Torque Test of Robot Actuator (Brushless Motor Robotic Joint)

[@Skyentific]

Do not hesitate to leave your comments! This will help to promote this video and channel. Thank you!

[@AltMarc]

Just made a small donation for a cheap "emergency stop button", on Aliexpress...

[@marcus_w0]

You're using the wrong kind of drive with the wrong kind of motor for your purpose. These drives are intended for high speed, high accuracy linear motion, not rotational torque. That's what stepper motors are made for.

[@franktkalcevic5342]

@@marcus_w0 All high end (expensive!) robot arms use servo motors and gearboxes. You can't beat the power to weight ratio, accuracy, or speed.

[@marcus_w0]

@@franktkalcevic5342 Yeah - they use Industrial SERVOS - not a hobbygrade brushless rc motor made for rotational speed, which could be translated to directional speed, but not rotational torque! these motors don't like low rpm, and they aren't made for constant stalling. These motors are like horses, they want to run free :) Servomotors as to the motors, also have a entirely different gearing than this one.

[@franktkalcevic5342]

@@marcus_w0 How do these RC motors differ to the industrial servo motors?

[@exol511]

Forgot to say last time! Since you are using a disproportionately large reduction on the second phase of the transmission I see some play with the belt under load in the direction where there is no tensioner. You might want to put a second tensioner to reduce it (you could also use two tensioners single screw that moves both tensioners towards one another thus pinching the belt loop between them).

[@tamgaming9861]

Great to see you at work! Your printed designs are looking total professional! Would love to see a tutorial how you messure and how you calculate that it fits always so well in the end of your constructions.

[@basementadmin]

Great work. Without doubt, you'll find the best way

[@Skyentific]

Thank you! You are always leaving motivating comments! Thank you!

[@cezar17negru]

Hello and congratulations again for interesting research! I would recommend you to use some kind of endstop setup in order to make the tests more safe. You can make a mistake and input setpoint 100000 instead of 10000 and then the arm might hit something. If you have a positive and negative limit switch on your rig that can cut the power to the odrives it will be much better. Keep up the good work.

[@otonielguilhermedarocha6535]

Hello my Friends. Congratulations for your videos. I'd like to ask you two questions. Which printer are you using? It seems to be a very good 3D printer. Which material do you use to print?

[@royarad8406]

Put another belt tensioner on the other side, it's where you belt slips. You can see it bending up and then hear the sound. that should help with testing but in the long run going with gears is always better.

[@mrraimundo130]

Yeah, the holding torque of a BLDC will never be the same as a Stepper. The gear reduction and all those belts cause too much backlash, and the encoder struggles. Someone else suggested this as well, but try to put the encoder on the output shaft

[@cedricpod]

Szilágyi Szabolcs use an electronic / electromagnetic brake

[@mikegardner5859]

Just a thought... If you need to supply 50 Amps, don't you need heavier wiring to the motor? Voltage drop will contribute to the lower power at the motor.

[@Skyentific]

Good point. This is the original cables from the motor. And they do get very hot at this current. I agree that these cables should be way larger.

[@TheRainHarvester]

Didn't the O drive have a battery rail for surge? (In addition to the power supply) in that way you can get away with a cheaper power supply while the battery provides surge startup and can also be recharged when the motor slows via regenerative power.

[@EliSpizzichino]

I would try with higher voltage. Do you have the 48V version of odrive? Voltage is essential for torque. Moreover these bldc motors shines at high RPM

[@Skyentific]

Good idea! Thank you! I do have the 48V version. I can try this!

[@justtestingonce]

Skyentific high rpm with a reducer, seen a guy pulling 10kg with a bldc and harmonic drive.

[@shengzhiwang2143]

Thanks for sharing the video! Would like to know if the cooling platform for ODrive could be open-sourced? That looks cool.

[@kokotmkokot4926]

For your task you need an engine with a large number of turns, that is, with a very low kV. Thank you for the video!

[@kokotmkokot4926]

ru.aliexpress.com/item/Yuenhoang-1-N6384-150KV-DC/32921203343.html?spm=a2g0v.search0604.3.8.1e1972f9TzTW7x&ws_ab_test=searchweb0_0%2Csearchweb201602_2_10065_10068_319_317_10696_453_10084_454_10083_10618_10307_10301_537_536_10902_10059_10884_10889_10887_321_322_10915_10103_10914_10911_10910%2Csearchweb201603_6%2CppcSwitch_0&algo_pvid=36a44b43-92e5-41b4-975d-5ba993043026&algo_expid=36a44b43-92e5-41b4-975d-5ba993043026-0

[@Skyentific]

Thank you for the comment! In this project I use the motor from the same family. With the lowest KV available (350).

[@kokotmkokot4926]

@@Skyentific Извините, это сообщение напишу по русски, чтобы точно передать суть. В Вашем проекте от мотора не нужна высокая мощность, а требуется высокий крутящий момент. Момент определяется величиной магнитного поля статора и ротора. На ротор Вы повлиять не можете и магнитное поле постоянное (постоянные магниты). Магнитное поле статора определено ампер-витками, то есть количество витков умноженное на протекающий ток. Образно: если взять провод в четыре раза тоньше нынешнего и намотать в четыре раза больше витков, то текущий момент Вы получите при токе в четыре раза меньшем. Только упадут обороты из-за увеличения индуктивности обмоток. Это упрощенная модель, но для понимания происходящего достаточно. К сожалению, этот двигатель не очень подходит для проекта. Удачи и успехов в реализации идей!

[@JonnyDeRico]

Nice setup, but I would consider building a housing for the electronics. If you ever see a electrolyt cap exploding... and you give him all chance to hit, putting your head allways above the boards...

[@Skyentific]

Thank you for your comment. I will consider to make the enclose for electronics. You are absolutely right: safety first!

[@franktkalcevic5342]

When you make the motor move, why doesn't it stop at the commanded position? It seems to stop short and creep up to the target. Is this an o-drive limitation, or does the PID loop need tuning?

[@sufianrafaqut8313]

Which Lipo Batteries and Motor combination was that?!

[@Flame6332]

This is very good work, but I see the main issue being your gear reduction. I am working on a very similar product, but I'm using motors 1/3 the total watt output, on 3S, and not even using an oDrive to cut costs, just a simple hobby car esc for 15 dollars. The solution is a gear ratio. Brushless motors truly shine at higher RPMs, there peak theoretical power output is at half of their RPM. So I'm using PID on these simple ESCs and a 1 to 800 gear reduction as I want my robotic arms max speed to be 10 RPM and my brushless motors max RPM at 11.1 V is 16000, so I want to run at half of that. And lastly, your issue was that gear trains have backlash that hurts accuracy, and you're not wrong, so I put my 3D printed encoder on the last joint or second to last stage of reduction in my gears, which I am running a 1 to 9, then a 1 to 9, and then finally, a 1 to 10, to minimize efficiency losses in early nylon 3d printed spur-pinion gear. I would highly recommend at least gearing your system up much higher, moving your encoder to the final joint, and running your oDrive in velocity mode, then using simple PID to control your motor, or you could even go a step further and just put a second encoder on the final joint and build a simple positioning algorithm that sets position of the motor until the target position of the joint is achieved. Anyways, there's a lot of solutions here I believe you're overlooking, as that reduction is not going to cut it for a robotic arm. Simple as that.

[@pavelzhdanov]

What kind of gear do you use for 1-800 reduction?

[@Flame6332]

@@pavelzhdanov I will be using 3 nylon 3d printed gears of a 1 to 9, to a 1 to 9, to a final 1 to 10, increasing the thickness of the gear as i move up the reductions to accommodate for the higher torque

[@pavelzhdanov]

@@Flame6332 interesting. What type of gear have you used? What is backlash? How big it is?

[@Flame6332]

@@pavelzhdanov ...sorry, I'm still in assembly phase, however, the backlash won't be an issue regardless, as I'm placing my 3d printed encoder on the final joint, so the motor can accommodate for any backlash

[@Flame6332]

@@pavelzhdanov for example, 1 degree of backlash inaccuracy on the final joint can be accommodated with 2.2 full rotations of the BLDC motor. And I'm using gears of increasing size, but a ratio of 11 on the motor to have plenty of diameter to cool down, the secondary stage would be 9, with the bigger gear being around 16 cm in diameter, and the final stage would be the thickest gears of the robot arm, a reduction of 8, with a gear around 20 cm in diameter

[@cristiantalos4042]

This is really cool! Two questions: do you use a hotswap circuit when connecting two lipo batteries in parallel (to prevent one battery from discharging into the other one if they are not the same voltage)? Do you think that by adding a derivative component to your (PID) controller, that it will be able to reduce overshoot caused by the inertia of the system?

[@szabi19994]

Take a look at harmonic drives, or "strain wave gears". Maybe you can build one into the new joint.

[@СашкаБелый-ч6м]

What the reason why you use in this moment brushless motor instead of step motor? (in first step of your arm you use step motor)

[@Skyentific]

Great question. The short answer: the brushless motors have much more power for the unit of mass. Thus I would like to make a more powerful robot arm.

[@ramadavince6611]

@@Skyentific no step motor has more torque than brushless. Brushless has high rpm tho..

[@obsidiann1]

I am learning english with your video :))) Thanks

[@IgorKovba]

Did You take in to account weight of aluminium arm?

[@Skyentific]

Nope, I forgot. But it is quite light. I will check the weight of this arm.

[@yann516]

Hi! You are forgetting something in your calculations, it is the static torque needed to hold the metallic arm. So the global torque is (measured mass+1/2 mass of the arm)*length of the arm. Also a little detail: you wrote 0.34 cm on the board, I think this is a mistake (should be 0.34m ou 34cm) ;) And one last thing: I think you should try to make the dynamics calculations instead of statics ones. This would give you a better idea of what you can expect from your future robotic arm ;)

[@pusatberk4193]

good job thanks can you give your motor product link?

[@siddharthshekhawat5502]

My suggestion is to use screw machenism instead of belts ,,it will significantly increase the holding capacity as well as torque.

[@Skyentific]

Yes. But this system will not be a backdrevable. I would like to have a good back drivability to use the current on the motor as a kind of force sensor. But I agree that the system like on OpenDog with the ball screw could be a solution.

[@СашкаБелый-ч6м]

@Skyentific, I see you use black 15mm belt. If in big speed it have backlash, you can use the same size white belts, they are armed with steel wires and not stretch out.

[@rushbigmoney]

Good Stuff, thanks for sharing!

[@Skyentific]

Thank you for commenting!

[@dimstr8714]

А вы не пробовали снять осциллограммы по фазам движка при удержании? Ну вот категорически маленький момент на 50 А получается, не должно быть так.

[@augustosouza6402]

Which motor do you use ?

[@peterrizk4915]

How can i change a motors torque?

[@pavelzhdanov]

Even if you make 1,5 kgm, what about the weight of motors, hand etc. What is final load of do you want to reach? And how long manipulator are you going to build?

[@Skyentific]

This is going to be the joint for the elbow. The final goal is payload of 2kg and human arm reach. And of course I accounted for the weight of the motors. I would not put any hand on it. It going to have interchangeable end effectors.

[@pavelzhdanov]

How are you going to transmit torque to joints?

[@MadHatter123456]

Are you located in Switzerland?

[@Skyentific]

Yes! Exactly. In the French speaking part of Switzerland.

[@MadHatter123456]

@@Skyentific Nice! Love your videos! Helped me quite a bit! I'm using an odrive for the main spindle on my DIY-lathe... I love the odrive, such an amazing piece of tech! Keep it up, you're making great content!

[@casperfigge]

try to look in to change the fans speed to the workload the arm maks but great job

[@genghizalseitov3807]

Both stepper motor and power supply as well

[@Otakutaru]

It looks and sounds amazing. Man, really good work I love it!

[@crzprgrmmr]

To cool such a brushless motor you should route the air through the motor, not across the outer rotor. Those are designed to have airflow through them.

[@Skyentific]

It is going exactly through the motor. I have very narrow gap between motor and housing, like this most of the air is going inside the motor.

[@crzprgrmmr]

@@Skyentific sorry I didn't see it. Thanks for the heads-up, looking forward to the evolution of this project

[@khellafsamy]

Thank u for charing

[@shouryaumang6211]

Hey which 3D printer are you using?

[@Skyentific]

Prusa i3 MK3.

[@Spirit532]

The joint looks incredibly beefy, but it's still a toy. *50 amps* and it can't hold just a few kilos... You need a proper gearbox - the cheapest(under $50), crappiest planetary or even worm drive reducer will perform *an order of magnitude* better than your belts, even in terms of backlash(but it'll still be bad, and unusable for a proper robot).

[@gdipaolo]

This tensioner ilder is drowning some W and therefore you loss torque. Use a proper length and remove the ilder

[@Skyentific]

I think you right. This idler is badly positioned.

[@obsidiann1]

А вообще, хорошая (правильная) идея использовать для позиционирования не обычный сервопривод или шаговый мотор, а простой бесколлекторный мотор с энкодером. Инженерный респект тебе :)))

[@woowooNeedsFaith]

16:23 Torque is not measured as kg *per* meter (kg/m)...

[@Skyentific]

I know. :) But my English is bad. Of course it is kg*m.

[@TheRainHarvester]

A grocery shopping robot! :-)

[@dasraiser]

I'm surprised the torque is low as you have around 60:1 ratio on the motor.

[@in_science_we_know]

Hello! I think your problem is mech. Maybe you will try gear or planetary reductors? Its most powerfull (worm better) than yours gear-gear-gear. Pleas see AR2 diy robot, its pretty cool and compact.

[@hengji5855]

Increase the torque of the motor with planetary reduction gears(用行星减速齿轮增加电机的扭矩)

[@Skyentific]

Unfortunately they usually have backlash. But I would like to have precise robot joint.

[@hengji5855]

@@Skyentific Try to fine tune the PID parameters. Using a larger proportion of the gear unit, using metal gears, and the belt also produces a large error, my mobile robot can reach 0.1 mm horizontally. In fact, other sensors and programs can be used to correct these errors. Hope it helps you:-) (尝试微调PID参数。使用更大比例的减速器，使用金属齿轮，并且皮带也会产生很大的误差，我的移动机器人可以水平达到0.1毫米。实际上，可以使用其他传感器和程序来纠正这些错误。希望对你有帮助:-))

[@chrisalbertson5838]

Have you considered using a planetary reduction drive stepper motor? You get 25Nm at the output shaft continuous or 40Nm momentary on a 50x50x120mm package and use only about 2 amps. Cost is about $80. Stepper motors have maximum torque at zero RPM which beetermatches your testing read this spec sheet www.omc-stepperonline.com/download/23HS22-2804S-HG100.pdf

[@christophemalvasio5569]

you should use gears instead of belts/pulleys : you did reach the belts limit in this video also again if you put all motors in the base no torpue used to move motors in the arm ;)

[@exol511]

I do not believe the belts are the problems (toothet belt holds very high torque without streching). Do also consider that belt drives are fairly easy to reduce backlash compared to gears. To correctly use gears the entire construction needs to be way more stiff than a 3d printed one.

[@christophemalvasio5569]

@@exol511 you are right on backlash but the belts used here don't look good

[@exol511]

@@christophemalvasio5569 I think that has mostly to do with the fact that the belt is not tensioned on both sides on the second stage of the reduction (which is also disproportunate). Also the belt looks like a basic T5 belt which has way more strech then a GT3/5 belt for example or some other high torque belts. In general the entire construction is not stiff enough to make maximum use of a belt or gear transmission

[@ramadavince6611]

Increase your bldc to 6384 or 80100

[@justtestingonce]

O Gravity! O Gravity!

[@ulissessalvador]

O problema desses motores e que tem muita velocidade mas tem pouco força

[@grzegorz2169]

You should use stepper motor for this application. Small BLDC motor has high power, but look at RPM vs Power graph for your motor. This motor is designed to work at high RPM, so you need use gearbox with greater gear ratio. This small motor has also low torque, so it can’t handle this load. I started build my own robotic arm an I decided to use stepper motor. If you use cheap powerful BLDC motor you have to use expensive gearbox with high gear ratio. I think you made mistake calculating torque. Your arm has also mass so real torque is lower.

[@exol511]

There are multiple reasons why use servo motor. You a closed loop system that will be able to detect stalls, compensate for errors and position more accurately than a stepper would. While there are so called closed loop steppers those are basically just used to detect stalls and have no compensation or error correction (also they are quite expensive, while this on the other hand is not). Do also consider that with encoders on the robot arm you can back drive the actuator to a position you want by hand and then save that position by reading the encoder, which is a far simpler and elegant solution.

[@grzegorz2169]

@@exol511 Yes, I know servo is much better, but is also expensive. I have stepper motor + driver, it cost me around 200 USD. It generate 3.1 Nm at 200 RPM, 2.1 Nm at 600 RPM. For gear ratio 1 : 15 I have output torque around 30Nm and rotation speed 360°/1.5s. This is quite good torque and speed for this price.

[@exol511]

@@grzegorz2169 thing is he already did a build with stepper motors and they added too much weight. Do consider that the actuator will be 3d printed, thus the entire construction is way too flimsy to support the heavy weight of stepper motors on top of the payload. Also Odrive is basically the cheapest servo drive on the market at 119 euro, the brushless motors and encoder are also fairly cheap and light, compared to a stepper.

[@Skyentific]

@Grzegorz Pierog And what is the weight of your stepper motor? I think it is close to 2kg. Which is way to heavy for my project. That is way I want to use a brushless motors. And this is way on the industrial robot arms they also usually use brushless motors.

[@grzegorz2169]

​@@Skyentific Yes, it is heavy. But you said that gear ratio is 1: 15. This is good for stepper motor not for smal BLDC. For example small BLDC can produce 0.1 N at 3000 RPM. For robot actuator 360°/1 s is enough. In this case gear ratio is 50: 1, but output torque is 5 Nm. You didn't say how much torque can produce your engine. I think not too much, but it can rotate very fast. So you should use motor with very low KV (has lower speed), and you need bigger gear ratio.

[@Vod-b4n]

простой червячный редуктор будет тех же габаритов и, как минимум, с теми же характеристиками...есть безлюфтовые...

[@СергейКравцов-п5т]

Сперва сделай этот самый червячный редуктор, умник

[@Vod-b4n]

@@СергейКравцов-п5т злой какой, эти штуки покупаются

[@СергейКравцов-п5т]

@@Vod-b4n ни разу не злой. И не покупаются такие редукторы, в лучшем случае вместо червячного колеса будет косозубое, а редукторов с настоящими червячными колесами не найти, а если найти, то будет стоить, как крыло от Боинга. Помимо этого у него очень много вопросов к точности изготовления. А тут печатная деталька, чтоб поиграться и потрогать конструкцию. В принципе в этом и смысл канала. Тут, судя по акценту, наш бывший соотечественник, все комменты с лучами добра и только наши соотечественники льют критику. Ну похвали ты, человеку приятно будет на родном языке прочитать комент, судя по его технической подготовке, он как минимум догадывается о существовании червячной передачи и без твоих бесценных подсказок. Будь добрее и скромнее

[@СашкаБелый-ч6м]

А тоже самое на русском языке где-то есть? Жутко интересно! И мужик очень творческий.

[@Skyentific]

Спасибо за комментарий. На русском к сожалению нет. Надо мне попробовать субтитры на русском добавить.

[@vladimirzaguliaev6023]

Есть классный проект OpenDog. Посмотрите. Только он тоже на английском

[@SequoiaAlexander]

Hey this is great work! I also work on 3D printed brushless motor joints (see my channel if interested) and I am so happy that others are working on solutions to this important problem! Thank you for sharing your work!

[@skyhacker6]

gear + wormscrew = better torque and none reversible mecanism. The belt style assembly is too fragile.

[@Skyentific]

I would like to have reversible mechanism (to be able to detect the forces, using the current on the motor).

[@vladimirzaguliaev6023]

Так и не понял как он из обычного мотора сделал шаговый :(

[@DigitalWorldBG]

Там енкодер есть с обратная связь.

[@СашкаБелый-ч6м]

1) Как ни странно, но ВСЕ стабилизаторы для камер/смартфонов тоже всегда сделаны на бесколлекторных моторах! Может они лучше удерживают позицию? 2) Бесколлекторный мотор точно нельзя назвать "обычным мотором", ведь там специальный контроллер, да и внутри они идейно схожи, с той лишь разницей что бесколлекторные расчитаны на большие обороты и работают по трем толстым проводам, а шаговые расчитаны на маленькие обороты и управляются по четырём тонким проводам.

[@DigitalWorldBG]

Крутящий момент - ето важно! У постояно-токовой мотор независимо он безколекторной или не, он на много выше чем у шаговой мотор и ето самоя важная разница, но и управление боле сложное и без енкодер не обойтись.

[@vladimirzaguliaev6023]

@@DigitalWorldBG Спасибо за ответ. Буду искать теперь энкодеры

[@Skyentific]

Благодаря энкодеру и специальному контроллеру (ODrive). У меня есть видео про этот ODrive.

[@nezhinHizZ]

Like, if the author is similar to Javier Bardem :)

[@fugufish247]

Get a lower Kv motor!!!

[@Skyentific]

It is already low KV motor (350).