Building an Internal Cycloidal Robotic Actuator

this video is sponsored by PCB way I like building robots so naturally I spend a lot of time designing robotic actuators and this latest design is what I'm calling the optimal actuator design because apart from being fast and strong its most important feature is within you see the gearbox is located in the center of the motor which makes the actuator super compact think two in-one shampoo you get the same benefits for Less space designing an actuator this way meant that all the parts had to be custom made including the motor which was a process that was quite interesting so anyway let's build an internal cycloidal robotic actuator like most things in the engineering field this idea was done by MIT first if you look closely at the actuators on the MIT mini cheetah you'll notice that there's a planetary gearbox in the center of the motor which is exactly what I plan to do except I want to use a cyto gearbox but more on that later first let's design the motor this is going to be a quasi direct drive actuator which means that the actuator will have a low gear reduction so that it acts like a direct drive actuator that is by having high speed and high efficiency the caveat here is that because of the low gear reduction quasi direct drive actuators need to use a high torque motor like this Eagle power brushless motor that I used in my previous actuator design this motor being an OutRunner motor has a high torque density simply by way of having a large gap radius I want to design my motor similarly so let's take this apart and see what's inside if you boil it down a brushless motor is essentially just copper wire and magnets this is the rotor which is made of permanent magnets called poles this motor has 40 magnets or 40 poles the stator has coils of copper wire that are called slots which make up the three phases of the motor this motor has 36 coils or 36 slots if you remember from physics class a coil of copper wire can create a magnetic field when current flows through it this is called an electromagnet or solenoid the stator of the motor itself is made up of laminated silicone steel sheets stacked on top of each other with thin insulation in between this is a key component to the motor design steel being an iron alloy is ferromagnetic which means that it becomes magnetized in the presence of an external magnetic field with this Ferris core the magnetic field of the coils is magnified which makes the motor more powerful this same Advantage is applied to the rotor as well the ring on the rotor is made from steel which magnifies the magnetic field of the permanent magnets now the reason that the stator is made from stack sheets instead of just a chunk of Steel like this is because stacking the sheets breaks up the Eddie currents which prevents energy loss from heating these Eddy currents are caused by the changing magnetic field of the coils another important thing about brushless motors is that on any motor the total number of slots is always a multiple of three because the motor has three phases and the total number of poles is always a multiple of two because you can only have two poles North and South that would make this the most basic brushless motor that you can make if you were to spin a brushless motor and monitor it on an oscilloscope you would find that each phase outputs voltage in a sine wave with each phase being offset by 120° this is also known as three-phase AC I actually tried this using an oscilloscope at my University you can only see two waves because I had to use the third phase of the motor as Gra round so you'll just have to trust me on the third one since a motor can also act as a generator if you have two brushless motors you can make one drive the other which is pretty neat and with that let's build the motor that we're going to use for the actuator I want the motor to have 40 poles and 36 slots which is the same as the eagle power so I bought a similar size Stater from AliExpress this one is 81 mm in diameter the plan is to eventually machine the rotor out of steel but for proof of concept I just 3D printed it for now I bought these cheap 10x 5x2 mm magnets from Amazon and I'm going to glue them to the wer in alternating polarity using some JB Weld epoxy something that I found out while gluing the magnets is that JB Weld is 11% steel which means that it should actually enhance the motor's performance though probably not by a lot I'm going to be using 22 gauge enameled copper wire to wind the Stater these windings are done in a specific pattern depending on your slot and pole configuration I use this website to figure out out my winding pattern which I'll link below I got about 17 turns on each slot and side note winding this stator took several hours to complete and most definitely gave me arthritis if you ever plan on doing this my recommendation is that you don't the phases of the motor can be connected in two different ways in Star configuration the phases all share a neutral point in Delta configuration the phases are essentially in series I'm going with star configuration because it looks like that's what the eagle power uses now after spending all that time winding the Stater I found out that it would be just too small for the actuator design that I had in mind so there goes 5 hours of my life anyway I decided to buy a bigger Stater that is 100 mm in diameter as well as being compatible with my design it should give me more torque this time I decided to use 26 gauge coer wire which is much thinner and should get me more turns per slot since the wire is very thin I decided to divide the roll into six mini rolls and wind the Stater with six strands of wire I did six turns of the six strand wire on each slot making 36 turns per slot doing it this way also made the process slightly less painful I also 3D printed a bigger rotor and used slightly bigger but still cheap 10x 5x 3 mm magnets with a bigger rotor diameter I decided to use 42 magnets instead of 40 magnets making a 36n 4 42p configuration and after winding this stator history repeated itself I realized that I was going to have to fit a gearbox in the center of the Stater which could not be possible since there were strands of copper wire overlapping the center which created just enough of an inconvenience for nothing to fit in snug now on a more positive note while making this video I bought a new 3D printer Pro tip if you you ever find yourself designing parts that don't exist to make your printer work it may be time for a new one I bought a bamboo lab A1 and it's awesome they aren't paying me to say this I just really like this printer or maybe it's just that my old printer was so bad that anything slightly better is an upgrade that's besides the point this printer can determine the flow rates of different filaments which gives really smooth prints it's also extremely fast but in my opinion the best feature is this sick tune that it plays after each print quick FYI while editing this video I got an email saying that this printer had been recalled due to a cable issue in the heated bed isn't that just absolutely fantastic anyway back to the video I figured that the best plan of action was to sit down and design the actuator so that I knew exactly how to build the motor a big part of this process was designing the gearbox as mentioned before I'm going to be using a cyclo drive or as I call it wobble Drive and since the gearbox is going to be baked into the motor design I'm calling this an internal cyal actuator the main body of the gearbox is a fixed ring made up of these roller pins you then have this eccentric shaft which is off center so that the cyal disc can roll over each roller pin to create the gear reduction the gear reduction is then captured by this output shaft and the gear reduction is equal to the number of loes on the cyal disc here there are nine roller pins and eight loes on the disc making an 8 to1 gear reduction now there is one problem since the disc is off center the gearbox is imbalanced which could lead to lots of vibration imagine you held a weight in one of your hands and try to spin to balance yourself out you would need to add a weight to your other hand but it doesn't matter because you'll be depressed either way the same applies to the cycal gearbox not the depression but the counterbalance if we just add on another disc that's 180° out of phase with the original disc then the gearbox should be perfectly balanced as all things should be in my opinion cyal drives are quite interesting the shape of the disc is made by Rolling a smaller Circle onto a bigger Circle and tracing out the path of one of the points on the smaller Circle to do this in CAD I just used equation driven curves another interesting thing about cyal drives is that the dis maintains contact with each roller pin at all times which means virtually zero backlash this is something that can't be said for planetary gear sets best of all is that with cyoa gearboxes you can make pretty much any gear reduction that you want in the same amount of space now just like the example gearbox I want the reduction of my actuator to be 8 to1 since quasi direct drive actuators generally have reductions under 10:1 I started by designing a cyclo gearbox to fit in the center of the Stater and then worked on building a prototype making this design was a bit tricky I had to go through several iterations to get it right the main goal was figuring out the smallest screws bearings and spacers that I could use without compromising the actuator strength I genuinely spent weeks getting the designed to this point but once everything was built it worked pretty well I'm also glad to see that it's back drivable which was one of my main concerns with the design as planned it fits right into the center of the stator now that I know that the gearbox works it's time to get some parts machine for the final actuator which brings us to today's sponsor PCB way PCB way is an online service for all of your Machining needs from 3D printing to CNC Machining to sheet metal fabrication injection molding and of course PCB making I wanted to get the rotor of my motor and the fixed ring of my gearbox CNC machined and after uploading my files to the site I could choose among a variety of materials that PCB way has to offer I ended up choosing mild steel 1045 for the rotor and aluminum 6061 for the fixed ring after that I received a quote almost instantly and then placed my order and got my Parts 10 days later I mean just check out how nice these came out here's the fixed ring and here's the rotor don't let your access to Machining hold you back for making your projects head on over to PCB way and take a look at their multitude of resources to figure out what's best for you you can check out PCB way today by going to the link in my description all right so now that we have the metal Parts it's time to assemble everything I first work on the Stater I'm going to be gluing the fixed ring to the inside of the stator using Loctite 648 retaining compound it's going to be a press fit but I may have overcompensated a bit so I'm going to need to sand it down this isn't the right way to go about it but a file and sandpaper is all I had on hand eventually it fit and I added Loctite to prevent it from being Twisted out of line once that's dried it's time to coil the Stater hopefully for the last time but knowing myself probably not once once again I'm using six strands of wire to do six turns on each slot which is essentially 36 turns per slot now it's time to work on the rotor this time I'm using really strong 10x 5x 3 mm n52 grade magnets these aren't your regular fridge magnets like before I'm gluing the magnets to the rotor with JB Weld in alternating polarity I have this magnetic viewing film that allows you to visualize a magnetic field I can use this to check if any of the magnets are weak here you can see the difference between the fields of the 3D printed rotor and the steel rotor the magnets on the steel rotor have a higher magnetic field concentration pointing inwards which means more flux less speed higher torque and higher efficiency this is why DC motors have a flux ring seeing something that I designed become realized in metal for the first time was quite amazing with the rotor and Stater built I think it's a good time to build a mock motor because there's a good chance that I winded the Stater incorrectly to test it out I'm using an ESC and a speed controller knob and to my great surprise it actually worked by the way that clip that you saw from the beginning of the video was the inertia of the motor causing a short which ended up burning the microcontroller that I was using as a 5volt regulator this is your reminder to always clamp down high-speed spinning objects I also went ahead and determined the KV rating for the motor KV is essentially the rate of RPM per volt of a brushless motor the eagle power is rated for 90 KV which means that it'll spin at 900 RPM if 10 volts is applied to it for the test setup I have the eagle power motor spinning my motor at 300 RPM I'll then read the RMS voltage of one of the phases on an oscilloscope after doing some calculations I found that the KV rating for the motor was 104 which is great to make sure that I did the test properly I also found the experimental KB of the eagle power which I calculated to be 108 which is not 90 as the manufacturer claims but it's close enough to know that my testing has some credibility from this testing it appears that both Motors are rated about the same and finally it's time to assemble the actuator this ler is going to be part of the actuator housing so I'm mounting the two together I added threaded inserts to the top and bottom of the actuator housing to give me different mounting options now it's time to build the gearbox which will essentially be the same process as building the Prototype gearbox the Ecentric shaft mounts directly to the rotor I'm then boring out these m3x 5mm spacers to distance the bearings it's a lot easier to build the gearbox when the spacers don't have threads building the gearbox is essentially a game of stacking with really tight tolerances the rotor needs to fit into the housing but because of the magnets it wants to lock onto the Stater eventually though things do line up and it just pops in on the bottom of the rotor is an 8x2 mm incoder magnet that's going to be read by an odrive S1 FOC controller that has an onboard encoder think of this controller as a wait to turn a brushless motor into a Servo motor except along with position control you also have velocity and torque control FOC controllers are essential in robotics because they allow you to dampen or stiffen the actuator in software which leads to more Dynamic robots and with that we have a one-of-a-kind internal cycloidal actuator now let's see just how well this performs [Music] is [Music] [Music] [Music] testing went great until I heard this so I opened up the actuator and it appeared that the coils got really hot especially while lifting the weights and that caused the plastic warp which threw the rotor off center I also realized that I forgot to add any sort of lubricant which made matters worse now let's talk specs and results the actuator is approximately 125 mm in diameter 84 mm in length and 1,23 G in weight the air gap of the motor is 1.23 mm this is the gap between the rotor magnets and the Stater the smaller this Gap is the more flux you can get the phase resistance of the motor is 75 milliohms and the phase inductance is 41.0 five microhenry now in terms of Motor Performance the top speed on this thing is about 209 RPM at 22.2 volt the max torque output is about 16.17 n m for reference the planetary actuator that I made with the eagle power had a 9:1 reduction and a Max torque output of 16.36% as I mentioned before the real benefit of this design is compactness the internal cyclo actuator is about .9 in shorter than the planetary actuator and if you ask me .9 in is a lot I'd go as far as to say that9 in is more than enough the backlash on the actuator was a lot more than expected I think I could fix it if I worked on the tolerances more now at the end of the day I don't actually plan on using this actuator for future projects while this may be the optimal actuator it isn't necessarily a practical one at least not this version it costs almost $400 to make which is a lot more expensive than the eagle power actuator however in the future I do want to make an even smaller version with the 81 mm diameter Stater fully out of metal now that would be awesome this project also gave me a great idea for a future quadrupedal robot design for me this is the best part about engineering you never really know if the thing that you make now will inspire something even better later if you want to see those future projects be sure to subscribe to the channel this project is fully open source and you can download the cad and bom at the link in my description that's all for this video thanks for watching and see you in the next [Music] one oh

Channel: Aaed Musa

Views: 512,413

Keywords:

Id: GakFB8Tdd98

Length: 19min 0sec (1140 seconds)

Published: Wed Feb 14 2024