According to Andrew Boerder “industrial ac servos with just a few exceptions are all triphasic brushless motors.”
I see the confusion. AC -> DC -> AC. The servo drive re modulates the DC bus to variable frequency 3 phase AC.
2 Likes
It’s certainly possible but something that needs some research first. For the same load, the remaining two diode bridges will be conducting notably more current (we’d need to make sure they’re sufficiently oversized) and the ripple on the DC Link will be significantly increased. Three phase power has relatively consistent power and results in fairly low ripple even through a simple six pulse front end. Single phase power has zero power at the zero crossing, so it exhibits a large power ripple and single phase rectifiers need much more DC link capacitance. We should make sure the increased ripple doesn’t interact with the servo controllers as it can often make closed loop control unstable.
I agree, but in practice I have run many inverter based power devices on single phase which were not specked for it. I may have actually run the arm on single phase when I got it working. If you are concerned Its not hard to add capacitors.
Its fairly application specific, but this system is much more complex than a simple VFD. I’ll admit I can’t picture any damage trying it, under low accelerations and loads. But I have to believe the system will work better with the proper input power.
I’ll counter your generality by saying that I’ve spent a significant portion of the past few months modifying three phase rectifiers to run on single phase input power. Wether it works or not dependent on many implementation details. Proceed with caution.
Plastics SIG could mount a spindle on it to use for trimming plastics/composites. It would be the only 5 axis capability in the entire space. For plastics/composites work this would be a big deal.
The only hold back for these arms are the controllers, the are old and difficult to do anything truly useful with using modern software (such as Fusion 360), unfortunately the drivers for the motors are directly integrated with the old controller. I did find an open source driver for the motors on that arm, which would enable us to put a modern controller on it which in turn would enable use to use CAM software like Fusion to generate path for it.
The servos are AC with resolvers for encoders.
Then covid hit…
3 Likes
@jast, could you humbly split the topic of the robot arm from this proposal. I think both are significant enough to warrant the divide. Josh as OP isn’t concerned about a singular example piece of equipment, and Chris W and others are deserving of discussion of options for this arm. Personally, I am highly in favor of attempting to rework it, as the physicality / mechatronics of assembling that many motors consistently is the harder part, versus boards. I certainly think there is some knowledge to be gained from engaging Steve Alaniz, Jose Quinonez, Wichosky, and others who have absolutely demonstrated in the past the ability to work on prior CNC systems and determine faults, and repair/replace
3 Likes
Yeah, I managed to fix the wiring to get the unit moving but the power driver doesn’t seem to be delivering enough current at times, and we can’t get the unit to pass the power on reset stage in automatic mode at all. The controller is pretty archaic, and the vendor can’t get us any useful information for repairing and debugging it. We have the manual but it assumes you’re buying replacement parts for most things rather than troubleshooting on-board.
With new drivers it would be worth it, but that’s a notable endeavor.
We powered the unit in the 208 mode (the controller can run two arms simultaneously with some other hardware based on my research, so it did not seem to require full power rating); when machine and science were both in the south workshop we swapped the plug for the Bandsaw to run it (@TBJK I can’t recall, is the bandsaw 3ph?)
Would the controller work with 480? Mayhaps, but given the trigger conditions and that manual movement of the arm functioned fine after fixing the busted wiring, I doubt that was the cause.
The manual mode we were running it in has a far lower stall torque and operational speed. It’s not the torque that’s the problem, it’s how fast that heavy object could move when the unit is running properly. We could drive it super slow, but again custom controllers at this point.
Indeed, but we also have to consider long term upkeep of said controllers. Would it, for example, be better to invest in a Pocket NC if we just want more degrees of freedom? Getting parts for these arms will be problematic on top of maintaining the controllers.
If we did keep it, I would agree with this.
I’m not at all certain what you mean by not enough current.
Chris I tried to figure out what kind of command the power amplifiers took, but I never got that far. Is it ±10v, or does it have three inputs, one for each phase? If the servo amplifiers can be retained then the hardware from Mesa electronics should make it straightforward to replace only the control.
1 Like
One other thing that might could be used is the ethernet interface. There is software for a PC that might could control the robot. I didn’t look into it far enough but I know it is made to use a null-ethernet cable to a PC.
I can look into the actual servo boards.
@Chris_Wischkowsky do you have a link to that open source controller that you were telling me about?
The London Makerspace got them working with LinuxCNC using Mesa/STMBL
From my perspective, the robot arms are cool, but science is not doing anything with them. We are trying to get our area, neater and running smoother. We have had multiple people and groups of people tell us they planned to get the arms going but as they are nothing has really happened recently, so we don’t want them there another several months just to see nothing happen. I understand @Chris_Wischkowsky was looking into it, but we need something going.
So options that i see are…
- We sell the arms
- Some other committee that wants them takes them out of our space that can handle them.
- We see significant progress being put into getting the arms running with a planned handoff to another committee/location
@kobin is the incoming chair so really whatever he says on the issue goes.
1 Like
I don’t have much of a horse in this race (aside from wanting to use robotic arms at the 'Space), but have some thoughts I’ve not yet seen mentioned in this thread.
I visited the Innovation Lab at the Bill J. Priest Center for Economic Development several months ago. They have a variety of robotic arms, including matching Staublis. They also offer training programs on using such arms.
One staff member in particular seemed very knowledgeable on industrial automation of all sorts, especially industrial robotics. I suspect he’d be happy to help get them running, purchasing them from the 'Space (he mentioned he was aware of them and interested in acquiring them), and/or helping us explore other robotic arm solutions and training the first generation of trainers.
https://sites.google.com/view/billjpriestinnovationlab/home
After seeing the variety of robotic arms available at the Innovation Lab, I tend to think a Collaborative Robot (cobot) would be a better choice for the Space, for at least two reasons:
-
They’re designed to work alongside humans and can halt on unexpected collisions.
-
They offer easier programming (including a record feature, where the bot can be manually manipulated into various positions, and those positions replayed in sequence).
The Staubli’s are pretty cool, but they seem to be more of a project than a tool (at least, this has been the case for the last few years).
While I have no doubt the minds at the 'Space can eventually turn the Staubli’s into useful tools for the membership, would other options be worth considering instead of, or in addition to the Staublis?
1 Like
Well there’s your problem.
how do you figure? Anything it can do we can do better - 'cause we be people.
Chris said he could set up the arm to trim plastic pieces.
Since the last update, I’ve assembled a STMBL v4.1 open source servo drive and I’m ready to test it with a spare servo. If that works, then I’ll build 5 more…
2 Likes