Who says you can’t teach an old robot new tricks? Nobody, actually. That saying is about dogs. But it applies to robots too, at least judging by the way this late-90s industrial beast was put to use in a way it was never intended: as a giant CNC router.
The machine in question is an ABB IRB6400, a six-axis, floor-mounted industrial machine that had a long career welding at a Eurorail factory in Austria before [Brian Brocken] made its acquaintance. He procured the non-working machine — no word on what he paid for it — and moved the 2-ton paperweight into his shop, itself a non-trivial endeavor. After a good scrubbing, [Brian] tried to get the machine started up. An error prevented the robot controller from booting; luckily, there’s a large community of ABB users, and [Brian] learned that one of the modules in the controller needed replacement.
After fixing that — and swapping out the controller’s long-dead backup batteries, plus replacing the original 1.44 MB floppy drive with a USB drive — he was able to bring the machine back to life. Unfortunately, the limited amount of internal memory made it difficult to use for anything complicated, so [Brian] came up with an application to stream coordinates to the controller over a serial port, allowing for unlimited operation. With that in place, plus a simple spindle mounted to the robot’s wrist with a 3D printed adapter, [Brian] was able to carve foam blocks into complex shapes. The video below shows everything from delivery to first chips — well, dust at least.
This build seems to be a significant escalation from [Brian]’s previous large-format CNC machine. He must have something interesting in mind, so stay tuned for details.
Very jealous.
Nice. I knew a guy that did the exact thing way back around 2010, commercial though so not a lot of info about. At the time I worked for a place that refurbed old robots, they were just starting up a partnership when I left to offer custom CNC robot cells.
Loved working on those early 6400s and 1400s, the quality was head and shoulders above everything else (Including the newer ABB models). Everything was way overbuilt, well balanced and modular.
We’d often get robots in with 10+ years of working life and no maintenance to speak of. Fresh oil in the gearbox, new batteries, a coat of paint and they were good as new.
Later models by comparison usually had a gearbox full of glitter and all sorts of wobbly bits.
As someone who has been shoulder deep in the posterior of many a 6400 series ABB replacing a base loom I don’t know about exactly “no maintenance.” They were decent though. Auto spot welding robots might have been a bit more abused getting into tight places though.
Certainly better than the older ASEA units we had. I didn’t work on the 6600s long enough to know how well they stood up in comparison.
Yep, that’s peak capitalism: well built machines that last. With an established reputation, businesses then push inferior versions of their machines. They are built much cheaper, need more repair/replacement, and have a much large profit margin.
Alternatively: We built robots that were way overbuilt for their projected use, hence were more expensive and less efficient than they could be so we optimised them.
Not everything is an evil capitalist plot – over-building stuff is wasteful in many ways. If every iPhone was designed to last 50 years there would not be more iPhone 1’s still running, there would just be a million tonnes more stuff in the e-waste chain rather than still in the ground.
Or it could be a trade off between the design engineers, the budget office, and the customers complaining of high initial purchase price. Some companies are willing to trade off higher buy-in prices for increased cost of ownership or maintenance intervals.
The wobbly bits and glitter in the gear boxes may also have been from the units being pressed into service for jobs that were well outside design parameters. Sales officers are notorious for making inflated claims that directly contradict what their engineers have told them.
A good example is the M5OD manual transmission used by Ford. It was pretty much bullet proof so long as you didn’t exceed the tow capacity or pop the clutch too often. Push it too hard though or try to “snatch and jerk” a load to get it rolling and you were looking at burnt clutches and blown slave cylinders. The latter being a true pain because the damn thing was internal.
I always wanted to do something like that as I watched these robots follow my commands behind fences and safety systems during my career as a controls engineer. However after all those years of drilled in safety protocols it makes me cringe to see one so free and easy in someone’s garage. There were several times I was surprised how fast they moved to the wrong place with amazing power. I wish him a safe future !
could be worse https://youtu.be/w77-ILRwSEA
What software technology used to program the ABB IRB6400?
Normally they have their own programming language – RAPID.
Yeah all the industrial robot brands have their own specific language (though they are all pretty similar fundamentally) – RAPID is turing complete (so it can do loops/logic/IO) but can also be used sorta similarly to GCode. You can program them directly on the controller, via ABB’s RobotStudio or a number of third party apps – RoboDK, Powermill Robot, Grasshopper Plugins, ROS etc. But before anyone gets too jazzed on the idea of replacing their cartesian router with a robot note these things are a lot more finicky and less robust for this kind of work. You have a chain of six rotational axes vs three semi independent ones so they are a lot less rigid and have a ~super fun~ quality of having different resonant frequencies depending on axis configurations :)) – so not great for proper machining but okay for stuff like grinding, carving foam and more point-to-point operations. Also mind the industrial robot biz is pretty slow to update so hardware compatibility can be a bit of a maze.
I used to program OTC, ABB, and Kawasaki robots to do fancy repairs for a variety of welding applications. The OTC programming language was HORRIBLE, one step above assembly. Kawasaki was a nice simplified version of C. ABB and RobotStudio was a full on simulation environment with a language similar to basic. Given the choice I’d probably only work with ABB robots.
can this controller use drip feed to exceed 15,000 code limit?
Was called a “side job” for the first 30 years of my life unless you wore a large and loud hat typically with feather(s), fur coat and occasionally a cane. Colors may have included purple or green with leopard or lion print.
Wow, something seeing that kind of precision in something so large is amazing.
On the “could be worse”, look at this robot theatre that was stopped.
https://www.svt.se/nyheter/lokalt/ost/robot-intar-teaterscenen-i-norrkoping
It’s nice to see the old guy staying busy in retirement.
Now, the question is whether that arm, being so big, has strength enough to be used for milling tougher materials: wood, aluminium?, steel???
For a moment I thought it was going to be this chap:
https://forums.lr4x4.com/topic/104487-learning-cnc-milling-first-steps/?do=findComment&comment=927134
Sadly (probably for commercial reasons) he never posted a follow-up of it in action.