vertical X axis bearings, the third is the moving plate carrying the spindle. I used two pieces at the back because:

1) There are no access-to-mounting-bolt problems for bearings on both sides of a single piece.
2) There’s some wiggle room to adjust the Z axis travel direction to be exactly perpendicular to the Y.
3) Doubling up on the 12mm thickness adds stiffness.
4) It's easy to remove/replace the complete Z axis assembly for alignment or modification purposes.

The plywood prototype Z assembly was adequate for milling the clearance holes for recessing the bolt heads in these plates and using a centre drill to mark the positions of the fixing holes but I elected to do the actual drilling on the manual drill-press. Brackets made from 6mm aluminium offcuts hold the moving proximity sensors for both Y and Z axes, the sensing plates for these are made from 2mm x 20mm alminium angle, one of the few materials used for this machine that I am able to buy locally.

16mm unsupported rails are used for this axis simply because I had all the bits left over from something else. They don’t appear to be a source of significant flexing compared to all the other possibilities for movement on the machine. A (dubious) advantage of these is that the flatness of the plates used does not affect the straightness of the vertical travel of the mechanism. That’s all down to the rails themselves. Once again fitting the 1610 ballscrew on the same plate as the bearings left just a small gap to be shimmed when attaching the ballnut.

[John McNamara]

Gee Kit

The Aztec Sun Gods will be pleased with your machine build. I like the emphasis on sturdy, using "Found" materials, Its a credit to you.

[Kitwn]

Gee Kit

The Aztec Sun Gods will be pleased with your machine build. I like the emphasis on sturdy, using "Found" materials, Its a credit to you.

Thank you John. High praise indeed from someone of your standing.

Kit

[Kitwn]

When I built my first Heath Robinson machine the choice of controlling software seemed to be between MACH3 which almost everybody was using and LinuxCNC which was free. I went with LinuxCNC running on an old Windows XP era machine built from the best bits from a couple of dust-gatherers I obtained for $20 the lot by advertising on our local ‘Buy-Sell-Swop’ group on facebook. This is still working well.
The original $25 Chinese breakout board is still working though I did have some fun with noise pickup when changing from using mechanical microswitches to proximity switches for limits and homing. That fun and games is documented elsewhere on the forum.
Four 1.9Nm stepper motors and DM452T digital drivers from Stepperonline provide the motive power. These are way superior to the TB6600 based drivers I originally used, the motors being much quieter and cooler in operation.
Power comes from a pair of 36V 10A switch mode supplies, one for the two X axis motors and one for Y and Z. I have had no problem with SM supplies as long as the rated current is well above the total peak current required by the motor drivers. If I were starting again I’d use a higher voltage linear PSU and drivers to suit but I can still rapid at 9m/min with this arrangement.
The electronics is mounted on an MDF baseboard housed in an old PC case. It isn’t exactly pretty but functions just as well as a more expensive, professional looking box costing more than a case (or two) of fine Western Australian Shiraz. 5V for the BoB and 12V for fans is provided by the original PC power supply.
The VFD for the 2.2Kw water cooled spindle is mounted on the wall rather than in the box. Partly because the box isn’t big enough but also because I know from professional experience that poor cooling airflow is death to solid state power electronics so I wanted it where it can breathe freely.
Water cooling for the spindle is provided by a tub full of about 15l of distilled water with a 240V submersible aquarium pump.

[Kitwn]

This entry isn't the 'Build Log' it's the Un-Build Log! Ironically after spending so long designing and building, re-designing and re-building and at last getting the beast nicely aligned and ready for some serious work I've had to take it all to bits and put it into storage, having just sold our current abode in Western Australia, where it will likely stay for a good year or more until I have retired and we're in a new home in Tasmania. Talk about bad timing! I suppose the fact that we decided to put the house on the market now may have been a contribuitng factor.

Anyway, here's the Un-Build Log video...