Very good, now my mind is ready!
APT stands for Automatically Programmed Tools or any variety on that theme ... It's both a language and a program, came from MIT originally at the same time that CNC was developed / invented / designed, and by many of the same people. Then the aerospace industry took over development ... it was the very first ANSI standard and most of the aircraft built before 1980 would have been designed with APT. It was still common into the 90's ... the wikipedia page is not bad but if you're lazy ...
Nice, now that I put to work those old rear sprockets, I seem to recall that in the school we studied a bit about APT around 1988/1989... along with DBase-II/III, FORTRAN, TurboPascal, and a few other things... that would explain why it sounded slightly familiar but I could not tell why...
But now I recall what was the use of APT there!, I think it was used in company of an orangey italian mini-lathe!
Anyway, sorry, you're right... I always use Google and Wiki for everything, but this time I given up too soon when I noticed a lot of non related results. So, double thanks. Yes, here is the APT Wiki
First step is to define the part using geometry statements. Common entities include circles, lines (in fact it's all 3d so a line is just a specialized form of a plane), cones, spheres, parabolas, hyperbolas, surfaces, mathematically-defined arcs, whatever. If you've played with FORTRAN the conventions will be familiar ...
CIRCLE1 / CIRCLE, POINTA, DIA 6
LINE2 /LINE, 12, 17, 44, TANTO, CIRCLE6
and so on ... gazillions of ways to define each entity.
Yes, I know what you mean. I can see easily the beauty behind such no non-sense approach compared to any amateur class GUI, but I would be lying if I say that my actual rythm of life can support such thing for routinary work.
I mean, most part of the time I use CAD to develop a model into a virtual environment, before to build the real thing... but that's the key point: 99.99% of the time I'm not doing it to produce anything serial. My actual workflow depends on CAD mostly to produce blue-prints that then I use to build equipment by hand! So, strictly speaking, most part of the time I only use CAD but not CAM... even if here and there I use CAM too to control a couple of very specialized tools that I've built here.
The next section is tool movement commands - turn on spindle at 3000 rpm clockwise, turn on coolant, start from point A, rapid to point B, go forward to the intersection of plane one with circle three, turn left on circle three at four inches per minute to the intersection with plane five, rapid to the start point, turn off spindle, turn off coolant, rewind, fini.
Very nice, looks like the kind of techie's black-hole in which I use to fall here and there!
...but I can't help thinking how good would be to write some good wrapper to put everything under some GUI! Don't hate me!
Rewind refers to tape, of course. Did you know that SR-71 missions were controlled by a computer running paper (mylar) tape ? I think until the very end. Funny.
Heheh!, nope, I was not awared of that! ...very funny, I guess the dummy mylar-tape sequencer was a lot less heavy than fully equiped PDP-6! A couple places here in the city used to synchro slide projectors by using dummy sequencers with tapes like that. I think our Municipal Planetary used something on these lines, but I think in this case was real paper (not mylar).
These days maybe that's too much work for people but compared to an HP-41, it's a dream. Also, as you can imagine, it's quite a bit more powerful than graphics programs if you are more of a programmer than a clicky-the-picture sort. For families of parts, just change the dimensions on your geometry. Run the part definition through the interpreter again. Viola, new part, same program. GE built entire families of turbine engines with a spreadsheet and APT loooong before MySQL was a gleam in Larry's gimpy eye.
Yes, I know that geometry constraints are still a bit hard under FreeCAD, and you got a good point there... but it is not so difficult to find the way to do it right, even if the app itself still has some bugs...
The program is in the public domain - hard to find the source code though, it's so old and not sexy for the anti-aliased fonts crowd.
Again, I think the problem is that I say "CAD" and you hear "CAM"!
...of course an integrated CAD/CAM environment is always better, but even if the CAM side of FreeCAD is still too young, the CAD side has little or nothing to do with the first versions... this thing evolved a lot, and for a lot of things you can use it with confidence... even if you end exporting the models to some machine trajectory optimizer.
Easiest way to try is to get a trial copy of Personal APT (web search) which is a subset of the complete language. Bob Drewry was involved in the original program, he's retired and messes with this in his dotage. He's created a Windows shell for today's users, I much prefer the DOS version. Price for the whole thing - program, manual, postprocessors - is around $300. Or was, last I checked. It's worth it. The manual alone is a full course in machine tool programming.
Well, I'm looking at the site right now: http://www.cncfocus.com/
...I see what you mean. I can't say it is not interesting, and looks very powerful.
Anyway... what is wrong with brand-new apps natively oriented to G-Code / RS-274?
It seems pretty soon FreeCAD could be doing it well: Organization ChartRobot Workbench
There were several other versions of APT out there but I believe most have fallen by the wayside.
Well, there are tons of CNC oriented stuff, and people wants to do whatever they do every day faster... I can see clearly that I'm not the only one hurried to get the work done, but I need to accept that I still hold it with some lyrism (i.e.: refusing to switch to MS Win even if many things are targetted to such so called OS)
The military still uses it, I'd imagine. They've got all those B-52 parts to keep in inventory.
I've messed with every free cad app I can find, you can bet on that
I don't like any of them very much. To tell the truth I am surprised there aren't any nice ones. There have been a few with promise but they seem to have fallen by the wayside. They are all so busy playing with their dicks that they forget the purpose of a CAD program - to draw/design stuff. 80% of the screen is icons. Hey ! Where's the damn part go ? Over here in this postage-stamp-sized mini-window ?
No, no, no Hamei! ...look at it (FreeCAD) now
What I'm saying is indeed that three years ago such app was a complete looser and now we (they) have something that is very-very cool and has lots of features that previously were only found into commerical apps! And believe me, has a lot of power under the cover... for first time you can do real CAD work with a FOSS app!
Just go there and download the actual thing! ...if you tried it when I posted this topic for first time, then you got the wrong picture of it...
Anyway, I understand your point: you're talking of mature industrial-grade CAD/CAM. There's no discussion about that. We'll have to wait some years before to see something F.O.S.S. including the whole real dealt.
There is also Irit and that BRM CSG thing. Neither of those seemed very exciting from the point of view of machining tho.
Nah, I understand. But APT is going to kick ass over anything from Open Cascade for the next century. Much five-axis work is still done in APT. Just a few years ago it was all
done in APT, because the graphics apps couldn't handle the math
Sure, I understand. There are lots of things that can't be done with OpenCASCADE on actual state, and even worst, there are still a few resilient bugs (some boolean operations, etc.)
Not to be mean but I can't see it being good for that, even. For small delicate work you want a very high-speed spindle and good rigidity and accuracy and repeatability. Little parts have details in the .0001"s. If the best your machine can repeat is .0005" (with luck) then you're going to be disappointed. That kind of accuracy isn't free. That little thing is so flimsy Ahnuld could crush it in one hand like a beer can. I can't see it being useful for anything at all.
Maybe for simpler shapes without too complex and/or too many curves it could work. I'm still more biased to think about some retrofit including some heavy milling machine, and so on... mainly because any real CNC, even if it is a many years old second-hand, is way expensive in my country. And I'm not even talking about BridgePort style machines. Anything with the "CNC" acronym costs many thousands here, even if it can't work without an investment of many thousands more!
Oh, well... time to eat something!
All the best,