tim   .


newpad.gif (50 bytes)

.   hunkin






CNC Video 

Cogs with 4mm spacing between the teeth, about the smallest thats practical. Below is the rear of the cog showing the dross - which can be tapped off with a hammer or ground off.

A small cog cut in 8mm steel. The one on the right was cut with a worn nozzle and electrode but even with new consumables the cut (left) isn't completely vertical though it still works as a gear!

A good test of accuracy is to cut a square and then a square hole the same size and see how they fit. The software automatically adjusts for the width of the kerf but the cut sides taper a bit so the parts never fit perfectly.

Holes have to be cut oversize because of the taper. I prefer not to enlarge the holes with a drill because the sides are quite hard and not good for drill bits. Also The start of a cut is a bit messy so when drilled out the holes aren't always exactly centered. 

The lack of precision can have have charm 


Automata video

I bought a hand held plasma cutter in the mid 1990s. Compressed air is swirled round a spiral ‘electrode’ to produce a focused jet heated by a large spark (roughly 200VDC at 10amps). The heat ionises the air making it a plasma (the definition of a plasma is an ionised gas). The plasma is conductive so it can transfer the energy and heat into the steel.

Its an exciting machine, spraying sparks everywhere as it cuts through steel up to 12mm thick. Its actually much less scary than an oxyacetylene cutter, which sprays globs of white hot metal about and uses the potentially explosive mix of oxygen and acetylene. I have used my plasma cutter a lot, getting good at drawing by hand with it. I usually draw the design on paper and stick it to the steel with spray glue. The paper burns a bit but not so badly that the lines can’t be followed. However I’ve long been intrigued by the idea of a CNC plasma cutter, which has the potential to cut more accurate and delicate shapes.

I've also used quite a few different CO2 laser cutters, also fun. Cutting acrylic is particularly satisfying because the edges are so polished and perfect. However, the list of materials suitable for CO2 laser cutting very limited. Acrylic is weak and cracks easily and the special laser plywood is horrid stuff. A friend tried cutting some proper 6mm birch ply with his 60w laser but said it made a lot of smoke and only just cut through. I liked the results cutting Formica laminate, but although it wasn’t on the list of banned materials I suspect it did produce toxic fumes. I’ve also tried cutting Acetal (Delrin). I use a lot of this material so it would be useful but found that it needs a lot of power. An 80w laser only just cut through 6mm black Delrin, and the edge was quite tapered.

What I really want is a machine, the size of a laser cutter, that will cut the materials I like – steel, brass, aluminium, birch ply and Delrin. Waterjet cutters could do the job, but need a minimum 25kw power to run, not practical for me down the end of a long lane in the country. Metals can also be cut with a YAG fibre laser. The price of these machines keeps coming down but still cost £30k and consume a vast amount of argon or other inert gas as shielding. A salesman told me that 6 bottles should keep me going for a couple days (each bottle costs about £50).

A plasma cutter is the poor man’s option. For less then £5k it just requires a 7.5kw single phase supply and compressed air to cut up to 12mm steel. It can actually cut any metal, but the instructions never mention brass, I guess because of the lead fumes. Plasma cutters produce a wider cut than a laser, (.5mm to 1.5mm depending on thickness) so the process is not as accurate and there’s usually a bit of dross to be knocked off the cut parts. But its still an amazing tool. The inaccuracies can have charm and in many situations aren’t a problem. CNC plasma cutters used to be enormous machines for cutting 8ft by 4ft steel sheets, but much smaller ‘hobby’ machines are now on the market. 

My machine is a UK made 'Swifty'. Unlike larger machines it has no z axis feedback to keep the torch the right height from the work. On small sheets this is less of a problem and I've had little trouble without it. It has a very ingenious feature to pierce at a greater distance before dropping for the cut. Its generally well thought out, though I think they ran out of time or energy when designing how the torch connects to the machine. The clamp doesn't grip strongly enough so the torch lowers a bit each time it drops on the work and the magnets connecting the torch bracket (a safety feature to enable the torch to break away if it bumps into something) are too weak so it sometimes comes off when you don't want it to. Fortunately its not difficult to modify both so its not a major problem. .     

The software isn’t as straightforward as a laser cutter. The vector drawing has to be converted to a DXF file. Most CAD programs can save files in DXF format. I often start with hand drawn shapes, which I tidy up in Photoshop and convert to vectors using Illustrator’s ‘live trace’ feature. (I find live trace (Illustrator CS3) particularly frustrating. It usually produces a double path, often open ended, I really need a better way of creating vectors from bitmaps).  The DXF then goes to a program that converts the DXF to a cutting path called a .tap file or G code. The G code is loaded into a second program which controls the plasma cutter. All this takes some time to get used to – I still usually do a ‘dry run’, running the torch round the path with the plasma switched off to check everything is OK.

Once a file is loaded, and the plasma cutter electrode and nozzle are is good condition, the accuracy isn’t bad. The worst inaccuracy is that the edge of the cut isn’t usually vertical, obviously more of a problem on thick material. Small holes usually taper so have to be cut 1-2mm oversize to fit a bolt. I’m still learning. 

I just made a useful discovery for cutting thin (under 2mm) materials. Although a TAP file would look perfect, it often cut off the corners. Here's the bad example of a row of houses 600mm long, cut from 1.5mm ali:

Its difficult to even recognize them as houses. I found it was caused by the torch moving too fast. The way the machine works is to keep the plasma cutter at full power for all material thicknesses, but to adjust the speed  so thin materials cut quicker. Maybe the stepper motors on the Swifty aren't powerful enough to cope with really high speeds - the machine is really intended for thicker materials. Anyway the solution is to program the cut for a thicker material and then turn down the power on the plasma cutter. (In this case I used a fine cut nozzle with the plasma cutter turned down from 45A to 30A, programmed for 2mm and slow cut in Lubelia, then 50% reduced speed in Mach 4. The result is a great improvement):



18 months later

I'm now quite confident using my Swifty, its a good tool but with annoying problems took a while to sort out. One problem was Solid Edge, the CAD program it comes with. This works but is so obviously a limited version of a 3D CAD program. Being large, it takes ages to load and exporting the files in DXF format is slow because its not Siemens native format. I looked at a number of other CAD programs and eventually decided on Librecad, because its only 2D and the files are DXF by default. Its not perfect, but I persevered and now use it quite well.  I also abandoned the Lubelia 'Swifty' software which converts the DXF files to G-code. Lubelia plasma software is normally expensive, I think its most unique feature is slowing down the speed before making a sharp turn. But its been brutally knobbled to fit the budget spec of the Swifty so is erratic and won't do useful stuff like adjusting the allowed accuracy of DXF lines that don't quite meet. I bought Sheetcam (which Swiftcut use on their bigger machines). None of the posh speed control of Lubelia, but I've gradually realised that the cut is just as good. I actually get much better cuts, because its so easy to adjust the settings for different materials and thicknesses. I eventually found that the annoying rounding of corners (see photo above) comes from the Mach 4 profile written for the Swifty. This sets limits on the acceleration and deceleration of the stepper motors to ensure they never 'jump' a step. Basically, the stepper motors are not powerful enough to turn tight corners above 2000mm per minute.   




about the site


links to
other sites

where to see stuff

contact me