How do I choose the correct robot and software for a laser cutting robot cell?

    Hello again , I posted on this forum last year and received invaluable information. A lot has happened since and I would like to ask for your help again.

    I represent a company currently occupied in sound and light. We manufacture, fabricate and customize lamps, lighting fixtures and installation solutions.

    We are looking to automate our processes and will soon be purchasing a robotic milling cell (purchased as an all inclusive standardized cell from the vendor). In addition to this, we also want to create a (fiber) laser cutting cell as we didn't find any standardized cells that suited us. The cell will be used both for cutting sheet metals (2d cuts) and for cutting into existing products such as a pre-assembled lighting fixture (3d cuts).

    The laser system will be comprised of an IPG source and chiller and a raytools cutting head, we will probably add a rotating table of some sort as well. The gas compressor and fume extraction parts will be sourced locally. The parts will be integrated by the same integration team that will assemble the milling cell. We are in touch with our local authorities in order to understand and comply with the safety standards.

    The two main questions I have are the following:

    1. We will probably use Sprutcam as the offline programing software for the milling cell. Will it be able to handle the laser cutting too?

    2. How do I choose the correct robot for the laser cutting cell? The robot will be used/refurbished, so purchasing a customized extra precise model is not an option. This is ok as we do not require high accuracy/precision, but we still want to be as accurate and efficient as possible within our limitations. Looking at the robots used in other laser cutting cells I saw many different options such as ABB irb 1410/2600/4400, fanuc 20iB and many many more. We would prefer using a kuka model, but we are relatively flexible with this. The point is, different companies present the specifications for their products differently, making it harder for me to compare between them. Do you guys have any advice as to what I should look for in choosing the correct model other than repeatability and reach? The cutting head only weighs a few kilos but I also know the higher the load capabilities of the robot - the sturdier and more able to absorb vibrations it is, should I get a high payload robot? Should I be looking at the rotation range and speed of the axes? Are differences between models that great? The most available robot to us is the kuka kr210/240. Would this be alright do you think?

    Thank you very much

    Edited once, last by dodkipod (March 29, 2021 at 12:15 PM).

    I would be very careful about buying a used robot. Many are "retired" from plants where they have been very heavily used and may have experienced joint wear.

    Old robots can be found cheap, and might even be in good shape, but may present problems getting support or parts. For example, there are lots of KRC2s on the market, but KUKA no longer makes parts for them. And finding people experienced in working on them is becoming more difficult. So getting anything older than a KRC4 entails some risk.

    The same holds true for other brands. Getting support/parts for secondhand Fanucs can have some extra complexities due to Fanuc corporate policies.

    The other item is: how are you going to program these robots? different brands (and models/generations within those brands) have different support for CAD/CAM conversion (which is usually a paid option, to boot). So the programming toolchain and how its supported on the robot is an issue to be careful of.

    Robots, even brand new, always have certain backlash issues that can't be gotten rid of (barring special calibration with something like New River Kinematics' SARCA module). For example, trying to cut a circle with the robot will give you an oval, even if the program is mathematically perfect. And the more worn the robot is, the worse this will be. Generally speaking, smaller robots are "tighter" this way, but it's never zero. And smaller robots sacrifice reach. And moving from 2D to 3D will increase these issues by 2-5x.

    A KR210 is a good all-round robot. Better if you can get it with a KRC4 over a KRC2, for the reasons listed above. If you want to go with a KRC2, at least make certain it's a ed2005 with KSS 5.4 at minimum (5.6 would be best).

    Thank you for your reply. We are indeed looking at a kr210 with a krc4 controller. If we choose to go with a different manufacturer we will make sure to get the most recent controller possible.

    The robot we are looking at is refurbished and should be in good enough condition. Could you elaborate on the backlash issue a little? I know we wont be able to cut a perfect circle, but I was under the impression it would at least be close. We don't need utmost precision, but if the issue is extreme then it will be a problem... Its this reason why I am trying to figure out what type of robot would be best. Any advise on this will be greatly appreciated.

    I am particularly interested in what you said about cad/cam support. We intend on using sprutcam for this, do you know anything about this program? Also, kss is the kuka software installed on the controller - correct?

    Thank you again for your help.

    KR210 is quite large and rated for 210kg payloads. are you sure you need such high payload to mount laser head? yes, KSS is "Kuka System Software" which is software installed on Kuka controller. About your questions on accuracy.... why don't you define it? don't you know what is tolerable for your application and what is the size of working area? what is your "close enough"?

    1) read pinned topic: READ FIRST...

    2) if you have an issue with robot, post question in the correct forum section... do NOT contact me directly

    3) read 1 and 2

    The backlash issue is hard to quantify. Unlike a CNC machine, where each axis's backlash is purely one-dimensional, the serial kinematics of a robot arm mean that each joint's backlash multiplies into each axis "downstram" -- the error becomes more exponential than linear.

    To complicate matters, the backlash of each axis can have very different effects depending on where the robot is in its operating envelope. Take Axis 2 -- it's backlash can "flip" depending on whether the A2-A3 link is forward of vertical (gravity pulling it fowards) or "leaning back", where gravity will act in the opposite direction. Plus the counterbalance unit adds another variable into the mix.

    There are techniques to minimize the effects. For example, for 2D cutting, keep the end effector orientation fixed at all times, and keep the arm in a pose where gravity effects don't change. Try to avoid cuts where any axis reverses direction mid-cut, and use "run in/out" motions to ramp into each cut when a directional change is required.

    Quote from panic mode

    KR210 is quite large and rated for 210kg payloads. are you sure you need such high payload to mount laser head? yes, KSS is "Kuka System Software" which is software installed on Kuka controller. About your questions on accuracy.... why don't you define it? don't you know what is tolerable for your application and what is the size of working area? what is your "close enough"?

    I am sure that we don't need such a high payload, but the small payload robots are short in reach. The kr210 is just easily available to us, that's all. It could also be a different robot with a similar reach for all intents and purposes.

    Its hard to define what's tolerable to us and what isn't. "Close enough" basically means that it looks close enough. The 2d applications are meant to allow us to cut metals ourselves instead of outsourcing like we do currently. If I need to cut a circle and rectangle in order to later weld them together (manually) into a lighting fixture then close enough means that it looks like a circle and will pass as a circle when looked at. If I want to cut an odd shape in order to create an "installation solution" (sorry, don't know the term in English) that will later be part of existing infrastructure in, say, a new event hall - it just needs to have the holes be more or less the correct size and shape so that the metal workers can fit and weld it into place (we will late hang lighting/speakers from this "installation solution".

    The 3d applications will basically be cutting holes into existing products. Things that are too large for a regular laser cutting machine, or have angles. Possibly cutting decorative features into existing products as well. So say I need to cut a hole into a metal tables legs to later fit a pipe through and weld it together - I just need the holes to be "close enough" to size and shape. If the hole will be completely elliptical then yea, that's no good, but if its pretty much a circle then its fine. I can cut a slightly larger hole to compensate for inaccuracy in cases like this. If I want to cut wavy lines into a metal lamp shade as decoration - practically no precision is required.

    Does this make sense? I tried to define it as best I could.

    Quote from SkyeFire

    The backlash issue is hard to quantify. Unlike a CNC machine, where each axis's backlash is purely one-dimensional, the serial kinematics of a robot arm mean that each joint's backlash multiplies into each axis "downstram" -- the error becomes more exponential than linear.

    To complicate matters, the backlash of each axis can have very different effects depending on where the robot is in its operating envelope. Take Axis 2 -- it's backlash can "flip" depending on whether the A2-A3 link is forward of vertical (gravity pulling it fowards) or "leaning back", where gravity will act in the opposite direction. Plus the counterbalance unit adds another variable into the mix.

    There are techniques to minimize the effects. For example, for 2D cutting, keep the end effector orientation fixed at all times, and keep the arm in a pose where gravity effects don't change. Try to avoid cuts where any axis reverses direction mid-cut, and use "run in/out" motions to ramp into each cut when a directional change is required.

    Sorry but English is my second language and though it's pretty good, I don't know many of the technical terms. "run in/out" = lead in/out? The "extra pathing" added to the beginning and ending of cutting paths? If so, are you suggesting that we use this for every new direction and not only for the start - end?

    I'm particularly interested in what you said about avoiding reversing direction mid cut. This sounds like great advice - thank you. So say when cutting circles, maybe divide the cut into 4 and have a lead in+out for each of them so as to minimize the effect of reversing the direction of the axes? The bumps this will create are not an issue - we can grind them away with an angle grinder. Does this sound reasonable to you?

    SkyeFire said: For example, trying to cut a circle with the robot will give you an oval, even if the program is mathematically perfect. That's how I lost my way to build a laser cutting machine with the help of a robotic arm. Thank you Mr.SkyeFire! :thumbs_up: :smiling_face:

    I'll probably do something else. I'm going to buy an old Trumph from a friend, with his solid cast iron chassis, and I'm going to mount a fiber laser source on it. It is possible to replace the mechanics, drivers, motors, etc ..., but I use a very good bed. Any ideas?

    I don't know what a Triumph is, but I assume it's an older CNC of some kind?

    LinuxCNC is an open-source system for putting new controls on old CNCs, or for supporting the conversion of manual 3-axis mills into CNCs. It also supports 3D Printing, plasma cutters, laser cutters, and other kinematics.

    There's a series of videos on the YouTube channel "RotarySMP" that details the conversion of an old Schaublin CNC lathe to using LinuxCNC. And there's a lot of different YTers who've documented converting manual mills to CNC.

    Mach4 is not open-source, but in my limited experience, it's pretty good, and it's not terribly expensive.

    Both of these can work with a wide variety of controller boards and motors.

    Sky fire is correct, use Linux cnc, it supports stepper motors, and encoder feedback dc motors and more so the world is your oyster in that regard.

    In terms of using robots to do what your looking at it's certainly possible, but that requires things like Kuka.Cnc and a calibrated arm, but that also depends on your budget and the tolerances you require

Advertising from our partners