@@bc8010 Not any machine, but it can certainly do work on much larger objects than any similarly sized CNC machine, since the object doesn't have to fit _inside_ the machine.
I love the IRB6400! I spent so much time working on these and the older S4 and S4C controllers. We didn't have the USB function, and we had to get really creative when writing our programs to optimize them for space. Great work on this! As always be safe! These robots have the ability to get up to 6m/s! 😵
It's an amazing machine 😁. will do so, haven't gone above 150mm/s yet and milling usually happens at 30-40mm/s so not that much speed but you'll never know with these machines of course.
This is so cool! My first job after university was ABB robotics and I developed the SW controlling the electrical drives for the S4C controller. However, the IRB6000 (later rebranded to 6400 when relaunched with the S4 controller) was designed for spotwelding and outperformed all competition for 1 inch, 2 inch and 5 inch movements between spots. The main challenge was the compact gear boxes with natural frequency far below the frequency of the spotweld moves. Just cutting EPS is a simple job for this great machine!
You gave this robot a second life, that's great! It's so precise, can't believe its repeatability is so fine! Once upon a time, I've designed and made my own desktop cnc mill - I've implemented my university graduation work 10 years after I finished my study. It was an interesting experience. Best regards, Max.
This would be PERFECT for making fiberglass molds! In the yacht industry we have massive 5-axis CNCs that we only cut dense foam on to create a "plug" the plug is then sanded and polished to a mirror finish (with high build primer "duratech"). Then we wax the plug, spray with gel coat, and spray chop strand fiberglass on. From there you lay up additional layers of fiberglass, and once you've pulled the foam and fiberglass apart you have a mold! The foam you cut here is standard styrofoam or what we call 1lb foam. We cut 10lb and 40lb foam as well. For the biggest parts we use the 1lb foam, and do a rough cut (undersized by 1/4") with a huge endmill, then we apply a layer of thick 2 part epoxy putty to the rough cut foam, then we use a smaller ball endmill to do the finish pass. You have an awesome opportunity to start making sweet carbon or fiberglass parts!
Adding a 7th axis (linear track for the robot to move along) gives you a lot of scope for such things. Some yacht builders (there are also specialist plug makers) are doing exactly that using robots.
Isn't 40lb foam incredibly expensive to buy. And I assume heavy. I'm in the process of saving up for a cnc machine to make foam plugs. A robot arm would be killer because of its massive z axis
@@crazyg74 Believe it or not is not all that helpful. Because with a mold you’re still constrained by draft so the part can actually release from the mold.
@@CoolAsFreya I meant in terms of overdoing task at hand, and its capabilities. Not about precision. Simple 3D mill with couple nema 17 and aluminium extrusions would cut this foam just as easily. Hell, even a 3D printed robotic arm would do the trick and cost 10 times less at least.
@@zonemyparkour they may just be using foam for testing and may use other materials later. There is a very good chance that a 3D printed robot arm won’t be as precise as this unless you spend a lot on proper gearboxes but you still have to deal with the arm flexing.
@@zonemyparkour true but I don’t believe the hole point of that video is to make the cheapest and easiest way. In my opinion it’s to show that he get it running and can program the Maschine. He can use that Maschine for things much more than a 3d printed home mad arme is capable.
Very nice video, I have to say I thought a robot arm of this size would of been louder, it's very impressive how silent it is. Nice precision as well, I assume all the joints/gears are harmonic?
Thanks man, I was also surprised by the sound (or rather lack of it) it made. You can barely hear it when standing next to it. Yes the gearboxes used in this robot are harmonic. This robot inspired me to design and 3D-print a harmonic drive, just to understand it a bit better 😜 ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-1jqYdFrZqFI.html
@@SetKat-Alex That's really interesting, thanks for the info, I never get to be around robots, i've actually always wondered which ones are louder and quieter. I have some (somewhat affordable) harmonic drives that are FOC driven BLDC motors and they are surprisingly quiet, actually dead silent, it's always impressed me.
@@maxgood42 at full speed, you can only hear the motor drive, but not the reducer at all. We tested an IRB 460 with a 25kg box (that's nothing for the robot), at full speed and 2g Accel, it really is silent. (i think we only heard the controller fans, as tbe door was open). Maybe I'll post a video of the testings if I've some time.
In Fusion 360 if you haven't, I would look at the smoothing feature within most toolpaths. It limits point density in the program. Makes it easier for slower controllers to keep up with the program. I would do a test with smoothing off and smoothing on set to a value that you deem reasonable and compare the cycle times.
Yes i have experimented with this feature and it reduces the files size by a lot. Also changed the default resolution from 0.1mm to 1mm which is also a huge improvement and barely noticeable on the larger parts, might turn even turn it up for larger parts. I also set the lead in and lead out distances to 0.0, these are arcs and arcs have a lot of points so this is also a huge improvement.
Hi Brian, superb show case of how we can give new purpose to these old machines! We did sort of the same for a irb6400r200-2.8 s4c controller. We’re in the mids of turning it into a large format 3D printer. If you need some info contact me. I can explain more in detail wich components we used for storage and battery upgrades. Also we would love to come in contact with like minded people from around the globe! Regards Marcel
Hey Marcel, thank you for your comment. Would love to get in touch with you. You can send me an email with your information. Email adres can be found on my RU-vid channel or my website (see description of this video).
Hi, me and a friend recently became the owners of two IRB6600 robots with IRC5 controllers. We are both finishing our engineering degrees this year so we put the robots in storage for time being but we plan on restoring and recommissioning them early next year. We are still a bit unsure about a few things and we have a lot to learn. It would be nice to be in contact with people who have experience and who would be eager to answer a question now and then. Would you please let me know how I can get in contact with you if you don’t mind.
I have an irb 6400 with an s4c+ controller that I am trying to 3d print with. I would love to connect and see how you are doing it. I have not gotten much past getting the thing running yet.
Wow, easily the coolest thing I’ve seen on YT this year! Congrats on scoring it and mad props for getting it running again! (And holy moly, 16 *micron* repeatability?!? What kind of encoders does it have? (Even with the encoder on the input shaft and a high ratio on the harmonic drive, it seems insane to me that it can have that level of precision out at the end of the arm, after going through all the joints. Crazy tech!) (And I have no need whatever for such a thing let alone space to put it but I’m still intensely envious 😁)
Thank you very much 😁 i tested the repeatability by going to a far point, making sure all axis moved, and then returning to the test point. Did this about 15 times and 0.016 was the difference between the min and max. I also have no idea how they did this, I occasionally design and 3D-print different type of actuators myself and even getting it to mm level is very hard. Hats of to ABB for making something like this 😉👍
36000 lines encoder is pretty normal stuff, 10x interpolation gives 360000, 360/360000 => 0.001°, which is about 0.017mm/m, about the accuracy that thing has. Also largely depends on whether it's like a knee joint or fully extended in rotation, with knee joint nearly straight, very little change in distance vs encoder angle. Put like a Haimer probe on that thing and test full 3D repeatability ;-) Or ball bar test, put like a ball bearing on a table like in 4:35 and dial indicator into the robot and look for circular error
ABB uses analog resolvers to determine motor shaft angles. The signals from each resolver are evaluated by the Serial Measurement Board (SMB). The SMB also counts motor revolutions and sends the data to the motion controller. This calculates lots of stuff to compensate for mechanical things like flexing or vibration of the robot arm, so you can hit the bulls eye at low speed as well as full throttle.
Dito、epic。Something emotional about sprucing up a retired beast of a robot and bringing it back into service。No doubt its much more capable than we think。
Awesome, the accuracy is impressive to say the robots probably got quit a few thousand hours of use previously. I worked and installed large industrial laser cutting machines back in 2000’s , LDV /BYSRONIC , kinda lost touch now but think it’s nice to know there’s people like yourself having a go . 🇬🇧👍. Lots of respect
This is a great project! Much respect! Since I work a lot with CNC machines, I would consider this robot good enough for softer materials. With metal on the other hand you could likely run into problems with shattering.
That robot seems in really good condition, we have one very similar at work, in production since 1998. There is a between step Z0 instead of fine or z1. Z1 might round some corners, z0 should not, but neither stop at the point like fine. Nice job with the program conversion !!
@@BrianBrocken Maybe it's only available since S4C+ but you can build your own! Declare a variable of type zonedata in program with a zone like 0.001mm. It's self-explanatory on the teach pendent. Or you edit the BASE.SYS and add something like "CONST zonedata z0 := [FALSE, 0.001, 0.001, 1, 0.001, 1, 0.1];"
Admirable skills. Hat off. You give that huge robot a second life, another brain. Thank you for entertaining us and... of course instructing us as well Brian. I subscribed to your wonderful channel
The 6400 series are built like a tank. I programmed and operated 2 of these at work since 2000. They ran at 3-4 meters/second lifting up to 90 pound parts 24 hours per day, 7 days per week with little issues. The would pace parts on a machine spindle with +/- .0025" accuracy. The motion sensitivity makes them relatively safe to work around, and they have welding software for oscillation. We used Schunk grippers on ours.
I bought an old VMC with a punched tape reader. It also had a serial port so I wrote a drip feeder that looks surprisingly similar to yours. But I too had the issue where the controller would only take one command at a time so it had no ability to "merge" the path between commands so it would execute every line of code and stop then start processing the next line. Never did figure out how to fix it or if it was even fixable. Sadly I had to sell it at a loss. Great machine.
Total cost? I've always wanted a kuka arm. Something about the complex movement and all the joints makes them feel organic and alive when they move autonomously.
I knew those robots were strong, but I didn't realize how fucking accurate they were! That repeatability test was .0006 for all my Inch people. That's a lot for a dedicated cnc machine on linear guide ways and with ball screws, but with a big dangly arm like that?!? That is amazing.
Yeah, those don't work unfortunately (yet). Bought them at the auction for 50 euro. One has a fried EPROM and the other is missing some boards. These are the projects for when I have a lot of spare time (won't happen very soon😬)
Subscribed, which is a bad idea. I have a complete Fanuc M16i RJ3 cell that I bought with a project like this in mind. I decided to part it out when I realized much work the project would be. But maybe...
This is great for all processes that don't produce a significant load on the tool, because robots arms are not stiff enough to take these without bending. If you take a 20mm endmill and chew into steel, it will be chatter hell. But for foam and 3D printing... awesome!
i remembered after graduating university, i worked as an engineer in a local sheet metal stamping factory. the CNC turret punch still uses floppy disc for the program..we had to transfer the files from a crappy pc into the disk and hand the disk over to the operator.
The entire concept of Skynet coming out of some US military program is ridiculous. This is how Skynet took it's first tentative steps towards extinguishing all human life. And I get to say, I was there at the beginning. I hope your proud of yourself.
De meeste geluiden die het frezen gaven lieten me haren rechtop staan als verspaningstechnoloog. Maar hee wat is dit vet zeg, volgens mijn een flink partijtje moeite in gestoken om het een beetje te laten werken. Super knap!
Ik heb de verkeerde lijm gebruikt om de platen op elkaar te lijmen, deze was redelijk dik en harde volledig uit. Dit zijn de geluiden die je hoort. In zuivere isomo is dit geluid veel beter 😜
This is super cool. What'd you pay for the robot, if you don't mind me asking, and how much time did it take in total from getting it, to having it work?
I can't believe I just watched a 2-ton robot arm carrying out an intricate task with such precision that it makes me writing my name freehand with a pen look like a Neanderthal using a rock! UGG!!! BASH!!! UGG!!! 😳 😄👍
Incredible work! I've got a 6400 foundry incoming to my workshop and it's going to become an art installation before it's retired to join my other cncs cutting things. Would really love to pick your brains about the whole thing at some point if that's ok? 🙂
Great video 👍 The robot arm is the best way to do CNC,if the memory restriction could be resolved it would be amazing fast.Edge precision has the same restriction on his Mazak.
Cool stuff. I started out using a Motoman SK arm that still used floppy and serial communication. But I remember the user interface (forget what the term was) was really intuitive.