Linux CNC going Commercial...

Hello to everybody!

With my first Post in this Forum I want to introduce myself and describe what I`m up to.

My Name is Christian, I`m 30 Years Old an I work as a Service and Application Engineer for a small german Machine Tool Builder.
We are Building and Servicing small CNC-Milling machines from 3 to 5 Axis with Spindlespeeds up to 100000 rpm.
I´ve got over 15 Years experience in Industrial CNC- Programming, and the same time in making Steel to Dust by chipping it with Milling, Turning, Grinding aso....

So this is why I am here:
I`m absolutely Sick and Tired of the most CNC-Control builders whatever they are Named... The Prices are absolutely unjustified, and the Support is much too expensive. There is no Reason in 2012 to Pay 20000,- € or more (and in Fact it is much more if you want a reliable System) for a Piece of Equipment that didnt Change its functionality for over 20 Years... The only advances that where made in the last Century by controlling Machines via G-Code are Colorful Screens with a lot of unneccessary Bulls**t...

Ok, from my Point of view this is the actual Status.

I want to build a mostly Standardized CNC-Controller for mostly standardized applications... I think, the most CNC-milling Machines today are still 3-Axis-Setups, with arround 80% and the next Setup will be 5 Axis with around 15% of the whole Market.

I`ve been searching for a while and now I decided to make my first steps with LinuxCNC. For testing i got a "HAASE AL 640" standard 3 Axis Setup Machine with simple Stepper Motors. later i will Upgrade it to a Closed-Loop Servo System. But first i want to learn how to Walk before I try to run.
Actually it is controlled by a CNC-Graf-Controller by Boenigk- Electronics.

The Controller should-

-Use selected Hardare for maximum speed and reliability (But still no Special Hardware as far as possible... Controller-Cards will be needed for sure)
-Fit in one 19" Rack- Case
-be able to control different Drive-types (Stepper, Servo etc... My Dream: Linear Drives)
-operate with standard ISO-G-Code Postprocessors,
-have a Tool Management System (Tool-Type, Diameter, Length-Offset...)
-have a Preview/Simulation
-not cost more than 5000,-€ (Retail)
-in perfect Case support SercosIII by using Bosch/rexroth Sercansiii PCI-Cards (active and passive)

My Questions:
-Is someone actually working on a similar Project?
-Who is interested in cooperation (i can just offer my experience in G-Code-Programming, my Connections to Machine-Tool-Builders, Translation from german to English an English to German and Machines and Hardware for testing)
-Do you have any advices for the selection of the Hardware (Motherboard, CPU, Graphic ?)
-Any other advices, suggestions, Ideas

Greetings
Christian

just about everything you say is covered with Linuxcnc , apart from ( perhaps ) the sercos drives , but that should not be a huge problem given some working cards etc ..
although in most I know what your saying , but support for commercial systems can at times be huge , mostly in labour and travel etc , and can be transparent to the end user .

your machine given some sensible interface such as Mesa pci interfaces as well as some others , should be an easy conversion

Linuxcnc covers up to 9 axis and 5 axis machines have been done in the past

Linuxcnc will work on reasonable low cost pc motherboards such as Atom etc , see the wiki and the forum for more information too , it does not require highend motherboards quad core etc to work very well
I usually run and build machnes using solid state drives for my retrofits ,

fitting to a 19" rack is not much of a problem
i'm currently building touch screen based controllers



Dave

The "Standard G-Code Postprocessor" outputs G-Code ...

I checked the G-Code interpreter of Linux CNC, and it looks very good at all. There are just some little changes to be done to make it compatible to compete with the Big ones like Fanuc, Siemens, Heidenhain...

The Problem is just that nearly any CAM-Software has a little "difference" in its output.
It starts from Circle-Center Coordinates (some choose Relative , some Absolute) continues with Different behavior for Toolchanges etc. and doesnt Stop with things like different Program Head and ending- Sections...

Actually that is no Problem at all... you cam ask the CAM-Software Builders to modify your Postprocessor. But nothing works without Money, and some Times you can Pay several Thousand Euros, just for an optimized Postprocessor for your machine.

I will attach some Sample Data of an unchanged Postprocessor... its just the finishing Step of a fantasy Surface.

@cncbasher:

but support for commercial systems can at times be huge , mostly in labour and travel etc , and can be transparent to the end user .

That is what i want... I dont want to sell LinuxCNC... i want to make Money by offering a Low-Cost and transparent alternative to the established CNC-Controllers. So somebody has to intergrate and Service it. When you are retrofitting Machines you know the Potential of just offering the Service...
I`m very exited of seeing Linux CNC at Work with my Machine... The Most interesting Benchmark: Time for the linechange...

Linuxcnc is based on Fanuc , so has no problem with cam systems . I use it with the likes of Solidcam / Camworks Hsmworks , and of course theirs many more
theirs no actual standard for NC controls hence why theirs so much out there , every vendor adds his own methods

cewematec wrote:

I want to build a mostly Standardized CNC-Controller for mostly standardized applications... I think, the most CNC-milling Machines today are still 3-Axis-Setups, with arround 80% and the next Setup will be 5 Axis with around 15% of the whole Market.

Christian

I might mention the Pico Systems PPMC (Parallel Port Motion Controller) an interface
to analog velocity servo amps that has a motherboard with plug-in boards for
encoder input, 16-bit DAC output and digital I/O. For more complex machines
you just add the boards as needed to get the required input and output points.
See pico-systems.com/osc2.5/catalog/index.php?cPath=1
for more info on this system. The basic package for up to 4 axes is
US $780. (Disclaimer: This is my company.)

Jon

Thanks for all the Information!
I will start in the next 2 or 3 Weeks by ordering some Hardwarecomponents and keep this threat Updated, as things happen.

Maybe my Boss will find my "Hobby activities" interesting too, so i can use some Infrastructure of our Company... I`ll talk to him the next couple of Days.

Hi Christian,

where in Germany are you located?

I am sitting in Hamburg and have some different machines (3-4 axis) running, all controlled by LinuxCNC. Most of them used commercial.
...and all driven by very different hardware.

A "standard" system would be very helpful, sometimes I am asked to do a machine conversion, but it is not always easy to find working hardware for big-sized machines.

I am using LinuxCNC since many years now, maybe I can give you some help, if needed...

best regards,

Martin.

Maybe I missed your intentions here, is your goal to build a turn key LinuxCNC based controller box to sell on the open market? If so I'd say go for it but don't expect it to be easy going. You would likely get some play from the hobbiest or DIY crowd, that is users that want to and are capable of configuring the software themselves. However from the larger machine tool community I would expect signiifcant resistance unless you can offer engineering support as an option.

Here is the thing you may see those support contracts and expensive training programs as nuts and at times I would agree. The problem is many manufacture rely upon those support contrast to keep costs contained and to have somebody to "blame" when glitches pop up. I see this all the time when engineers go to one of the large industrial automation shops in the area and buy outwardly expensive drives or controllers when similar hardware can be had for one third the cost as imports. The problem is the imports don't offer the security of a local applications engineer. You and sometimes even myself find that frustrating but when you are an overloaded engineer even a few hours saved here or there are a big deal.

What I'm trying to say is that if you are to offer such a controller, at a reasonable cost, you will have to consider support issues to gain signiifcant mind share. The fact is many engineers simply don't have the time. Think about it a bit, LinuxCNC is pretty good so why is it in limited usage? I think the simple answer is that for every engineer or business owner that has time to tinker with it, there are 10,000 that are so swamped that the wouldn't even consider it.

@Turbspeedskater

Thanks for your offer!

I am located in the southern Harz Mountains... the town is called Walkenried.

You`re right, the missing Hardware for big machines can be a Problem...

In my company we are actually working with Sercos III and after some reading and asking I think the Solution of the most Hardware-Problems will be to run Linux CNC in combination with SercosIII.

I found some interesting iNFOS for combining Linux CNC and SercosIII:

One is from the University of Stuttgart:

www.isw.uni-stuttgart.de/lehre/themen-fu...e-arbeiten/?t_id=386

another one (according to this Info the SercosIII protocoll is already open source since 2009, so it will be legal to integrate it to LinuxCNC)

www.pro-linux.de/news/1/14983/sercos-iii...veroeffentlicht.html

I also found a Linux Kernel Driver for the Sercans III Cards (Active/Passive) from Bosch Rexroth:

sercansiii.sourceforge.net/

So I think the Migration of Seros III to Linux CNC is just a matter of Programming... Unhappily in dont know how to code Software...

The actual Plan (Just a Fantasy):

-A minimum (no unneccesary Packages) installation of LinuxCNC on SSD
-A second SSD for Backup
-Somehow build a system with very low latency (Optimized for LinuxCNC)
-Standart Ethernet-Connection
-SercansIII Active- Card for Drive Control
-Multiple Parralelports for the connection of Peripherial Devices (Robots, Toolchanger, Cooling-Systems, Touch-Probes, Tool length sensors a.s.o.)
-Setup Wizzard for standard milling machines up to 5 Axis
-A controlpanel with Keyboard, Mouse and Touchscreen.

Maybe build 2 different Controller Types: One with SercosIII for new Machines and total conversions, another for retrofitting and low cost conversions if someone wants to keep his old Drives or use Stepper Motors.

I`m not sure about all that and actually I just start to work and experiment with LinuxCNC...