What's new
What's new

Robot ignorance, have some questions

macgyver

Stainless
Joined
Aug 2, 2012
Location
Pittsburg, KS
My background is cnc routers mostly with experience with programming and setting up 5 axis machines as well.

I have had an idea to put a router spindle on a robot for awhile but really have no idea how robots are programmed and set up. My main question; is there a cam software that I could use to program one like a router or mill?

Are there certain brands or controls of robots that would make this easier to accomplish?

I am mainly thinking about it as it seems there are used machines on the market that are a much lower asking price than say a 5 axis router and they would also be much less expensive to move and get in the shop.

Maybe this should be the highest concern, but what kind of tolerance or accuracy should I expect to get from a machine that is in good condition?

I ran across an add for one 1.5 hrs away for pretty cheap that got me thinking about it again.

Thanks for any help,
Jason
 
We are doing this now for a customer using a Fanuc robot w/15kW 20,000 rpm spindle. They selected a software CAM package that takes care of translating from part geometry to robot code. We are not yet that familiar with it but will be soon.

The other option is to convert the controls to CNC using something like “Run my Robot” on an 840D Siemens controller. This is a superior solution from a controls standpoint in that you have true coordinate system based adaptive tuning built in for better path integrity and accuracy while also being able to program in Cartesian X, Y, Z, A, B, C coordinates while the CNC manages transformation to joint coordinates automatically and seamlessly for the end user. This allows you to use your CAM package of choice which simplifies a lot of tasks you take for granted like tool management, optimized rotary axis paths, cable windup, etc. This is most commonly done with Kuka and Comau robots because of superior stiffness and accuracy although Kuka quickly fell out of favor when acquired by the Chinese recently.
 
A lot of CAM softwares can support this. MasterCAM has a plugin for machining with a robot arm.

I suspect one of the major issues is that the accuracy of movement on robot arms is really poor. At least relative to a traditional machine tool. It must work well for some applications though, as it seems to be gaining popularity for machining foam and plastics.
 
I had a Fanuc 420i that was part of a NASA program to 3D print plastic buildings on the moon (it did it here, not on the moon). The resolution wasn't great, like 1/16" or so, but repeatability was good. It was also done after that robot ran auto production welding for 15 years.

The thing you need to realize is when you get a 2 ton robot that can move at 180 degrees a second with a 350 lb payload not only can it cut parts from sheet with a router it can also take out walls, injure and kill people. You need a very large dedicated safe area for a large robot.
 
I have a 420 iw robot and router head if you are interested mastercam will program it. I used the teach pendant to cut doors windows and lock bolts 5 k for everything I got
Don


Sent from my iPhone using Tapatalk Pro
 
They have virtual light curtains easy to set up robot is pretty easy to teach also
Don


Sent from my iPhone using Tapatalk Pro
 
,,,Maybe this should be the highest concern, but what kind of tolerance or accuracy should I expect to get from a machine that is in good condition?
Jason
Robots not great on positioning under cutting loads. It's rotational joint and moment arm thing.
Smaller robots better at position but not as stiff so tradeoffs
My first deep conversation on such in the early 80 with a guy I met on the bus to IMTS, He was all in and trying to build a company with spindle in the hand of a robot. (no pc cam back then)
We spent a good four hours not going to the show but sitting at the hot dog stand talking about the how and whats and drawing stuff on napkins. Truth told for me better time spent than walking the floor.
Depends on what you are making. Going to sculpt a face bust from a scan and this may be the best way. Making a milling cutter, crankshaft or bone screw not such a great idea.
Bob
 
If you can live with something like +- 1/8" and are cutting something soft (foam, fiberglass, wood) then with a decent bit of struggle you'll be able to do this. It works better if you're making a pile of the same thing or have an on-the-head feedback mechanism, because robots are very repeatable, but they're absolute garbage for accurate. There are also systems for doing a global calibration map for the robot, then pre-deforming all your paths to account for the global calibration map, but they're the cost of a used 5 axis router by themselves because you'll need a system capable of measuring in volumetric space to significantly better accuracy than you're trying to hold. So laser interferometry or one of the very high end creaform style scanners.

For a one-off, if you want CNC router tolerances buy a used CNC router. It'll go a lot easier and it'll still be worth something in five years if you go to sell it.
 
If you can live with something like +- 1/8" and are cutting something soft (foam, fiberglass, wood) then with a decent bit of struggle you'll be able to do this. It works better if you're making a pile of the same thing or have an on-the-head feedback mechanism, because robots are very repeatable, but they're absolute garbage for accurate.

Our Fanuc M20 will position to .08mm and the LR Mate to .01mm ( that's .0004" in merican).
 
Our Fanuc M20 will position to .08mm and the LR Mate to .01mm ( that's .0004" in merican).
Under varying load?
Say like holding a spindle and entering a cut?
Put a .0001 DTI on the end of your LR Mate and push on the end of the arm a tad. I think your indicator won't have enough travel.
Free space vs dynamic loading. With that in mind one thing that can be done somewhat reasonable using a robot is spot drilling.
Most think of a robot as a arm type device, gantry style ones begin to approach or even pass what would be a low cost router type unit but the rotations or ability to "sculpt" is lost.
Technically a 3-5-6-8 axis machining center is just a robot with a spindle.
Is a Hexapod a machine tool or a robot?

Conventional type of center base, arm and rotational joints. ... not so good for stiffness out on the end no matter how good they will position and repeat in free space.
Bob
 
I guess it depends on what you have collision detection set to, or if turned off.
Turned off the robot will increase servo power until it reaches the position, or goes into overload.
 
I guess it depends on what you have collision detection set to, or if turned off.
Turned off the robot will increase servo power until it reaches the position, or goes into overload.
Why am I gonna guess that you have never tried to do this.
The alarm will not go off as the servos are in position, the end of the arm is not, the cpu thinks all is fine.
Put that DTI out there and see. It is not the servos or feed feedback, it is mechanical not seen by the control system.
If the computer can not see it it can not fix it. That is the big problem.

This kind of the holy grail in part machining for the last 40-50 years.
If you can nail it you are going to be a very rich man.
Bob
 
Why am I gonna guess that you have never tried to do this.
The alarm will not go off as the servos are in position, the end of the arm is not, the cpu thinks all is fine.
Put that DTI out there and see. It is not the servos or feed feedback, it is mechanical not seen by the control system.
Bob

We unload (4) 48" levels with a Fanuc.
They had such a high polar moment of inertia that when moved too fast, they would cause the robot to over travel, then the arm would try to back track to reacheive the position.At the right frequency, the motions would sync the arm motion change with the motion of the parts springing back.
It would get into a feedback loop and literally fling the parts out of the eoat.
 
Robots not great on positioning under cutting loads. It's rotational joint and moment arm thing.
Smaller robots better at position but not as stiff so tradeoffs
My first deep conversation on such in the early 80 with a guy I met on the bus to IMTS, He was all in and trying to build a company with spindle in the hand of a robot. (no pc cam back then)
We spent a good four hours not going to the show but sitting at the hot dog stand talking about the how and whats and drawing stuff on napkins. Truth told for me better time spent than walking the floor.
Depends on what you are making. Going to sculpt a face bust from a scan and this may be the best way. Making a milling cutter, crankshaft or bone screw not such a great idea.
Bob

Ingersol Milling Machine developed the octahedral hexapod milling machine. The idea was to be able to construct a portable milling machine that was capable of a machining a part that would fit in the machine base foot print. Setup time was to be only several hours once on site.

The design has most of the elements in tension, eliminating many of the vibration and positioning issues that a traditional robot arm has. This comes at the cost of needing a substantial PLC to handle the math calculations required for positioning. This is probably overkill for a simple router in wood or foam but will do metal very well.
 
I agree a weak and spindly robot arm can be forced out of position and that doesn’t change the cpu assessment of the eoa position.
 
Having integrated a number of LR Mates - I can categorically state that accuracy expectations of anything better than +/- 0.010 inches on a repeated basis are a pipe dream. AND - any hope for accuracy on path is completely dependent upon speed of move. This is with no crashing, no abuse, just day in and day out positioning when making 1000's of parts a day.

 








 
Back
Top