linux compared to centroid and acromill

I have a large 3d router running acromill. My control card is crapping out and I want to retrofit with either centroid cnc or linux cnc. The machine is running dc servos. I will be doing alot of 3d milling with the machine. I like the idea of using linux cnc but im worried about the performance of the system. Is linux a true professional control system? I have also been told to stay away from control cards without onboard dsp also. Any comments on this are surely welcome. My acromill system has worked great for a long time and i want to make sure i get at least the same level of performance.

thanks

there are many linuxcnc users around the world using it in a professional situation , yes you will have a learning curve etc , theirs no getting away from that
over all i have retrofitted more than 40 machines of one type or another , and never had a loss in performance . if anything it's been a magnitude better .

i would suggest a mesa interface for servos and their series of addon cards for interfacing servo's depending on the types fitted
or pico systems , although i personaly only have used mesa , so therefore i may be biased .

remember linuxcnc is a full performace system back to the pc as part of the loop and not a control card with a microprocessor , their is a vast difference , not only in the way it works

by all means ask questions, and their are many that will guide you .and assist

welcome to linuxcnc

sounds good. I am seriously considering linuxcnc at this point.. I am wondering if the learning curve has more to do with learning linux or getting the system running............both?? I don't mind the learning curve if im getting quality on the other end and it seems like that is the case here. I was thinking maybe the reason some people don't like the idea of not using a control card is because it gives windows too much control over the cnc and windows is evidently not as stable as linux.....would this be correct thinking?

thanks for the info.

why not just use the mill i already have? the control on it doesn't work so i may as well hook up & try it...........right? plus i need it to work with my router. I've already started downloading the iso and troubleshooting it so im pretty much committed to it at this point. I guess im just wanting someone to say.....Hey man, you're doing the right thing so rock & roll with it.

For 3D milling the standard LinuxCNC distributions are not perfect. They will work fine, but the older trajectory planner is quite slow when it is fed with a gazillon of small G1 moves unless you relax the tolerance on path following accuracy quite a bit. The new trajectory planner executes 3D contouring code a LOT faster, but is not included in the current 2.6 release as far as I know.

You can get it by installing a 2.7 development release. 2.7 and the normal LinuxCNC (2,5, 2.6) can even happily coexist on the same system.
But although I have zero significant problems with it it is not a production release yet, and you might want to ask yourself if you want to run a production machine on software that might be unstable.

That said: even if I had to pay $500 for a LinuxCNC license, I would. It is the only system that can be made to do anything I want it to do.

yeah i talked to the guys at mesa today and they told me the same thing. I definitely need it to do alot of 3d. the machine bed is 10'x5' and the z is over 30 inches. I generally run programs over 30k lines of code at over 200 ipm. So it looks like i'll need the new trajectory planner. I have an older dual processor machine i was going to start with using the 8.04 download. I am wondering now if i should build a new pc for the 10.04 and 2.7 combo......................any suggestions? The machine has been offline a while so im not waiting on it right now.....I don't mind the learning curve and I'd love to get away from windows for this application so I'm pretty much committed to linuxcnc for now unless someone tells me I'm wasting my time.........doesn't seem that way though. thanks for the help.

With many small segments, the new trajectory planner and high feedrates system load does increase. Also the preview pane becomes slow on older systems when it has to redraw 30K lines every time you pan or rotate it.

I would build a new PC. Gives you a speed advantage and old PC's do die when put in heavy use after a few years of sleeping.
You don't need bleeding edge hardware so cost of a fresh mainboard+processor+memory+video adapter is low.

I do have a small shitty video online of one of my preliminary tests with the motion of the new mill (which is still under construction, BTW): .

The code is for a real (aluminium) part and uses constant engagement/trochoidal toolpaths at high feedrates (100-200ipm). Your situation is going to be easier; feedrates are the same, but your parts are bigger (I suppose). Video is only exciting for the person who already spent hundreds of hours building the machine (me, that is), but it also shows how well the new TP processes all those short segments (there is a shot of LinuxCNC on the monitor; look at the line numbers in front of the G code). In a synthetic test I have been able to put over 1000 blocks of G-code per second through the system.