The Differential Drive - A New Breed of Actuator

I designed, built, and tested a unique actuator I'm calling the differential drive. It utilizes two identical motors, and the output rotation speed is the sum of the motors individual rotation speeds (with an additional overall reduction). It's something I've never seen before!
Gain access to project files and support my work directly - / levijanssen
Join me on Discord - / discord
Follow me on Instagram - / leviajanssen
GoPro Hero 9 - amzn.to/3rq4wZq
Rode Wireless GO and Lavalier Go - amzn.to/2QDwx2O
Disclaimer: I may earn an affiliate commission when items are bought through these links.

[@flagship1701e]

At the end of the day, every engineer's true desire: " It does look good and it is shiny".

[@witoldgrabowski9263]

Levi, you're an awesome man. One day, you'll save us all. (hopefully, you'll long be done by then, with these dark and awkward DARPA projects) !

[@dvkerns]

Brilliant. Really fun to watch, and it does look cool as hell. Edit tip for next time: drop the bg music, it’s not adding anything or making the video itself any better. Regardless you’re an instant subscribe and I look forward to the next.

[@holgerh.1580]

The principle in a Vario transmission from Fendt, which has been around since 1995, is similar, I think. There are no 2 motors, but the power from the motor is divided into 2 paths and combined again. This achieves what you want to achieve.

[@tduffin2010]

"differential Drive" patented by JBL in 1994. But yours looks neat.

[@kellyaquinastom]

Search for differential drive jbl patent "1994" -loudspeaker turns up nothing. Any links please?

[@2blkSSs]

@@kellyaquinastom I found a list of JBL patents and none of them sound remotely close to what he’s talking about...

[@2blkSSs]

Are you trolling or what? And why does this comment have 20 up votes? Haha do you have 20 different account?

[@tduffin2010]

@@kellyaquinastom literally typing "JBL differential drive patent" into google - the very first result is a white paper on their patent. IF you can't find it, you are banned from using a computer forever.

[@iandrake4683]

I think we've all been where you are.

[@AOZMONSTER]

But since the speed is the sum of the two motors, if you ran both motors at the same speed, wouldn't you have single motor stall torque at two motor speeds?

[@frivolas]

Sorry but i just saw 19min of you talking. Can we have another video of the actual mechanism? Did you build a mathematical model for this mechanism?

[@robertphillips9017]

You just described the “transmission” of the Prius, except that the Prius uses three input motors, mg1, mg2, and ICE.

[@johnmcfarlane9416]

Exactly. Ford/Toyota eCVT. I made one based off the linear toyota variant to dynamically share output from a windmill with two generators and one motor, one gen for low winds and the other geared for high wind. The motor/braking takes some power off the gens but it speed sets the distribution between the gens. Cost effective, no; Fun yes.

[@leonschmidt7]

Check out the Fendt Vario or AGCO CVT transmissions

[@felixar90]

I knew I had seen this before but couldn't remember where. Hybrid cars. That's where. Also some CVT were there's an electric motor who's sole purpose is to vary the gear ratio. I think modern tanks use something like this.

[@HelloNotMe9999]

Not precisely... A combiner gearbox, which is what the Prius has, is a little different. Similar, yes, but not _quite_ the same. It's nice to see him make something to test his theories though.

[@webmonkees]

Another gearing to study the history of the Subaru CVT transmission,, they were one of the first to try.. www.subaruoutback.org/attachments/cvt-diagram-jpg.26499/

[@senjaz]

They are used in tracked vehicles and are called adder/subtractor gear boxes. Essentially two differentials linked together. They allow one high torque motor to provide the main motive force to two outputs and a second lower torque motor to add and subtract its speed to those outputs. It allows precise control over the relative speeds of the outputs even with motors of similar torque.

[@572089]

and with 3 differentials and you can use the "steering motor" to move torque from one diff to the other via the central, allowing for a constatant total output, but varying difference between the two outputs. Thats how you get fine control on racing tanks like the Ripsaw. this way you dont lose speed and dont put too much backlash on the steering motor at high speed while maintaining steering reaction time.

[@freescape08]

That's a very well known application, and I didn't even think of it. Nice.

[@BrokeWrench]

Can you mention some vehicles that use this? I had looked up this exact thing and found nothing

[@joemaser9122]

@@BrokeWrench cat differential steering

[@nou5440]

cant like cause 69

[@mandernachluca3774]

Little fun fact: The Toyota Prius is using a planetary differential to couple an electric generator, motor (and ice engine) and the driveshaft. This system is widely belived to be the most cost effective and efficient way to build a hybrid drive train. In german, we call this system a "Leistungsverzweigter Hybridantrieb". 😅😁

[@Eugensson]

toyota calls it eCVT

[@qztronic]

@@Eugensson in Toyota’s R&D lab it’s nicknamed the “power splitter”

[@FPSlacker]

Also the Chevy Volt's drivetrain www.greencarreports.com/news/1096942_2016-chevrolet-volt-powertrain-how-it-works-in-electric-hybrid-modes

[@liesdamnlies3372]

German: “CONCATENATE ALL THE THINGS”

[@Eugensson]

@@liesdamnlies3372 Luxembourgish: Do the same as germans, but in a most unexpected way, eg ZWEEEEËG

[@rklauco]

The best way to learn is to fail :) Congrats on not giving up!

[@ChrisTuttle]

Agreed. Also, documenting expectations/hypotheses and how they relate to test results is fun to watch and learn about. One of the best aspects of MythBusters has been executed well here! Right on Levi.

[@LeviJanssen]

Thank you, and I absolutely agree. The world needs more failures!

[@LeviJanssen]

To be mentioned in the same thought as Mythbusters is an honor, thank you

[@danriches7328]

Well, failure is how we learn and I really admire your full honest approach and not omitting the failures. I found this very fascinating, keep up the great work and we'll all learn something worth while. Awesome!!

[@kingmasterlord]

"I'm not saying I'm the first to ever invent this, but I am saying I developed it independently"

[@TheNefastor]

Also known as the "patent loophole" :-D

[@kingmasterlord]

@@TheNefastor or: Why Patents Are Bullshit I mean seriously what is sane about "I had that idea so no one else is ever allowed to think of it"?

[@TheNefastor]

@@kingmasterlord well that's not an accurate portrayal. It's more "I had an idea so everyone else must pay me to use it".

[@kingmasterlord]

@@TheNefastor it's not a race to have unique ideas. you don't deserve my money just because you had an idea earlier than I did. I'm not using your idea, I'm using mine.

[@TheNefastor]

@@kingmasterlord yeah except it's hard to prove your idea is original if someone else has published it first. You can't really prove your ignorance (short of being a brexit voter 😅)

[@TomasSab3D]

This is beautiful. And your style is flawless.

[@Thepiecat]

I love that you turned a "failure" into a video about what you learned. Quality video!

[@witoldgrabowski9263]

I'm learning now, how to turn any quality video into an unmitigated disaster, or irreversible failure. But, thanks for watching.

[@byronmckenzie5805]

@@witoldgrabowski9263 n. m m .mmm

[@stefancoppens4922]

The "reverse" differential drive you describe was used for a century (I believe) to actuate rudders on ships. Two large DC traction motors ran in opposite directions at high speed, with the rudder linked to the output shaft. The rudder could then be moved by varying the electrical power fed to the motors, one more and one less, resulting in a high torque at the rudder very quickly. This was before the days of sophisticated hydraulics, and avoided the need to spin up a single traction motor from rest every time a rudder movement was required.

[@Skyentific]

Interesting idea! Everyone is using cycloidal nowadays. :)

[@megadjc192]

all you have to do is invert one of the inputs and you've just created the mechanical equivalent to an electronic operational amplifier where the gear ratios determine the "gain" and you can even feed the output back to one of the inputs with a variable gear ratio and you can change the "gain" or input to output ratio of the drive unit.

[@fire17102]

Go on ...

[@DFPercush]

@@fire17102 "What is an op amp" kzbin.info/www/bejne/bXe8eadrjducgcU just substitute "rpms" instead of "volts". That's a pretty sick idea btw.

[@megadjc192]

@@DFPercush also torque equals current. And if you multiply torque*rpm you get horsepower which then can be directly converted to watts. The same energy you see in electrical systems

[@whtwolf100]

@@megadjc192 so what sort of applications or purpose would this serve? Sorry, not as knowledgeable about this stuff, but its really cool!

[@megadjc192]

The primary usage of this is in automatic transmissions as they allow the syncronized smooth change of gears where one of the inputs is connected to the engine then the gears are set at a specific ratio then the other input is forced to zero rpm thus allowing you to selectively change the rpm and torque without have to mesh gears together. A planetary gearset is a mechanical operational amplifier. Also, one can use this at a simple analog computer. For example, you can compare two rpms of two different rotational shafts then compare the output rpm and that output rpm will be the addition or subration of the two input rpms depending on the configuration. You can use this effect to accurately measure a high rpm source by converting it down to a lower rpm and then allowing an extremely precise measurement. Basically anything you can do with an electronic op amp you can do with a mechanical one. You can also integrate and differentiate the same way and this was commonly used in old mechanical computers. Also the mechanical op amp is completely reversibale unlike its electronic cousin so as long as the two outputs add up to the same input you can have any configuration of rpm and torque on the two outputs. This is how a car differential works. The limited slip differential just has clamps on both of the outputs so that no one output will equal the input in terms of rpm or torque. Otherwise the other output will be zero which is usually undesireable with a differential.

[@ddegn]

I'd really like to see a demonstration of the variable speed feature you described. Put encoders on the motors and output and test your theory. It might be my bias in favor of encoders, but I have a hard time believing this mechanism will allow precise speed control. As with standard gearboxes, I think changing loads will require adjusting the power input to maintain a desired speed. Excellent video. Thanks for sharing your projects with us.

[@ethanspi]

Agreed, he "confirms" his two theories but didn't even demonstrate those two theories at all.

[@tt54l32v]

Look up Nissan electric variable valve timing.

[@ronnetgrazer362]

I never thought of a rear differential as being a planetary gear pulled into 3D, or a planetary being like a rear differtial folded inside out and flattened. Thanks for that!

[@openlink9958]

I know it has been a while but timestamps would be useful

[@antitriangular2828]

With regards to the arbitrarily precise positioning; my understanding is that under applied load, in order to energise the windings; a motor controller requires either a sensor or encoder to know the exact position of the permanent magnets. I don't quite understand how this requirement could be removed - as soon as you add load to the mechanism, the rotation speed of the motors will reduce. So without an encoder or sensor; you end up losing most of your torque as the wrong coils are energised in relation to the poles on the rotor. Perhaps you're implying the use of back EMF to determine the rotor position?

[@ddegn]

I should have read more comments before adding one myself. You just gave a better explanation of what I was trying to say in my comment. I really don't understand how this mechanism can produce precise speeds without encoder feedback.

[@LeviJanssen]

Sensorless motors do have positional feedback, but it's achieved via measuring the back EMF across the coils, and sufficient back EMF is only generated at speed. So, a sensorless driver can maintain a specified velocity, compensating for external load based off the timing of the back EMF. Arbitrarily precise movement can then be achieved because the actuator can stall when the motors are at or above that threshold speed.

[@antitriangular2828]

@@LeviJanssen Thanks. Thought that might be the case So, to summarise; back EMF can be used over a certain speed threshold and accurately maintain that speed Run both motors above this rpm threshold. run in opposing directions to "stall" the output (motors are still running at an rpm above back EMF threshold) output entirely depends on ratio between the two motor speed+direction.

[@jaymoser7953]

You should look at the drive system for the Sherman tanks, they use the difference of 2 motors to do forward backward and speed control. It is a very old idea

[@spankeyfish]

It's also used in mechanically controlled gear hobbing machines.

[@ilikec]

They had double and triple differential steering in certain tanks.

[@tHaH4x0r]

It is the first time for me as well to see this idea. Very neat. You might be able to write a publishable paper on it. Core XY mechanisms technically also double the speed achievable, although not in all directions thus strictly speaking don't double the velocity. Even a 'failure' is a success, because something new is learned.

[@howardbartlett3419]

This is pretty similar to a differential swerve drivetrain that some FIRST robotics teams have experimented with before. It is also similar to some types of robotic gripper/wrist mechanisms, since you can actually pull two outputs out of the differential.

[@BogdanTheGeek]

Power = torque*velocity so for 2 motors: 2*P = 2*torque*velocity which is (2*torque)*velocity or torque*(2*velocity), cant get more than 2 times the power :)

[@LeviJanssen]

Yup, I initially tried to explain my way around this, convincing myself that it had two modes in which torque or speed could be doubled separately. Of course, that wasn't the case, and my initial physics intuition won the day.

[@logan5018]

I actually worked on something similar to this in high school. I did FRC robotics and i think from around 2017 to 2019 we worked on making a "differential swerve drive" and used it on a robot in a 2019 off season competition. Idk if "swerve drive" is something commonly used outside of FRC, but it was very common in FRC, and it was basically just a system of driving a wheel using one motor for power and one for turning, and the diff swerve was designed to be an improvement over that. I know several teams have designs by now, and one team brought them to a on season competition as early as 2019. With our previous standard swerve drives we were already one of the faster robots, but we had virtually no pushing power, so i believe the intention was to gear the motors such that we could get similar speeds with increased torque, but im not super sure on that. Here's some documentation (though on a later version than the one i worked on): www.chiefdelphi.com/t/4143-diffswerve-2019/364655/7 And a video of it running (this time the one i worked on, it was a bitch and a half to mill & assemble): kzbin.info/www/bejne/mZ25hqlunr17mtk

[@DustinMaki1]

Small world. I was a control systems advisor and robot inspector at St. Louis, Smokey Mountain Regionals, and Champs. I've gone out to eat with your team and remember sharing my ideas for differential swerve drive, but using 2 drive wheels on the swiveling pod and using the ground for coupling. The object was to combine steering and traction functions with 2 motors instead of carrying the weight penalty of a motor dedicated solely to steering the swerve module. Gearing it eliminates the traction variable. Good work. MARS 4143 is one of my favorite teams in all of FRC because they innovate, share , and encourage others so effectively. I had this idea www.diyelectriccar.com/threads/differential-torque-regen-controller-design-principles.204403/ refined a bit since to fit the back of my bike www.guzzitech.com/forums/threads/final-drive-casting-mods.21263/. Cutting the OD of each ring gear to match a worm on each electric motor for a 3 input 2 wheel drive w/electric differential. Motorcycle turns a solid axle shaft. Each side of the axle drives a sun gear, the wheel output is taken from the planet carrier. A steering angle sensor and gyro help determine how much speed differential is required to get around corners. Its torque vectoring. I'm trying to work out the user interface for differential throttle/brake/steering. Right now it looks like left/right twist grips on handlebars that can rotate either forward or reverse from neutral position. Its weird rolling throttle forward to accelerate because of wrist mechanics, normally it only twists backwards. Engine throttle is converted to drive by wire combination of L/R inputs. Nice side benefit of electric reverse and nearly turn in place slow speed maneuvering with no additional effort.

[@kenworks6068]

Beautiful build quality. Great presentation and example of the value of lab tests in conjunction with thought experiments. The first computer I worked on professionally, used physical differentials to perform vector math based on heading to integrate acceleration, and speed into distance traveled in latitude and longitude for the F4 fighter jets. It worked well - Back in my day. The entire navigation and weapons delivery was done in analog computing and was build before Apollo.

[@LeviJanssen]

Wow, that's fascinating!

[@denispalmer1937]

I designed this and applied for a patent on it about 12 years ago, it is a great system never put in market

[@MrKhan-od6it]

May i get your patent number?

[@lernenmitrobin]

15:42 "you live, you learn." - Great dude!

[@Steevo69]

This is already done for agriculture machinery, high torque hydraulic motor (s) coupled to Diesel engine crankshaft to adjust for loads.

[@DaveEtchells]

I just returned to this video after having seen it back when you first posted it. It’s clever and very well-explained! I realized that this is essentially the same principle used in the ancient “north-pointing chariot”. There, the inputs were the rotation of the two chariot wheels, and the output was a pointer that s always pointed in the same direction, regardless of how you rotated the chariot. Anyway, a great project and explanation of the project!

[@samudrajs5409]

Shouldn't you have made the simplest version possible of this thing to make sure this even works and then move on to build the complex version? Asking out of curiosity

[@kirknelson156]

interesting concepts but I fail to see any practical use, typically you chose the motor based on need like torque and RPM, its why so many different models exist. to double up motors with the additional gearing your are complicating the design for what could be accomplished by just 1 motor. the only reason I can think of to double up the motors is space requirement, if the space allowed doesn't provide enough room to install a larger motor and you need additional torque then ganging 2 smaller motors might be an answer. but its far easier to join the shafts than a complicated differential gearing design. but hey keep thinking outside the box, I could be wrong and just can't see the benefit.

[@kevinl5554]

An interesting system use could be trying to get more than 3600 RPM out of AC motors . Say an ice cream factory that has a huge mixing dish that is 10 feet wide and 10 feet deep. You could use a system like this with one motor for the initial mixing at low speed and then 2 motors to get close to 7200 RPM input for the full ingredient mixing. or vise versa.

[@andrewdowling6321]

This one. kzbin.info/www/bejne/fXjRgXiwoN-No9k And I am pretty sure this one too kzbin.info/www/bejne/bGjSZ5V8i9JrgtE At least you admit yours is pretty much a differential from the get go. So I think spinning both motors at max rpm in the opposite directions, then rapidly switching the direction of one motor would let you use the spinning mass of the system and direct it through the output, allowing for an increased initial launch torque.

[@AllenBarclayAllen]

4th hypothesis reverse with no gearing..! Why hasn't automotive industry it's done that yet its absolutely insane..!

[@butterbaybiscuits8694]

What a humble learner!

[@Unmannedair]

This is a nice idea. Honest critique: I don't like the belt and I think the tolerances probably need a lot of tightening. James Brunton has an excellent version of a single cycloidal that fixed the tolerances using bearings. Other than that, I think it's a brilliant idea. One problem I noticed, on your master transmission, your motors don't see equal load and travel. I suspect you'd see better results if your slave motor had a different gear ratio that reflects it's loading on the outer ring gear.

[@d4vidd]

thank you, great content! My subwoofer is going to explode while watching the video :) Just FYI, your audio channel's very low end has many high amp artifacts, maybe a high-pass filter could fix those after recording (or better fix the issue during recording). Cheers

[@bdr420i]

Why your glasses have double reflection??

[@rmkscrambler]

This sounds a lot like an IVT or infinitely variable transmission, versions are used in some hybrid cars and john deer tractors. Usually 3 motor inputs (john deer uses 3 cvt inputs) are used to control the sun, ring, and carrier of a planetary creating "infinite" gearing.

[@LeviJanssen]

I initially thought this would be an IVT/CVT system, hence the double-torque hypothesis. But, considering my final thought experiment and the experiment, this can't be the case. As far as understand.

[@sandervesik173]

You may want to look up epicycling gearing, inc patents for it - it does the same thing (i think), allow riving a shaft from several motors. Its not a new thing really, these were used in middle ages to make astronomical clocks and other gear that allowed for prediction of planet's movements - primitive mechanical computers. There are also books on design of these. Overall, very well done.

[@LeviJanssen]

A planetary gearset has one output and one input, not two. Though this can be achieved via a modified planetary gearset (or many other reduction systems), it is distinctly different.

[@shaunhendrickson1628]

If you come to the U.K. can I meet you. Thanks

[@wesleymays1931]

So rather than adding their stall torques together, you instead got the stall torque of the weaker motor. Interesting.

[@sublucid]

The easiest way for me to think about this is to just mount next motor to the output shaft of the previous motor. If you imagine a longer chain of motors, then I think how the stall torque distributes begins to make more sense.

[@holgerpieta7367]

And it also explains why Hypothesis 4 doesn't hold water: If you add up two values with independent errors, the error add up quadratically. So the precision of speed or position control of two coupled motors will always be worse than a single motor. Of course you can get below the minimum RPM of a single motor, but with a horrendous quality of control.

[@AnthonyFrancisJones]

Brilliant and quick explanation of the differential gear - not an easy task. Well done!

[@DrGoo-px9eb]

Awesome video, I like that you show where you went wrong, really showing people that engineering is about analyzing failures and not only about making perfect systems all the time.

[@allenwalker1379]

Thank you for showing us your whole development process and not just the successful parts.

[@busyonthefarm]

I believe this is how the combination gearbox of twin engine Bell helicopters (Huey/Cobra) works

[@robertsteinwandel6658]

What if you did this with a non back drivable gear set, like a strain wave gear for each motor. Shouldn’t the new stall torque be double the stall torque of each harmonic gear individually? I may be thinking about this wrong Ive only been thinking for the last minute or two, but it seems that since they cant be back driven, it would be the outer shell that would have to fail in order to back drive. Since the 2 outer shells are attached with a toothed belt, you would have to cause both gears to fail together in order to break it?

[@BensHacks]

One application for driving one shaft with two engin is in ships. The concept is called "Vater Sohn Antrieb" (German), "father son drive". I wasn't able to find anything on the internet with this phrases. The idea is coupling two motors to on propeller shaft to get the option to run any of the motor under optimum efficiency conditions under different load condition. The father motor has around twice the power of the son motor. So using only the son, the ship can cruise low speeds at maximum efficiency of the small motor. For maximum both motors are engaged. This concept is used on tugboats and a specially on riverboats. For example on some passages of the Reihn ships need very high power to go uphill but for going downhill they need only little power to enable enough flow at the ruder to maintain stearebillity.

[@BRNKoINSANITY]

I would imagine that the reason this type of gearbox isn't used is the same reason differentials blow up if you run a spare tire too long... The extra gear movement generates a *lot* of heat and wear. Additionally, most motors are already gear reduced. Better to get your increased speed by building out the engine or using a higher voltage/wattage electric motor.

[@cooperised]

Great stuff! Take a look at the 10100, an experimental British diesel locomotive from 1950. It used four engines that could each be coupled to the transmission independently, with a similar effect to your design. en.m.wikipedia.org/wiki/British_Rail_10100

[@AnubisSolvang]

And here I thought this video was like 5 years old. Oh well, at least you know it's been around for a long time and you can use the many examples to improve your design.

[@HandyDan]

2:20 this method of planetary gears is used in many hybrid cars, the Chevy Volt does this where the outer ring gear can be held and use the reduced speed of the planetary or the planetary gears can be held to overdrive the ring, which it uses to recharge the battery. Great video!

[@HuskyMachining]

This is cool shit..... glad to have found the channel

[@gumbykevbo]

The outboard spoilers (used for additional roll control at lower air speeds) on A B-1b aircraft use triple redundant motors driving through two differentials. I was the product engineer for the electronics that control them for most of the production run.

[@turboprint3d]

I wonder if you did the planetary idea, with say the planet gear coupled to one motor , then say the outer ring , or the planet frame connected to another smaller motor. It would be used primarily with one motor to drive the load, once up to speed the other motor could be turned on to add rpm to the output as load decreases . it could also be used as a clutch with zero friction parts other than whats involved in a gearbox . seams like there would be a few ways to configure such a system that might have applications more in the idea of durability , or improved throttle control . maybe even some sort of hybrid vehicle application . I feel like the best use of this would to have wildly miss matched motors , like one really low kv , and one really high kv . Its defiantly interesting to consider , nice work .

[@JustCuzRobotics]

Based on how you describe this system at the end of the video it sounds analogous to electrical batteries. When wired in series you get double the voltage (speed) but the same current capacity (peak torque) whereas when wired in parallel you get the same voltage (speed) but double the peak current capacity (peak torque) and also double the energy capacity (thermal mass). Directly coupling two motors is like wiring two batteries in parallel. You kinda made a way to drive them in series, almost as if one motor is directly attached to the output of another. If you imagine having a huge output axel rod with a motor on one end mounted to a fixed base and another motor in the middle, it seems obvious that the axel will only ever be able to resist turning with the stall torque of one motor, but to accomplish this, both motors need to be stalled. However if one motor is turning the rod from the end to the second at 100rpm and the second motor turns the end of the rod at 100 RPM the end is now at 200RPM.

[@BraxtonHoward]

Would it be more likely to work with planetary gears like in the first graphic?

[@Tvvistedfork]

so, im not too certain about the answer to your question, but i think the answer would be no. the reason being the motors are going to be exerting different torque on the output because the outer motor will have a different gear reduction compared to the inner motor. this falls back on the weakest link of the motor, what ever motor has the weakest torque rating that that said speed will fail first. the only way to remedy this issue will be to directly link the motors.

[@LeviJanssen]

It should work the same for all reduction systems.

[@fuhkoffandie]

I have something I'll make you scratch your Noggin. Ready for this? Simplify. Simplify. Simplify. You're on the right track.......

[@Sekir80]

Yeah, this reverse differential was the drive of choice when my friend built a robot in his university years. There weren't any precise way to control the joints and this was their (his? can't remember) idea. The story dates back to the eighties.

[@StephenWilliams1]

Links to anything?

[@robertoguerra5375]

Great and interesting experiment... I am an electrical engineer, but I think I skipped the class about "stalled torque"... going to learn it now. The speed combination or disaggregation function is well known in the Prius power split device (a planetary), a differential (as you said), and 2 motors connected in series (output shaft of M1 spins the case of M2, powered by a slip ring)... there's a Prius simulation video showing that ;) kzbin.info/www/bejne/e5DbdKaph52kh6s One way to model the torque of 2 motors adding speed is with "harmonic sum" of parallel resistors. 1/R1+1/R2=1/R3, but with torques... 1/t1+1/t2=1/t3... if t1 increases a lot, t3 starts looking like plain t2... but if t1=t2, then t3=t2/2, but the speed is the sum of the 2 input speeds.

[@nijram15]

New viewer here: cool video! Well done, really enjoyed the editing and pace. Small tip: I got a bit distracted by the reflections om your glasses though. Have a good one!

[@htomerif]

Just spitballing here but you're doing it wrong. You can get higher torque by summing the two motor speeds, but one of them has to be going in reverse. Don't think of it as the stall torque of either of the motors. Just let the ESC handle keeping the motors at the same (reversed) speeds and see what the "stall" torque of the assembly is. As far as the math is concerned, summing the two motors' stall torque obviously doesn't work, and the failure point is where you get cogging on either of the BLDC motors, but if you have two non-stalled motors putting several kw into keeping a stick still, all of that power is going somewhere and in this case, its literally going into keeping that stick still.

[@SergioGugliandolo]

I'm not quite sure about your idea of summing the torque: in my opinion it's doable, but with a different setup (yes, doubling the speed will not be possible, to have more torque some things need to change and you will not really summ the torque at the end). Let me explain. If you take a differential T shaped gearbox, with two motors, 10Nm for motor, you can obtain the 20Nm you were hoping for and the motors are not rigidly coupled to each other, because they chan still rotate a different amount (the difference in rotation speed will be translated as a "roll" of the output shaft on it's axes). If you stop one motor (infinite torque), and rotate the other one, you will still have 20Nm of torque because the conical gear is actin like a leverage, doubling the torque of the smaller motor. In this way you will have the roll effect I was saying earlier but still double the torque (not sum of the two motor, just double of the less torque one) By the way cool video, thinking about this I designed the joint for a robot I'm building with a university team, extremely cool video :)

[@Slider68]

As others have mentioned, the Toyota Hybrid Synergy Drive uses an identical system (Toyota calls it their Power Split Device) to combine the power from the ICE (Internal Combustion Engine) and MG1 (Motor Generator 1). The output contains another electric motor, MG2. By varying the rpm of MG1, this creates an effective CVT for the ICE. Electrical power for MG1 (positive or negative) comes from MG2. See the following for more information: hybridautopart.com/blog/toyota-prius-power-split-device/ BTW, Ford also uses an identical hybrid system in their Escape Hybrid and in their C-Max & Fusion Energi & Hybrid

[@Tragson3]

You have described a Ford ECVT. A gas motor input on one side of a planetary gear set and electric motor input on the other side. A variation of speed between the 2 makes all the gear speeds.

[@pauljs75]

You built a neat torque limiter though. Just replace one of the motors with a controllable braking mechanism. Seems reasonable that it could have use for traction control.

[@davidbeliveau-viel3084]

Great video and thought process! I think you don't have a torque problem, you have a speed control problem. When you apply torque to the output shaft, the motors' speeds will change a little as a result, allowing the output to move. You'll find the same difficulty in achieving precise position control, which your stall torque experiment was in fact testing. Any speed difference in the motors of just 0.01 RPM will cause the position to drift by 0.01 RPM. But as many have said, you independently came up with something similar to Toyota's Hybrid Synergy Drive drivetrain, which is no small feat. Kudos! Keep it up!

[@jesscast5122]

mmmmm......I got BORED so I left!..... this subject is not only NOT interesting but the dude is super geek........

[@codysalyer6212]

I did a lot of work with cycloidal gears in robotics, 8:1 though possible in cycloidal gears it falls outside the range where their benefits outweigh their challenges. Things like their resilience to shock loading and high amounts of inertia during back driving are all lost when creating transmission with such low ratios. Typically ratios for these gear boxes are in the 20:1 or greater.

[@nikitanugent7165]

It's not a new idea, but I appreciate how you walked us through your initial hypotheses and showed the errors. I can think of one other place this was used, and that was in a scam infinitely variable transmission design. All you needed was a second input that output just as much torque as the first, but is infinitely variable! I think your hypothesis about the arbitrary precision is wrong, however. At least as I understood how you explained it. Not only do the speeds and positions of the two motors get added, so do the tolerances on those values. So if you normally can control the RPM to +/- 0.5, then normally you could set one motor to 1 (+/- 0.5) RPM. If you use your setup to connect motors going 99 and -100 RPM, then you'd get (99 - 100) +/- (0.5+0.5) = 1 (+/- 1) RPM. So the range of speeds is double (0-200 from 0-100), but the tolerance is also double (1.0 from 0.5). There is still a way to make this work, I think. If one of the inputs is reduced ten times more than the other, then you would be able to use the second motor to "fine tune" the output speed. This would increase the precision of the output speed, but not the accuracy. The same goes for position: better precision, but worse accuracy.

[@pjnoonan2211]

I think what you're describing is a device called a "torque converter". As far as I am aware, it was developed in the early 1900s (1910s-20s) as a way to combine the power of internal combustion engines, with the relatively weak, but precise and quickly accelerating, electric motors of the time. The only device I personally have ever seen it "used in" were some experimental auto targeting anti aircraft guns. Judging by it's lack of being known to most modern people, it doesn't seem to have been very successful, even though the concept was sound.

[@Slider68]

That is a nice way of packaging a reverse differential. I believe you are slightly confusing the torque and rpm relationships though. What you have created is actually an rpm adder, not really a torque combiner. I'll try to explain. To keep things "simpler", imagine using a common, open differential backwards, like a typical differential in a car but without a pinion gear, no driveshaft and no axle housing (just the differential). Now mount your 2 electric motors where the wheels normally go and consider the output the rotation of the ring gear. Ultimately by using a vehicle's differential backwards as defined above you will get the following somewhat confusing relationships: Torque - The electric motor torques will always been equal to each other because they apply directly at a 1:1 ratio to each other through the differential (imagine locking the ring gear temporarily to understand). If the torque isn't equal, the motors will spin at whatever rpm is necessary to make the torque equal (again imagine locking the ring gear). - The output (ring gear) torque will always be equal to the sum of the motor's torques, which is the same as 2x either motor torque (because they must be equal). - If you lock one motor and drive the other motor you will still get the same 2x motor torque at the output. With the locked side, it will resist by exerting an equal torque to the motor that is being driven, hence again the 2x or sum of motor torque at the output. - Mathwise: To = Tm1+ Tm2 = 2xTm1 = 2xTm2 , where To = output torque Tm1 = motor 1 torque Tm2 = motor 2 torque - Because the motor torques must be equal, you can't really think of it as a torque combiner. - As mentioned above, if the torques are not equal, the motors will spin at whatever rpm is necessary to achieve equal torque. RPM - The output (ring gear) rpm will always be equal to 1/2 of the sum of the motor rpms. - If both motors are spinning at the same rpm, both motors and the output will ALL spin at the same rpm. - Mathwise: Output rpm = 1/2(M1+M2) , where M1 = rpm of motor 1 M2 = rpm of motor 2 - You could think of it as an rpm adder with each side having a 1:2 gear reduction. - by changing the gear ratios you can change the rpm ratios but it will always be an rpm adder. Where this is useful is if you have a motor that is sensitive to rpm, like an internal combustion engine, you can use an electric motor on the 2nd input to vary the rpm relationship between the 1st motor and the output. On top of this, if you use the 2nd input motor as a generator and use this electricity generated to power an electric motor driving the wheels, you now achieve a torque increase at lower wheel rpms (lower vehicle speeds), which is what you want at low speeds. At high wheel speeds you loose the ability to generate electrical power through the 2nd input so it behaves like switching to a highway gear ratio. This is why several hybrid vehicles use these as a CVT.

[@Shazbat5]

Look at the Toyota Electric Variable Transmission. It uses two 2:1 planetary gear sets in series, with an internal combustion engine driving the #1 set and the wheels, with electric motor #1 driving the #1 ring gear, and the #2 ring gear driven by electric motor #2 with the output from #2 driving set #1. Each ring gear can move forward or backward at any desired speed from +6000rpm to -6000rpm. This gives you any gear ratio you want from 1:4 to 4:1, using no, either or both motors to add torque or rpm, or subtract power to charge a battery or to run the other motor (or to start the ICE)

[@oldskool4580]

Awesome work, never give up, building anything is better than playin video games !!! Lol, Man I'm old. I believe you could possibly find extreme versions of this idea by looking into professional tractor pulling where Some competitors will have multiple 3000+ hp engines synchronized driving one differential. I know that's kinda backwards but I'd bet you are following a similar path.

[@jonwatte4293]

I think you're wrong about the "arbitrarily precise" thing. Each of the input motors have some number of phases. You can control the output as (P1+P2)/2, which is no different from gearing one of the motors down 2x. This comes from the "no encoders" assuming that the BLDCs have accurate speed derived from commutation. You can just use that same assumption directly from one motor.

[@simplyceem4145]

Any gear set is of course a lever system. Your design is equivalent to the arm of an excavator - you could move either input X amount and (if ratios are the same for both pieces) get the same movement at the actuator end, however the furthest actuator from the output / load end also has to move the next stage, increasing the length of the lever (This will be why you had to 'gear down' the secondary motor.) Move both sets together and you will SUM the output motion As you eventually found out the issue is, exactly like an excavator arm, that the furthest actuator from the load / output will suffer the highest level of loading because the overall length of it means the leverage includes subsequent stages :) Hence why excavators usually have 2 MASSIVE cylinders driving the main arm and only one smaller one driving the furthest limb! I rarely comment on videos but I felt like it was worth pointing out (not sure if anyone else in the comments said the same thing before me mind you?) Great content, thumbs up :D

[@oleksandrgrytsenko]

Am I wrong that this described mechanism is basically a Compound Drive from Machinery's Handbook (page 2214 for 29th edition)? : Gears, Splines and Cams / Other Gear Types / Planetary Gearing / Ratios of Planetary or Epicyclic Gearing / Fig.20

[@xaytana]

Your connection between a differential working in reverse and a planetary gearset more than likely comes from the fact that a differential _is_ a planetary gearset, where the carrier is the input, with the ring and sun as the outputs, with the planets/satellites/spiders allowing for differential movement; the differential is a unique planetary gearset because it's folded in on itself where the ring and sun are coaxial and equal in size. These have existed for awhile, though for some reason they're referred to as an 'adder' gearbox, at least in certain circles it's called an adder; and those certain circles make adder/subtractor gearboxes. In fact, any hybrid that uses a 'boost' motor functions in this way, whether that's EV main with ICE boost (Prius) or ICE main with EV boost (I believe this is what most of the supercar hybrids use, Porsche 918, the newer NSX, etc.), with the oldest patent I know of being in filed in 1969, granted and publicized in 1971, by Baruch Berman, "Power train using multiple power sources," patent US3566717A; of which multiple modern patents reference for hybrid 'boost' drives. Though, I don't think I've ever seen cycloidals used for this. If you continue iterating on this idea, I would suggest going back to basics, start with your original idea, the inversed differential, that way each input is 1:1, and the satellites are also 1:1 with the sun and ring, this is the absolute most basic setup as you're not dealing with gear ratios, only the speed ratio of input 1 and input 2 and their effect on the output. From there, unfold the differential into a more typical planetary gearset, where you'll have to mesh the motors to the gearset to achieve the 1:1, or direct drive the gearset and differ the speeds of the motors to achieve a similar output to the differential design. Then from there your can iterate on the cycloidal design, this is where you'd be leaving the territory of a widely known mechanism into the territory of a lesser known mechanism. I would like to see you iterate on this idea, but I think you do need to approach the base concept first, learn the basics, then use that knowledge to attempt the cycloidal variant. And who knows, maybe cycloidals are less than idea for this application, for whatever reasons you may eventually find.

[@CraigBurden1]

My gut says that the output torque would be: 1/Tout = 1/Ta + 1/Tb Ta and Tb being torque from motor A and B respectively. I'm not sure I agree with the use case of the arbitrary precision. Assuming the controllers using ZX are able to perfectly control speed. That would give you "position control" on the output because of the differential. However accounting for disturbances like a variable load torque would not be accounted for. So it isn't truly position control but speed control that is able to overcome the low speed limitations of a ZX controller. So yes you could make it look like position control but it is really just good speed control. Until of course you add the encoder on the output at which point you have all the parts needed for FOC again, so it is likely better to just use FOC. Although if the use case specifically needed the speed addition then I think you could make it work. Very cool idea! I'm excited to see if you manage to get functional. Perhaps a design with a epicyclic geartrain would be easier to build and debug?

[@michazarzycki1634]

Is it posiible for this designe to use stpper mottors, or BLDC motor are reuired? I was thinikng about it, and it seems that with stepper motors if primary motor would be power up, but wouldnt turn at the moemnt, and the primary housing was to rotate becouse of the secondery motor, the cycloid dysk would stuck in one position and resist the rotation of the primary houseing. If there is missunderstanding on my part, would anybody like to explain it to me? I would love to designe something like this using stpper motors becoise bldc are to expensive for me.

[@uavcrazy9562]

not going to lie, I had this exact idea, except just with a planetary gearbox with a driven outer gear, but I'm curious to if those could be either magnets, or somehow ran through eddy current as a frictionless drivetrain. my theory is that if they are slightly off in speed (a ridiculously low output speed) you would have to have a ton of torque (in theory at least?)

[@nibblernibbles3205]

Check out the "power split" transmission used in heavy off-road vehicles eg. Fendt Vario tractors. Often by splitting engine power into two paths: 1) a gearbox with fixed ratios 2) a continuously variable hydrostatic transmission. Then summing the two with a differential gear. Result is a high efficiency transmission with continuously variable ratio. Only a fraction of the total power is transmitted through the variable element, which typically has lower efficiency than the gearbox.

[@markmclaren3836]

I'm a retired aircraft mechanic. If you can get this thing to work fully, then it sounds like it might make a GREAT aircraft Flap and Undercarriage, and undercarriage Door actuator motor. Over to you, genius. M. M.cLaren, Brisbane, Australia

[@Side85Winder]

You are totaly correct it could be a very nice mechanism for a 3d printer if you are using nema 23's. They have good torque but are generally a little to slow. You wouldn't want to gear down the motor that much though as their torque is decent. I had my eye on a stepper online motor thats 2.8A 0.9° that can output 0.6nm of torque at 375 rpm. In the case of the sub 5min benchy prints this needs to be around 400rpm with the same values. Your mechanism would allow this setup to be faster and still have enough power to precisely move. If you come up with something i know some people that would be interested including my self.

[@dragancrnogorac3851]

Gotta disappoint you dude. Komatsu buldozer(probably more their machines) have this system for driving tracks. There is one main driving shaft and two speed alternating shafts for each track. Therefore you can speed up or slow down each track on komatsu buldozer without cutting power. So on one machine got regular efficient automatic transmission and on top of that one hydraulic pump and two hydraulic motors for speed alternating (turning purposes) I believe there are more machines with that mechanism like tractor with tracks and skeedsteer loader... Who knows

[@santiagovaldez5176]

I think there a 2 simpler ways of achieving doble speed and same torque: 1) make motor 1 directly spin motor 2 completely and connect output to m2's shaft 2) connect m1 and m2 doubling the torque and use a 1:2 ratio to double the speed and the torque of 1 motor

[@jdholbrook33]

Working with electrical / electronics on a metal topped work bench? You would be fired quickly in my shop. Please, for your own safety, get rid of the metal top before doing any more electrical / electronic work on that bench. Metal top benches are for mechanics, NOT electrical. Other than that. Cool video and very interesting. My mind is definitely stimulated.

[@yahku2231]

Remember you have shared the current from the same power supply ... to solve provide each motor with it's own power supply. Next look into syncros & servos It's a electric way to add two mechanical inputs to produce a 3rd output. Single phase inputs for 2 synchros, divided into 3 poles, that are interconnected at the stators... the 3rd motor is a servo, that detect the difference & then respond to turn until there is no electrical difference in voltage. This synchro-servo combination can have light-weight inputs on the 2 synchros, that then can drive a super heavyweight output, all depending on the power supply and size of the servo-motor.

[@vonkel85]

This was done in 2006 in the agriculture tractors. It was a CVT transmission. On a planetary gear set. The sun gear was driven by the engine RPM. The planetary carrier is splined to the rear differential. The ring gear was driven my a hydraulic variable displacement piston motor. The pump that supplied said motor was driven by the flywheel of the engine. It was also variable displacement. This gave the tractor great control over torque and speed. To me. It's the way CVT should be done

[@hamjudo]

With LEGO Technics it is easy to combine the output of two motors with a differential. My LEGOs are from decades ago. The gears in the differential from the 1990's are pretty wimpy and comparatively inefficient. The max torque through the gear train will be limited by those tiny gears. The total power through the system will be noticeably less than twice the single motor power, because a lot of energy will be lost in the gear train. The modern LEGO differential is better in both respects.

[@drew79s]

An IVT from a Toyota hybrid uses the same operating principle. The benefit is that you effectively "blend" the two ratios, allowing for a combination of generation and drive, as well as the fact that you can run the electric motors in reverse, or idle them, allowing the petrol motor to run without stalling even when the output isn't operating.

[@barefootalien]

Well. I mean, it is pretty. But yeah, the second and third hypotheses sounded pretty dubious to me. Electric motors have by far the _most_ torque when they're at zero velocity, with torque falling off inverse to speed. What you built is, in a sense, a backward differential... but a simple differential has a similar problem. If you lift a car with a simple differential up, and hold on to the tire on one side, you can drive the opposite wheel as fast as you want and feel very little force on the one you're holding. Considering you are nowhere near as strong as a car engine, this should be telling. (Edit: do not try this at home, or if you don't have a full understanding of the mechanisms at play and the vehicle's history. If you were doing this, got the differential up to speed, and then it failed in certain ways, you could easily end up badly injured as friction builds suddenly and makes the wheel you're holding onto spin with the force not only of the engine and all its momentum, but the other wheel's momentum as well!) If you wanted to actually make some kind of a go at this, you'd need to study things like limited slip differentials, or viscous differentials. I'm... not at all sure those would translate into the kind of stationary torque you have in mind, though. Ultimately I think the maximum stall torque would end up limited to the friction or viscosity of the limited slip system. If you think about it, this is no different from a normal actuator gearbox... only there, the thing resisting the movement against the motor is the sheer mass of the vehicle/device/workbench it's attached to, along with the fasteners and joints therein. With this setup, the thing resisting the movement is... well... nothing. Just whatever parasitic friction there is inside the setup. XD Like you said, you live and you learn.

[@BrainSlugs83]

You may not know it, but the differential shown in that ancient video can actually already do what you want. Just connect two motors the the "outputs" and use the "input" as the actual output. Also, they're really easy to build, take a look at the Lego differential, it's the same thing, but optimized for simplicity. (Edit: not claiming it can prove your original hypothesis, but that it can work in the reverse direction. But you can use this to create an automatic transmission with the addition of a slip gear.)

[@DB-thats-me]

Ummmm…Why did you add water to the torque test arm? You could have used a known weight at different displaced lengths. No error encroachment weighing the water. Heck. You could use a calibrated scale (ruler) as the arm and mark it to read torque directly. 🤓 Just saying. 👍❤️

[@protoborg]

You are wrong. The input is on the RED ring gear and the output is on the YELLOW sun gear AND the green planetary gears. Also, the output of the differential is NOT the sum of the engine. It is the engine DIVIDED by the the number of outputs. Also, an automatic transmission has multiple inputs to one output.

[@brown56765]

I think the speed isn't a huge benefit, as you could get the same thing just gearing up the traditional coupling by a factor of 2 (1x torque, 2x speed) But the very slow output with fast inputs is a huge benefit if you're coupling motors that die at low speeds (like ICEs) or non-encodered electric motors that have trouble at low speeds

[@jonwatte4293]

If you have twice the torque, you could use twice the gear ratio to get twice the speed. Anyway, Skyentific did a nice 2DOF actuator with a differential input, too!

[@JoeJoe-lq6bd]

Yeah. I was a little confused why, if the goal was just to spin at twice the speed, a simple change of gearing wouldn’t be sufficient.

[@fire17102]

I like your analogous to the differential drive, I think what happened is similar to a car sitting on it's belly, If your car is stuck in the sand in a way which one wheel (left) is touching the ground and one (right) is floating in the air, with an open differential, that right floating wheel will get spun rapidly and the left one will not get any power Your output is like the left wheel, but since yours isn't locked to the ground, it fails to hold a weight and drops I think it could be cool if you could design a *locking mechanism to your differential* which can be released after some speed is gained. Maybe even clutch it for smooth transition between modes Love the ingenuity! Keep up the amazing work 💪🏾