CNC Router frame under construction.

Some notes on making a welded steel frame.



Welding steel box section is a quick and low cost solution to producing the basic framework necessary for any CNC machine. With the right equipment and some practice it is not difficult to master the techniques and achieve very acceptable results. (A Google search for 'welding techniques' should bring in sufficient information to get anyone started). There is however one major drawback to welding which, due to heat and metal expansion with it's subsequent contraction upon cooling, introduces distortions into the finished work and If these distortions are considered to be inevitable then some way of correcting for any errors must be incorporated into the design.


My framework was constructed as accurately as I could, by carefully cutting each part squarely to the correct length and using jigs and clamps to support the work during the welding process. The completed frame was as square and true as I could measure with basic tools but the important rails to which the linear slides were to be attached were not flat to within +/- 0.25mm. To re-gain the required flatness I chose to use alloy shims on each of the four ways. These shims were bonded to the steel and then lapped flat, square and true. Lapping was achieved using a retired (300mm x 600mm) iron surface plate to which I fixed, with double sided sticky tape, emery paper and then turning the plate upside down it was slid back and forth along the ways, checking for square, flat and true at regular intervals. This process was extremely slow, however, the end results were well worth the effort.


Although it may well have been easier and simpler to have made a frame from proprietary alloy extrusions, bolted together and thereby avoiding any induced distortions I am certain that it would have taken around the same amount of time yet cost much more. If and when I do build my next CNC machine I will still opt for a welded steel construction for the framework but I think, that by careful design, I can confine the distortions to less critical areas.


Fig.1 illustrates a typical type of distortion (somewhat exaggerated) which is induced by welding two pieces which were initially touching each other. If the base plate was strong enough not to have deformed then tremendous stress would have been induced in the weld which at some time in the future would almost certainly be doomed to failure.


Rather than getting bogged down with stress relief techniques, I think that time spent looking at alternative approaches is the way to go.


The example of Fig.1 would have turned out differently had a small gap been left between the two parts prior to welding. I have observed professionals using the stub of a welding rod to produce the gap for the first tack, then removing the stub and leaning the upright to increase the gap slightly more, apply the second tack (on the opposite side to the first). Cooling of the second tack draws the upright back to the true position. A similar process is repeated for more tacks then the weld proper is completed. This technique allows the upright to be truly upright and minimises distortion to the base.

(1) Welded steel frame made from 25mm square box section just sitting there waiting for the paint to dry.


(2) Alloy strips 20mm x 2mm(shims), bonded to the frame where the linear slides are to be fitted, these are lapped flat and true to correct for the distortions introduced in the frame by the welding process (each pair of linear slides need to be fitted perfectly parallel and true to each other).


(3) To avoid the need for tight tolerance working, oversize clearance holes were drilled in the frame and tapped 'nut plates', made from steel strip 14mm x 5mm, slid inside the box section and used to secure the linear slides by way of M5 cap head screws. This method of construction allows for precise, parallel and true alignment of the slides during the final assembly.


(4) Nut plates for the X axis slides. The cut-away in the frame is for the X axis stepper motor. The original concept here was to use a belt drive between the steppers and lead screws, however this idea was dropped in favour of direct drive thus necessitating the cut-away to make space for the motor. It is not always possible to get everything right first time.


(5) Ball screw for the Y axis. Again oversize clearance holes in the frame to allow for precise alignment in the horizontal direction during final assembly. Both the X and Y axis ballscrews are of the pre-loaded type with twin circuits of ball bearings - the advantage here is that they truly have zero backlash.


(6) Flexible plate used to connect the nut of the ball screw to the table. The plate is wide enough to resist torsion but flexible enough to compensate for any misalignment between the ball screw and the slides in the vertical direction. It is a lot easier to build in some form of flexibility than it is to use shims.

CNC machines are inherently more dangerous than manually operated machines of the same type.

Keep your hands and any loose clothing well away from running machinery.



CNC Ballscrew mounting.CNC Ballscrew mounting.CNC router linear slides.

Here are some more pictures taken as the assembly work progressed.

CNC mounting for ballnut.CNC Router frame.

Only joking !

CNC Linerar Rails and Ballscrew.

CNC is only limited by our imagination.


Rounded Rectangle: CNC Home.
Rounded Rectangle: Construction 1.
Rounded Rectangle: Construction 2.
Rounded Rectangle: Construction 3.
Rounded Rectangle: Construction 4.
Rounded Rectangle: Construction 5.
Rounded Rectangle: Routing & Cutting.
Rounded Rectangle: Vinyl Cutting.
Rounded Rectangle: Lithophanes.
Rounded Rectangle: Engraving.
Rounded Rectangle: Making a Vinyl Cutter.
Rounded Rectangle: Soft & Hard Issues.
Rounded Rectangle: Micro Stepping.
Rounded Rectangle: Bob the Mill.
Rounded Rectangle: Making Printed Circuits.
Rounded Rectangle: Making a 4th Axis.
Rounded Rectangle: Tool Position Setting.
Rounded Rectangle: Low Power Lasers.
Rounded Rectangle: Gallery.
Rounded Rectangle: Related Links.
Rounded Rectangle: Site Map.
Rounded Rectangle: Sam.
Rounded Rectangle: Optic.
Rounded Rectangle: The X Files.
Rounded Rectangle: Bertie.
Rounded Rectangle: The Rover Jet Engine.
Rounded Rectangle: The New Laser Project.
Rounded Rectangle: The RF Laser.
Rounded Rectangle: Various Projects.

copyright (c)2016, all rights reserved.

Some retrospective thoughts for the design of a small CNC machine which is intended to use a spindle mounted cutter.


When designing a new machine from scratch ‘rigidity’ is the most important single factor and if you have some engineering knowledge then this can be visualised fairly easily, the choice of materials and how they are fixed together can be just as important as the design itself. If there is doubt then make a model or prototype to test the design (or part design) prior to committing yourself to making expensive scrap. Internet searches and the study of others machines and their construction methods, learning from others mistakes and looking at the quality of the work produced by others will undoubtedly be worthwhile here.


Although rigidity could be obtained by using great chunks of cast iron, this creates weight, which if it is a moving part has inertial mass and this, in turn, will require more torque (larger motors and higher drive currents) to avoid slowing down the acceleration / deceleration performance of the finished machine and ultimately the speed at which the work piece can be completed. Inertial mass can be advantageous in certain places on a machine but in general, for efficiency, it should be kept to a minimum at the design stage.


Rigidity and low mass is the key to a successful machine design, well almost but not quite….


When a machine is put to use there is another limiting factor which is not so easy to visualise at the design stage and this factor is resonance. All things have resonance (to one degree or another) and material density is one of the factors which will increase or reduce it. I have, for example, seen quite substantial, professionally made, machines with the spaces between the webs in the cast iron filled with a granite/epoxy compound which looks for all the world like concrete and although this trick does little to improve rigidity it can substantially reduce resonance. I think all machines will suffer from resonance but for a good machine this will not occur within it’s envelope of recommended usage, however, get the feeds and speeds wrong and tool chatter or resonance commences, completely destroying any expectations of a good surface finish and possibly even causing machine or cutter damage. Symmetrical designs may well be more prone to resonance than asymmetrical designs which have a cancelling or dampening effect on vibration due to the differing natural resonant frequencies of the component parts. In general, construction materials which have a high density will perform better than materials with a low density but there are exceptions and although they may look more professional, machines constructed from aluminium extrusion can quite often be out performed by machines constructed from MDF as far as the achieved surface finish is concerned.


If we consider a good quality, small, centre lathe and the unsupported areas of the machine (the gap between the point at which the cutting tool touches the work and the machines bearing surfaces or slides) this distance is quite small and so the machine performs well. But if we fit a 150mm boring bar or have the work protruding from the chuck by 150mm, then to avoid chatter and bad surface finish the speeds, feeds and cutter loadings will have to be adjusted accordingly. There will come a point, longer boring bar or further work protrusion, when it is impossible (at any available setting of speed or feed) to achieve a good quality of work or be able to take any reasonable depth of cut without resonance setting in. This must be considered in the CNC machine design - a small gap between the cutter and the machines supporting bearings will allow reasonable feed rates and depth of cut whilst a large gap will necessitate higher spindle speeds and shallower depth of cut for the same feed rates.


Another point worth noting is that our manufacturing tolerances (and subsequent quality of build) will depend to a large extent upon our skills and the available machinery and tools that are to hand. I doubt we could all make the component parts to NASA standards so some form of adjustment must be ‘designed in’ so that precise alignment of the finished machine can ultimately be achieved. If any machine is not capable of being set up ‘square and true’ then the work produced by that machine will almost certainly be a disappointment and after all, we want to own a machine that is capable of producing quality work not just average stuff.


I now have two, different, CNC machines which are best suited for entirely different tasks and neither would be good at doing the other’s job. Tweakie is a router, which works extremely well with wood and plastic etc. and Bob is a milling machine which is designed mainly for use with metals. The gantry style construction of Tweakie has large gaps between the point at which the cutter contacts the work and it’s supporting bearings, and as a result it’s capabilities are limited but it’s good point is that the working area is reasonably large. Bob, on the other hand, has extremely small unsupported gaps but has a much smaller working area with limited spindle speed and as such would hardly be suitable for working with wood. A comparison of these two different designs with their capabilities and limitations is important in so far as that an idea of the intended use for the finished machine is almost certainly a prerequisite for it’s initial design.


To summarise, the ideal machine should (as near as is possible) be rigid, with low mass, constructed from dense materials and have the smallest possible unsupported gap between the cutting tool and the supporting bearings or slides whilst trying to avoid any two or more parts having the same natural resonant frequency. Perhaps an impossible combination but it is something to aim for.