How does a fully configured Linuxcnc system compare to industrial controls.

Dbsharp wrote:

I have only ever used linuxcnc or similar on cnc machines and really don't know how well they compare to "real" industrial controls.  I'm getting a Fadal with the older 88 controller, and will upgrade to the 527 control or linuxcnc it.  Since this is no longer a hobby, but how I make a living (design/ prototyping/ machining), I value my time highly.  For the cost of the 527 plus upgrades, I could pay myself for quite a few hours configuring linuxcnc.

• • • 
      I'd really like to hear from the community opinions how Linuxcnc compares currently to the industrial controls. 
      
      

My Fagor 8055TC is linuxcnc, I also heard Hydenhain uses it, if that is true. I just finished a Fadal VMC conversion, which was already converted once to an obscure control. I scratch build common rail Diesel test equipment, and we make our own user interfaces and code the microcontrollers, wire up everything, build the entire machine. Unfortunately most of that does not translate to linuxcnc. We did however manage to implement timed events and write some of our sub-routines in C for the gearchange and mpg.

IT IS NOT for the FAINT of heart though, I am 9 weeks into my conversion and finally found out tonight how to use the Hal scope to set my PID correctly with NO overshoots, and no following error, well not much of one that is. We use PID on our CR test benches, and we also have a chart/scope to tune those in with no overshoots, so I understand what I need to see.

Documentation is an issue, not that there isn't any, you will simply find it all overwhelming and find that simple things like compiling files and what M64/M65 really do that is so simple will simply drive you nuts, because little bits you need are well just not in simple terms.

I spent countless hours in Fadal wiring schematics trying to find out all the connections for things like the 1100-1 J1 connector pinout, then searching for parts like a brekout board to din cable for it, J9-12 connectors and pinouts. Toolchanger will cause some headaches, RS274NGC will be a great enemy until you befriend it.

Reality, is this it was not necessarily hard but required perseverance, and some help from the guys on this forum, and mucho document searching.

Link to my conversion with the toolchange operational:

Thanks for all the replies. 

msrdiesel, If I choose this route I would love to see your setup as reference, if you want to share it. I just jumped into the "RS274NGC" section to setup a router to measure its tools automatically... things didn't quite work with the toolchange.ngc reference code I found here so I re-wrote it (and will post it once cleaned up).

footpetaljones, I am interested in learning more about the trajectory planner issue.  I get what you say about others following the work and having employees not be able to work on something, but since I am a single man operation and will probably remain that way for a while, I am not sure that matters much to me.  This is why I picked the Fadal, because I can do all the fixing myself cheaply.  I make most of my money on engineering, and often dont make as much on machining jobs I pick up, but I really enjoy the shop time and having the capability opens doors for design projects. 

Bari,
I have been doing mostly consumer good engineering, but have been looking to branch out into this area.  I have an EE I work with on a lot of projects who has done lots of firmware, automation, and controls. Between the two of us, I think there isn't much we couldn't do. I'm interested to hear how would you appoach this today if you were starting out? 

Regardless, I will run the machine for a while on the original control to get it mechanically proven out and gain some experience.

Dont fix what isnt broken. As a regular joe blow fanuc cnc/robot guy, the biggest linuxcnc pros that impact me on a regular basis
- file size over 512kb
- python interpreter (too bad mastercam doesnt post python code though)
- freedom

pros for fanuc cnc on a daily basis
- gui, they arent nice but they take care of business, easy to navigate in a way
- lack of freedom , makes thing easier in the jail known as work
- lack of accountability in exchange for money, "just call fanuc"
- all that other shit that money gets you, its so much easier and nicer than linuxcnc most of the time

So it depends on what your doing. For me, at my job, it wouldnt be an advantage. But my job is to be just another cog in the machine. If I were one of those people with freedom, I might think differently.

When Im home from work, its a different story. Nothing will do but linuxcnc, as it is the only thing that can do what I want.

footpetaljones wrote:

20 years and those machines haven't become orphans

Tech Support for my 20 year old machines
- ebay
- "man thats old, there's no one that worked here that long ago"
- " thats really old, I can source you a replacement [entire machine] "

My take. I have Lcnc on both my lathes, a larger Cincinnati and a Hardinge Hnc. They both had dead controls, they both make good parts. The Lcnc works as well or better then the original control. That being said, it is a ton of work and a HUGE learning curve to get the machines reliable. I have an old Matsuura with an Alzheimer’s Yasnac control , it’s gonna get a retro this year. That being said also, it’s gonna be proved on the bench before one wire is removed from a functioning machine I use it daily to make my living. I have two other mills, Tree’s with Dynapath controls. They are weird, but they work.

As for your Fadal, I owned a Fadal for 25 years. The control is easy, it always works. Lcnc has a few short comings, mostly in the tool table, it’s awful. I would look into the NexGen control , David Decaussin, who was the D in Fadal sells it. The control is based on a Galil card. It’s fast and bolt on. He is a nice guy.

Don’t get me wrong, I do like the Lcnc . The learning curve was s steep and until you get it down , it’s a struggle. For making a living and having only one machine , stick with an OEM that bolts on. If you used say, a Mitsubishi, you would have similar issues.

Get a cheap machine with a brain dead control and figure it out in spare time. If you use this Fadal to feed your self, go for bolt on reliability.

JR1050 wrote:

 Lcnc has a few short comings, mostly in the tool table, it’s awful. 

What is it that you dislike so much? Perhaps it can easily be made better. 
It is possible (but should probably be made easier) to switch to the conventional lathe T-for-tool-change paradigm. 
There is also the option to use a database for tool management, but I don't know of any ready-sliced implementations of that other than the demo:
github.com/LinuxCNC/linuxcnc/tree/2.9/configs/sim/axis/db_demo

The main issues I keep running into are:
The tool table isnt pinned to a tab, you can have multiple Instances open, which has caused a crash. That is my fault, but it’s easy to do.
It has no increment offset, adjustments to an offset have to be algebraically adjusted and manually inserted which leads to the next point.
There are multiple buttons, reload, rewrite ect. If they are not clicked in the correct order , the offset doesn’t get updated.Why have them all, seems like “ write “ would do.
There are no wear offsets, on a mill, not an issue, on a lathe, it’s a big deal as this is how you size your part.

In my case, I run a one man tool shop. Anything that makes things easier is a big plus. I’m used to being able to tweak an offset on the fly. In all fairness, I have a Tree vmc with a Dynapath 20, it doesn’t have any way to tweak an offset quick either, it’s great iron and if you are into conversational programming, it’s the bomb. I may or may not retrofit it. It’s goofy, but reliable.

I am for the most part very pleased with Lcnc and appreciate all the effort that everyone puts in and I don’t mean or intend to sound or be ungrateful. It just seems to me that most of the effort put into GUI’s are about making them sexy, not so much functional.
I am hoping a block of time opens up for me shortly, I’d like to attack the tool table issue.
For what it is worth , there is a guy who came up with a more standard feed hold , like Fanuc, Mitsubishi ect. It stops when you push the button, it would be nice to see that implemented. There is another thread also, it got over my head quick, but the idea was this guy had rewritten trajectory planner and claimed it reacted like a modern Fanuc. That’s cool, again, by the 2 page , it was past me.