Freesteel Blog

Berlin railway machining art

Thursday, January 10th, 2008 at 5:43 pm Written by: Julian

Here is a detail from the base of the ten metre tall metal sculpture Rolling Horse by Juurgen Goentz (2007) on the flank of the Berlin Hauptbahnhof exhibiting atrocious machining on its the panels, including a full cutter gouge where the tool appears to have been buried 5mm into the surface.

Possibly this was in preparation for clearing the next level, which didn’t happen. Normally you helix down, because this type of plunging motion is liable to break the tool, I am told. Nobody I have shown it to thinks it was left on purpose; there is no artistic appeal here.

The rest of the motion looks like an offset clearing strategy with extra segments in the acute corners to knock off the upstands. This is a pattern I have seen on WorkNC. I should have measured it carefully or taken a clearer picture. Whenever you offset the clearing passes by more than half the diameter of the cutter, you create the possibility of leaving triangles of uncut material at the corners of the contour, depending on how big the offset is. The uncut areas are usually triangular with one side being a concave arc, like the area of unpolished floor in the corner of a room that a circular floor polisher cannot reach.

In this example, the offsets look pretty close to the diameter of that mis-cut hole.

On Machining Strategist, Depocam, HSMWorks, and any other machining kernel that I have had something to do with, the offset Z-clearing strategy checks along each contour, measures the angle of the tightest corner, and offsets the next pass appropriately so as to clear all the material. This means that when the user specifies the step-over, they have to set a minimum and maximum value so that the system has a range to work with. The defaults are minimum = 50% of diameter, and maximum = 90% of diameter. Sometimes there are full-width cuts, which are unavoidable with this strategy, but as long as the minimum is never more than 50% (it won’t let you set it to more), it will clear everything.

This sometimes annoys users who want to give a definitive large value for the step-over and not be troubled with this geometrically compromising min and max nonsense. You will find in the user manuals — particularly for versions of non-MS/Depocam software from five years ago — advice on how to set the absolute offset for the clearing strategy. Users want to set it as high as possible to get the job done in as few offsets as possible, but if they set it too high they run the risk of leaving behind these upstands — and it will be their fault because they picked the number.

If you have a rectangular area whose corners are right angles, my calculations say you can go up to 85.355% of the cutter diameter before you get upstands, because that’s equal to 0.5 + 0.5 / sqrt(2).

The next best thing to choosing the appropriate maximum offset to avoid any upstands (to my knowledge a unique and identifying feature in MS/Depocam, replicated in HSMWorks) is for the computer to notice where they occur and clean them out specially. This is an act of retrofitting a correction onto a flawed strategy. I have seen it implemented in the form of little dashes of toolpath that nip in from the side of a contour one tool diameter away from the corner, and by other dashes that go out from the corner along the bisector far enough to clear all the material. It’s ugly because you get sharp 180 degree changes of direction that are not conducive to smooth high speed machining motions.

The primary example of such retrofitting is in WorkNC, as illustrated on their circa 2004 webpage (hosted by archive.org). The WayBackMachine shows a change to the WorkNC webpage in 2005, where these detailed and clear drawings of their strategies disappeared, so I don’t know if it’s still what happens.

The best answer is to use our now-quite-old-and-stable Adaptive Clearing strategy to make a lovely swirling finish to the part. No self-respecting artist would pass this clearing strategy up — if they knew about it — because it’s clean, unique, and looks good.

Unfortunately, I strongly suspect that the art of machining of metal is something that working artists just don’t know anything about. They most likely contract out the menial work to some tool shop with its cranky old expensive software and then pay no attention to the process, because if they did they would quickly notice how trivial it was, with an enormous potential for making gorgeous textures and strange mathematical shapes using kinder-garden level programming skills and a few days of experimentation.

I’d like to work with an artist on this one day. I’ve got a lot of ideas they could use. I can’t do it myself because I am not an artist — defined as someone who can sell their art. Anyone can make art; selling is the social certification which is not achievable by me. I mean, I find it difficult enough getting respected for writing software of the kind that very few people in the world appear willing to do.

Also seen on one of the Rolling Horse’s metal panels is this interesting spiral shape:

Analysis of the scouring suggests that it has been done from the outside inwards, which is unusual. The best theory I can find is given on page 4 of the WorkNC Version 18 brochure:

New Spiral Core Roughing Toolpath

We have designed a new roughing toolpath to generate optimised fluid toolpaths on core shaped parts. A feature of this toolpath is the low number of retracts due to a continuous spiral trajectory at each Z level. Machining times are kept to a minimum and tool service life is maximised.

although there is no actual core here. Given the quality of the rest of the machining on this base, its aesthetic appeal is probably accidental.

Still. If the Computer-Aided Manufacture (CAM) industry weren’t so invisible, its software development threads couldn’t have survived in such a stagnated state of mismanagement by the entrenched system of corporations, giving me a chance to beat it with code written in a spare room in Liverpool.

We’ll see how the benchmarks perform when I finally get through with this round of constant scalloping stepover work. Often it gets difficult to keep up the motivation to plug at it day after day. Sometimes it takes a dose of big-headedness to get the will to carry on.