Dane.Kouttron

[09.02.22]  Blue-Bot Upgrades: Vision and Hardware


So here's a quick reminder of the layout of BABOT, three theta axes, each with a brushed dc motor and servo controller

Adventures with waking up the monstrously large Blue SCARA robot of were quite good.


Some software and hardware upgrades were implemented to increase performance, and they are listed here.

    
The mechanical issues: Planning out a Harmonic Drive Replacement
So here's a quick reminder of the layout of BABOT, three theta axes, each with a brushed dc motor and servo controller

As you might imagine, any sort of issue with one prevents the robot from painting in any useful format, well as you might have guessed one of the gearboxes failed. This was fairly concerning as well, its a harmonic drive and they are both highly specialized and generally hard to replace.

Theta-0 started making grinding noises and eventually failed about 2 hours later, I was somewhat dreading this happening as its really unlikely to find 'new old stock Japanese 80's robot arm parts'


To diagnose Theta-0 I opted to open up Theta-1 (a working gearbox).

I took off the motor assembly and the harmonic is visible looking down-into the robot vs up-underneath. Fortunately the motor & encoder both have screw-terminal connectors so the whole assembly can sit aside on a table, vs hanging off the side of the bot. Shown far right is looking directly down into the robot.



The motor, which sits in the cup of the harmonic gearbox has an eccentric 'wiggler' which is what drives the harmonic procession. As the motor rotates the wave-shaper proceedes around to deform the central spline. The fit between the two is really tight, so its important to remove the wave-shaper from the gearbox while keeping the two concentric. This resulted in a lot of 'hands' prying with screwdrivers around the mating flange that connected the robot to the motor. Given the number of unknowns with this robot, i took it slow, but eventually it wiggled off.



Looking into the Theta-1 drive you can see the deformable 'cup' gear as well as the outer ring gear. The two are connected to opposing sides of the robot, with the ring gear tied to the upper part of the robot arm (towards the base) and the flexible spline coupling connects to the lower part of the arm (towards the end-effector). Without the wave-shaper they jusssst clear each other, so nominally you can move them without the gears munching on each other, but just barely. The beauty of this design is opposing sides of the same gear are 'in contact' at the same time, so you have incredibly low backlash.



Time to inspect the two and get a good idea for how to repair the harder-to-reach gearbox. So, the ring gear actually has some interesting design aspects.My first thought was: How do you possibly pull this out? Its not like a gear-puller would fit in there and even if it did it would destroy the tiny cutesy teeth. There are actually 3 threaded positions on the ring gear. im fairly sure its for pulling out the ring gear. By threading in 3 M5 loooong cap head screws and using them to back out the ring gear, slowly, incrementally it came back out.



Ok so here's what i was talking about, shown in the documented version of the ring-gear photo, there's 3 positions that are threaded and three positions that are un-threaded. For removal, bolts go into the three threaded positions and the process of threading them in pushes against the robot frame. As the ring gear is alligned with three installed pins, this is a slow process as it needs to stay concentric. During the re-install bolts thread into the robot hull through the non-tapped clearance M5 holes.



Time to re-install. Note that the harmonic gearbox had a lot of goopy hardened grease, the process of inspecting the teeth resulted in it being cleaned and as a result it needed more vitamin grease. For this I used automotive wheel bearing grease, as it was available and seemed to be 'about right' in terms of grease-ness.


The mechanical issues: Tearing down the sad gearbox
Here's the theta-0 motor. Note the zip-ties are holding in a cap for one of the motor brushes. The motor assembly has a recessed circular interface such that it stays concentric with the robot 's theta 0 gearbox. Eight M3 screws hold this whole beast in place. Note there is no gearing or anything inside this assembly, its all big ol' DC-Servo. The base has a concentrically mounted optical encoder as well.



After removing the motor the first thing I noticed is the theta-0 harmonic drive is frigging huge in comparison to theta-1. Keep in mind the disassembly is upside-down, and inside the robot. Nominally i could have lifted the whole robot up, into the air and rotated it about, but that became an increasingly difficult option. So upside-down it is. I began by using a 3/8 impact driver to loosen the 6 M6 bolts that kept the harmonic flexible spline in place. Note you can kinda see the 'its sheared apart' just from looking underneath.



Ok, better photo of the sad harmonic gearbox. The flexible spline is only held in place by the burrs generated when it sheared.



Here's the removed motor for theta-0. There was a lot of wiggling as it is upside down, and inside the robot cavity.



A first look at the greasy sad gearbox. This was kinda unexpected. The flexible spline coupling sheared right at the interface between the mate to the robot and the flexible bit. I was honestly more afraid that all the teeth sheared off, given the noise it was making.

Next up: what material is this? Some digging online indicated that the flexible splines generally are a magnetic-stainless steel (400 series). MMM, its weld-able....



How are the teeth holding up? To check the status of the teeth, I wiped down all the gooey old grease and cleaned each tooth with a.... Dental pick. Surprisingly, they were indeed all in tact. The failure could have been fatigue driven and at some point the repetitive stress probably tore/sheared off the thin walled part. Here's where it gets interesting.



So, given that there did not appear to be any real interference between the outside of the flexible harmonic, there didn't really stand a reason to indicate that it wouldn't suffer too much from a weldment. My only hold-up was that this is a flexure. Flexures are delicate and changing the material properties could end poorly.

Now to figure out how to hold everything concentric to attempt a weldment. If i wanted this to work again I would need to make sure it was nearly perfectly concentric, and the jig to hold it like that would need to be made out of something that could survive the welding process without sticking to it, so no plastics  / 3d printed parts.




I was conveniently bringing some waterjet parts towards WPI one weekend and was fortunate to have an insert turned to hold both parts concentric. The insert started as a 4" 6061 aluminum round. A large draw-bar nut was used to keep the part under compression.

The thread for the nut was single-point tapped by Thomas Kouttron [link].



Here's the new mandrel that's going to be my welding jig. Having the sheared parts on hand allowed for a fairly snug fit. After the photo (below) was taken, threads were added for ~1/2-13 so a capturing nut could squeeze and hold the whole thing together.



The broken harmonic drive with and without the mandrel:



So here's the jig holding everything in place ready for welding. I was somewhat concerned that the tig welding would be really difficult as the aluminum mandrel is a comically large heatsink, but the small relief angle was enough of an air gap that things were not too bad. 




As it turned out the insert was a bit short so a steel round was used to prop the teeth up off the table. If the flexure was sitting on the table, weld current would flow through and nominally cause pitting. I scoured the mating surface with a flexible sandpaper  block and then proceeded to Acetone-wash the surfaces before welding.



I was super anxious about welding this part as there was little room for error (a stray arc on the teeth would be really hard to correct). The welding actually was quite quick. I used a stainless filler rod and cranked up the cover gas flow. Ok i went a little overboard with filler rod. The big issue was the incredible thickness difference between the top and flexure. I started at the thick mount surface and made my way down in small bursts to the thin walled flexure, trying not to chew it up.



Here it is, pre-cleanup. I was fairly careful about thermal buildup, but in the welding environment it was difficult to see any discoloration. this fixture also did not have a channel for a back cover-gas. Stainless generally needs a cover-gas for both sides for it to remain stainless, if I were to do this again, I'd try and boring out a path for rear cover gas. Otherwise it came off the jig without too much struggle. Its 'quite structural'  Given that this was initially (probably) hardened, i used a file to check if it still was, on a face right above the gear teeth. The file didn't catch in, so that's a good sign!











(There's other photos in the photo gallery)
Concluding Remarks:

If you have questions or comments, ask below or send over an email.
Comments:
HTML Comment Box is loading comments...


Dane.Kouttron
Rensselaer Polytechnic Institute 
Electrical & Electrical Power
631.978.1650