Visit brilliant.org/MichaelRechtin/ to get started learning STEM for free, and the first 200 people will get 20% off their annual premium subscription.
This video was sponsored by Brilliant
Thank you for watching the video! Links to the CAD for this project are below as well as links to the hardware needed to assemble it.
CAD Files:
drive.google.com/drive/folder...
Follow me for more updates on projects:
Instagram: instagram.com/michaelrechtinyt/
Twitter: / rechtin_michael
Hardware List (Affiliate Links):
20x27x4 Bearing: amzn.to/3PkoUrF
3x10x4 Bearing: amzn.to/3PqHsXd
45x58x7 Bearing: amzn.to/3v0mnee
5010 Brushless Motor: amzn.to/3cto6lJ
M3 Locknuts: amzn.to/3oe4AMv
M3x16 Cap Screw: amzn.to/3B6UJQm
M3x6 Cap Screw: amzn.to/3IQVzmc
Small addendum to the video. The final torque value I gave was correct but the values shown on screen were wrong. The load cell was measuring weight so the maximum weight seen was about 9500g not newtons. Sorry about that.
Links to stuff I use (affiliate links):
AnyCubic Kobra Go: bit.ly/3quyQnL
Creality CR-10S: amzn.to/3BuAiM2
Ender 3 S1 Plus: amzn.to/3xv7zW8
Elegoo Mars 3: amzn.to/3UwVvh1
Creality LD002H: amzn.to/3QXhpqJ
Creality Wash Station: amzn.to/3S1NTBi
Visit brilliant.org/MichaelRechtin/ to get started learning STEM for free, and the first 200 people will get 20% off their annual premium subscription.
Don't use white lithium grease on plastic parts it will eat away at the plastic parts over periods of time rather use a silicone base grease or grease that safe on plastic.
UPGRADE TO CURA 5.10 PLEASE
Would love to see it attached to something like a scissor lift to observe how much useful work something like this can actually achieve. Robotics is interesting, but there is an entire world of more common workloads which require something along these lines. As such, something like a scissor lift is applicable to a variety of related workloads, such as solar trackers or simply lifting a load.
Make a screwdriver 9,5Nm should be enough
Hey mate, I haven't read through all the replies but this might have already been covered. Yes, your test rig had issues but I wouldn't be worried about inertia so much. First off, before testing again be aware of your limits, if the gearbox is the limit then that's fine, it'll just break. If the GBox holds up then your current draw will max out in an instant and you'll blow an esc or a lipo, whichever has the lowest current handling capacity, that can get expensive, dangerous, both. Maybe throw a cheap automotive fuse inline. I'd lift a load, essentially you just make a winch and keep adding weight until you can't. You just make the drum a known diameter and you can get meaningful results.
The problem with your torque measurement now, is that because of the high velocity you are coupling a significant amount of intertia. The motor inertia might seem very small, but because of the gearbox it is amplifier by n^2. None of the cheap brushless motors actually quote rotor inertia, but generally its somewhere around 30 gcm^2. That means that with your 64x reduction gearbox, you are dealing with an inertia at the output of about 0.0123 kgm^2 (note the change in units). Guesstimating the speed of the output shaft, its probably around 0.3 rotations per second, 1.9 rad/s. This results in an angular kinetic energy of approximately 0.0217 J. Thats quite a lot of energy stored, especially if you want to try to slow it down in a very short period of time. Point being is that you are generally measuring much more than just gearbox output torque. Torque testing should indeed be done at zero speed. Also, are you aware that there are specially wound BLDC motors meant for slow speed operation? These quadcopter motors are optimized to work in a constant high velocity, but the ones optimized for slow speed are much more well suited for most robotic and servo tasks. You can easily find relatively cheap ones if you search for 'gimball BLDC' and similar search terms.
My thoughts exactly. Plus the rotating gear stages add their own inertial energy on top.
that is true for real torque measurements. But as a max torque measurement it is ok. It doesn't matter where the load comes from, but you can say, that the max output torque for the transmission is at least 9.5Nm without destroying the transmission.
@@thrownchance I disagree with you, shock loading is inherently different than static loading. Torques might seem torques independent of where they come from, but you are disregarding the concept of energy and how it is spread over time. The only thing that can be concluded is that a (single) peak of 9.5Nm will not destroy the transmission. I put 'single' in brackets, as you might exceed the yield stress and start plastic deformation or fractures, especially in the infill. Only a proper fatigue test would be able to tell you whether or not it can actually deal with these peaks without damage. An example for the difference between shock and static loads with materials might make more sense. A relatively stiff material (such as regular PLA) will have much higher peak stresses in the material, which can lead to broken parts. Something less stiff like PETG bends more, hence spreading the same impact energy over a longer time thus lowering peak stresses, which means it might not break for the same impact as PLA. It however should not be concluded that PETG can take larger torques as these are (again) shock loads. It is very likely that for the example materials PLA can actually carry a larger maximum torque due to the higher yield stress.
Are you sure about the unit conversions? joules aren't a big unit, so 0.02 joules sounds small. As in, it's the inertia of a baseball that is moving at 2km/h (equivalent of falling a whopping 1.5 centimeters) Definitely shouldn't be able to break the arm on its own, but I guess it could influence the result somewhat.
@@alexbv475 Yes I am sure of the unit conversions. As a double check: rotor inertia is 30g cm^2 -> 0.03kg cm^2 -> 3e-6 kg m^2. Adding the n=64 gear ratio: 3e-6 * 64^2 = 0.012 kg m^2. Rotational kinetic energy is 0.5*I*omega^2 = 0.5*0.012*1.9^2 = 0.022. It might not sound like much compared to your example, but a baseball is 150gr, quite heavy. If you drop this from 1.5cm onto a loadcell, it too would produce high forces. The key problem is that you have something that you drop onto something very stiff. This means that a lot of energy has to be dissipated in very short time period, and thus peak forces are quite high. A pillar in a carpark can sustain tremendous loads, but if you smack it with a hammer, which has relatively low amounts of energy, the high peak stresses will still allow you to do damage.
Great design. You are measuring 9500N of force by the load cell however, your load cell has a max capacity of 200N!
???
Really cool design, thanks for sharing! If it is accurate, that amount of torque out of a 3D printed gearbox is wild. +1 for adding an encoder for better low-speed control.
As you have concluded, measuring torque while using an ESC won't really tell you much. There are some very nice alternatives to ODrive now that would work well with the size of motor you have. In particular: the Tinymovr (R5) and Moteus controllers. They are much smaller than an ODrive, come with the encoder built in, and mount directly behind the motor.
Literally cheaper to buy the equipment to design and build your own driver with an STM32
Wow, you could get an entire stepper with more torque and a much higher RPM than his motor+gearbox for less money than the >$100 controllers you mentioned. Seems like hacking an ESC that supports position feedback would be a much better idea
You could try testing how much weight the gear box will lift at given ratios. Knowing that would help determine what kinds of applications you might be able to build this into. Long Term speed test for heat and wear would be good too.
Isn't that what torque is?
No the problem is the motor is too fast. Unlike a theoretically optimum electric motor, it can't produce 100% of torque at zero speed. It can only produce torque well while moving quickly, he explained that. A sensored motor is the solution.
The major blunder in the test comes from the fact that when the shaft is stopped in the load cell, the motor stops, thus we don't det the stall torque but instead the torque necessary to annihilate the angular momentum of the motor's hub, shft, planets holders, planetary geers, ... everything that spins gets some angular momentum to annihilate with a counter torque if you want it to stop
There is no "stall torque" to unsensored brushless motors. Just like normal internal combustion motors - they only produce torque when rotating at a decent speed. Controllers use back-EMF to recognize the phase switchpoints and when the rotor is stationary - there is no back EMF. ESC sends some impulses to start the rotation, but they are random and weak, it just twitches around. This was perfectly explained AND demonstrated in the video, by the way.
This gearbox is one of the better ones i've seen on YT! Good job. Btw since you can't do a static torque test, it would be nice to see what weight it could lift at a given radius. For example, build a simple drum with 50mm radius with a string attatched to some weights. Then you could try different weights and speeds. Also, this would enable dynamic endurance testing which would be really interesting to see. And, maybe even try different filaments. Maybe nylon or petg as gears?
What I find most impressive about this quadcopter is the potential it has for DIY enthusiasts and makers. With the ability to design and print your own quadcopter without having to purchase expensive parts, it opens up a whole new world of possibilities for those interested in robotics, engineering, and design.
Absolutely mind blowing. One of the simplest gear boxes I've ever seen. Thanks for sharing :)
Wow, thanks. One or two weeks ago I designed planetary gearbox for 5010 myself but I didn't get to assembly. It's nice to come across same idea, u saved me couple of hours of tweaking. Thumbs up!
These would be perfect for hub motors in a build. I could definitely see this being extremely beneficial.
That 3d printer is amazing. I've never seen print quality/ such a small bead from the nozzle in a hobbyist's printer.
A truly masterful design
This looks amazing.
This is really cool design. Prints pretty easy too. Fantastic work.
Thats a really cool design, the rotating planet core that is just one free spinning stack of gears is quite fascinating. -1 for the torque test, because the peak is highly influenced by the inertia
Cool design, excellent explaining!
Says it feels very smooth as the GB siezes up
It would be neat to see it taken to the extreme where the final output has such an incredibly high reduction that it can be embedded in something unmovable and still function. I remember seeing a similar setup where the output was embedded in concrete but it will functioned because the extreme reduction summed up to basically no movement on the out put. Wonder how many reduction gears you'd need to achieve something similar.
TY, This is exactly what i was looking for
great and simple one, easy to mount....thank you! Ill try using it for a powered cart for hauling stuff around the garden
amazing!
A quick way to electronically reduce the speed and increase your motor's torque is to desolder the connectioms to the three phases ( at the stator) and rewire them in a wye configuration if they are currently wired as delta. This will cause current to enter one phase and be forced to energize an additional coil on its way back out of the motor. The scope of this mod is to reconfigure the phase connections from parallel to series. The motor will have more energized coils at any given moment, thus increasing its torque. I've used this trick before with good success and my motors live longer and run cooler. Keep in mind that you will decrease the maximum speed. That is the tradeoff.
Nice design. with a few slight modification in particular of the end piece you could make them stackable and destackable without requiring reassembly of the individual modules.
Great video, as always!
Nice grinding gears sound!
This guy needs more subs! Awesome content
Nice job designing and printing ! To really measure the torque, you could use a pulley as an end shaft, and try to wind a cord around the pulley while refraining the wire with one of those luggage weighting hooked, until you stop the motor. So you can get rid of the starting couple & the inertia 😉
Great design, I am in the process of automating my pool, and these gearboxes will be great for automating the valves, e.g. I can automate the flow to bypass the solar heater if the pool temp is too high, or remotely control flow distribution from pool to spa, etc. I'll probably be using steppers and arduino IOT. Thanks for your community spirit.
That's really awesome
Nice work!
Extremely interesting content, stellar editing and format. Subscribed.
I'd be really interested to see a similar thing but with different combinations of gearbox modules with different reduction ratios. With this, you could design a clutch mechanism that disengages the gearbox modules so you can get a variable gearbox out of it.
use simplefoc to control the driver at low speeds, even without encoders. it works great, and can be upgraded to encoders. there is even somewhere a board which lets you do a control by amperage
The stm b-g431b-esc1 is capable of doing basically everything, but it is a hassle to work with.
This has a great application for a trolling or outboard motor for boat. At least that's what I'm investigating
I think you can reduce a lot of the sound problem if you build conic gears instead of plain ones, they are more efficient in high speeds too.
Very good project ! Great
The going brushless bit floored me🤣🤣🤣
that is brilliant! those would work well with stepper motors
+1 for integrating a sensor and odrive in a future video. I'd love to see what kind of torque is possible to get out of these low-end BLDC's with 3d printed gearboxes. This is a great video. Keep up the excellent work!
yeah it would be great to see how far this could go!! keep up the good work !
cool!! thank you!
Amazing dude! SUBSCRIBED
Very cool man. I really like how that came out, and the modular setup. It'd be interesting to see how big of a fish it could reel in.
Just mount the arm with the load cell to the gearbox case and mount a longer shaft to the output shaft, so you can grab it with a glove to apply a breaking force to the output shaft. Now you can do continuous torque measurements at different rpm's without the influence of inertia!
coooool! good job 👍
Another idea may be using different motors on this gearbox (Brushed DC, Stepper, etc), and perhaps making different mounting plates, but the core+output being a “universal gearbox”of sorts.
The Jim Lahey clip earned my subscription Sir. 👍
Love to see this ..converted to magnet gear box
I love this idea!
This is just a wonderful piece of engineering. Thank you for sharing this. Have you considered expanding the systems with levels of other ratios? So have a level of x3 or x2 or x5. Then you can stack them to get to any final ratio you want.
INSANELY ACCURATE LINIER ACTUATORS
Great design👍
That's really cool! The only problem i can imagine is for every stage added you get more friction, but i guess you could just beef up the motor used if so.
Waw, really amazing ! No go and build a robot arm with this !
you got me with they go wrrr
well now i finally understand what is torque, thanks
It would be very good to see a metal sheet laser cut version to be a fast and cheap option to produce. Great project.
I see any Mr Lahey reference I hit that like button.
Newly found your channel while searching for an alternative to the ODrive since they're going closed source now... Wonderful channel you got here, and wonderful stackable gearbox! Have you done a split-ring planetary gear at all?
Planetary gearsets are 3-way adapters for gears. This design uses 1 of them as housing, and the other 2 change ratios like a car transmission. An old-fashioned manual transmission could do this more simply, or an Allison transmission is more elegant. Or add another motor (it can generate too, right?) and just 1 planetary gearset and you have a hybrid transmission. All 3 of those are modular designed to change while the vehicle is in motion, without having to stop and bolt in other parts. But this is a good learning exhibition - a launchpad to talk about transmission designs. Thank you for making the video.
Very neat
Really dumb idea considering how fast a BLDC motor can go, but what if you connect the motor to the output to backdrive the whole planetary system (i.e. instead of 64:1, make it 1:64). As GLaDOS would say: "Speedy thing goes in, speedy thing comes out". Doubt the small motor could keep up, but would be fun to see how fast you can make a thing spin on 1:1, 1:4, 1:16 and so on and see where the point of diminishing returns kicks in.
Dude you’re so funny, the edits caught me off guard a couple times hahhaa 😂
Man I wanna make this on a industrial level, compact modular motor it could be revolutionary on repair and replacement in the field
this is AMAZING MEN !!!!!!!!!!!!!! 🤩🤩🤩🤩🤩🤩
The most impressive part of this video isn't the gearbox, but the fact he used a bone stock CR10 to do it!
Great design, I would be interested in the backlash of the gear box. If it is almost backlash free, I could use the design for a 4th axis in my CNC machine
Nice dude.
Used your amazon affiliate to the materials for this build. Thank you for this amazing build.
im actually going to pick the files up and see if i can put them in any of our machines to see how well it holds up against actual mechanical force compared to raw movement will take a while but ill let ya know
You were asking in the video for better methods of measuring the torque capabilities. Automotive actuators are tested as following: PULL OUT TORQUE: Mount a weight on a lever that hangs straight down before the test. The motor then tries to raise the lever as high as it can. The further the motor turns, the further the weight is away from the vertical axis and the more counter torque it will see. The limit is called the pull out torque, which represents the counter torque it can handle during movement. PULL IN TORQUE: Mount a weight on a rope winding and lift the weight with the motor. The weight does not change over time (if you ignore the diameter of the winding growing with the rope winding up), but by making the weight heavier you'll find the limit of torque that can be applied from still stand. The pull in torque is also a nice automated lifetime test if you want to measure the durability. The gears will be worn only in one direction (counter torque does not change direction). Take an Arduino to generate a servo signal for the ESC, this should automate it easily. Be aware that you also test the capabilities of the ESC: their behavior when detecting blocking is a pure software thing.
you could make a simple dynamometer by introducing a clutch between the output and the load cell so that way you would get it's dynamic torque.
Creative video, thanks for sharing :)
depending on rotation speed, the 16:1 would most likely work as a very heavy duty Power Wheels Brushless motor conversion/gearbox kit!
Clever!
Cool project! I’m a hobby machinist, so I’m super curious about what the measurable differences are between machined tool steel gears and box which are hardened, and the printed version. Power, load, speed, noise, cost, and reliability come to mind. I have no opinion; it would just be interesting to have data that guides when, for example, plastics, steel, or hybrid designs are optimal choices.
You could try using a disk brake off of a motorcycle or even a bicycle to create a torque dyno to get more accurate torque readings at any speed.
I think what's missing is the ability's to hot swap out the reduction. Maybe a inner shaft that engages only at selected ratios controlled by an outside sliding mechanism, or the rings could twist lock into place with a manual switch to unlock a retractable back stop where said rings bottom out on the previous one.
0:52 "+/-/Magic?" Love the humor! Subscribed! 😂😂
you need to put this on an RC car Next 😁
Nice project! Would be great, if you could design a rotary encoder step on the output side for automatic backlash compensation
is it really necessary to have the outer shell have gear teeth on the wall? maybe it makes sense for stabilization but it feels like youre adding unnecessary friction. i would model the walls to lock with the motors so the motors stay in place but the gears down scaling your ratio could benefit from removing that extra friction happening with connecting with the wall gear teeth... just a thought
Loving the How Ridiculous reference :)
I've been using very similar looking motors in my RC tank. Easily capable of ripping the 3D printed tracks apart...
This gearbox could be scaled up and used with a 40v lawn mower motor for a DIY ebike conversion. Motor input speed is 3000 rpm. The gear box output speed will need to be 300 rpm. The final drive reduction to 70 rpm can be achieved with a 12 tooth to 52 tooth chain ring on the pedals.
"3 wires? That's kinda scray." instantly subscribed bro I love your humour
What's the current draw per gearbox added? Since you're possibly not as efficient as possible with printed gears that are slightly worn and/or metal, I'd want to know that before I built a circuit to control something like this, or code to drive it
You could make a shredder and use the gearbox for the shredder axis to see if the torque is sufficient to crush plastic parts (it could aldo be a cool thing to watch)
Another idea could be using different gear profiles, but an otherwise identical format. Could maybe try and get a series going akin to that “Design a Fan Blade” series although I don’t know how to build that hype i guess! Herringbone Gears are the ultimate if I remember correctly, although there is probably some downsides i am forgetting.
Among other things mounting the gears requires split gears, twice as many housing parts, requires more dexterity to put together; not insurmountable... Much quieter, more torque, many advantages!
Perfect for a jack or precision moves
Nice work. What's the weight of the first stage including motor, and each additional stage?
it makes me unconfortable how good your ender 3 prints
Mount it to a mountain bike brake disc and then add the load cell to the caliper. Will allow you to measure the torque while it's spinning
3:44 very smooth indeed
that is fucking cool as hell. you earned a sub
Applications: Mechatronics, like robotic grabbers, RC crawlers... You could make some wild automated kitchenaid mixers, or dough machines, cutters/log splitters.
You have too few subscribers, this is an engineering masterpiece.
Congratulations! you made another screwdriver)
You should try to control the bldc motor with an odrive driver or similar to get more control