Thanks for this. My gut was telling me not to trust the torque and this example encouraged me to do more research, which has lead to figuring out a lot of unknown unknowns. Also, nice robot!
The torque rating is max efficiency output before complete failure. It's hard to duplicate on the bench, especially with no load. It can handle a 35kg load before failure, that's not the same as outputting a 35kg load. Speaker amplifiers are the same way.
Have you tried implementing springs to make the load more balanced? Like, instead of pulling up the entire leg by only using a servo motor, implement a spring that holds the limb at the top most position, and let the servo rest. You could even implement a pull spring and have it function as a 'knee'.
I haven't tried that yet, and seems very interesting. My concern is that while the spring will help in one direction, it may add load in the opposite direction? But honestly I haven't thought it through yet and it may really help. This makes a great case to do some simulations...
@@MakeYourPet It would only balance the load, making the motor 'struggle' equally in both directions. But, you do need to know the physics behind the spring, sooo...
Spring physics is actually not that complicated (in most cases), and there's a linear relationship between the force and the spring displacement. www.jamesspring.com/news/spring-constant/
Yes, and I really want to try one of those in the future. They don't give you a complete 3D freedom really, but probably can be very useful to cut down the costs.
That's true, I've lost like half a year trying to find the problem in code, power supply and schema, until I found out that claimed torque is just a big lie. Now I'm always checking the real torque before making the project Also big issue is that servos accuracy drops dramatically when torque comes closer to its limits, so I'm going to attach angle sensors to work closer to servos limits. Btw, how much ur robot weights? It looks like 1.5 kilos maximum
Ok, so I see that you're calculating total weight to mean 'total weight of the assembled foot', but in practice isn't the load required by each joint closer to the end that much less? Meaning that you could use weaker (and thereby cheaper) servos near the feet?
Yes, you are right. The Tibia servo is under less load compared to the Femur servo. That's why I am consider the Femur servo (max load) here. The Coxa servos rotate orthogonal to the load so they can have much less torque too. So the load from less to most is like: Coxa < Tibia < Femur. I actually mention this in one of my older videos. You can use cheaper servos for Coxa, but usually it is not worth it for the Tibia. But then you have to deal with different sizes and potentially different voltages that complicates the whole design. Total is Total weight of the assembled robot, and in tripod gait 3 legs need to hold the body.
I am new to electronics, but my understanding from this video is that any kind of 35kg Servos will work with this project. But in the video part 1 of 3, testing the cheap servos of 12kg without supporting any major weight on them, they were failing and restarting. I am assuming that those cheap servos didn't work for another issue rather than their supported stall torque weight. right? do we know the reason? I found some cheap digital 35kg servos on Aliexpress (around 18euros) and I would like to know if they are suitable for this or not.
The 12kg servos failed for other reasons. I didn't dig too much into it, but probably because of a wide dead band and low res potentiometer. I was going to make shorter legs and use smaller step sizes to cover for the low torque. Also those were so cheap so were easily replaceable and I wasn't that worried about overloading them.
![Yeat & Summrs - GO2WORK [Official Music Video]](http://i.ytimg.com/vi/NEFEwfPcXRA/mqdefault.jpg)
Thanks for this. My gut was telling me not to trust the torque and this example encouraged me to do more research, which has lead to figuring out a lot of unknown unknowns. Also, nice robot!
This is so friggin cool. I love the videos and just watching your robot move.
Ok made it 6 hours late this time. Thanks for another great video, I was wondering about which servos to get and this video is perfect.
The torque rating is max efficiency output before complete failure. It's hard to duplicate on the bench, especially with no load.
It can handle a 35kg load before failure, that's not the same as outputting a 35kg load.
Speaker amplifiers are the same way.
They just straight lie? We learn something new everyday. Thanks for the head's up.
Good video, thanks for explanations and sharing your experiences
Great concise video!! 🙏
Have you tried implementing springs to make the load more balanced? Like, instead of pulling up the entire leg by only using a servo motor, implement a spring that holds the limb at the top most position, and let the servo rest. You could even implement a pull spring and have it function as a 'knee'.
I haven't tried that yet, and seems very interesting. My concern is that while the spring will help in one direction, it may add load in the opposite direction? But honestly I haven't thought it through yet and it may really help. This makes a great case to do some simulations...
@@MakeYourPet It would only balance the load, making the motor 'struggle' equally in both directions. But, you do need to know the physics behind the spring, sooo...
Spring physics is actually not that complicated (in most cases), and there's a linear relationship between the force and the spring displacement. www.jamesspring.com/news/spring-constant/
I might try that on my underpowered quadruped
Great video 🔥🔥
I have another idea to reduce number of motor in a limb. We can use complaint mechanism to achieve degrees of freedom and it's also 3d printable
Yes, and I really want to try one of those in the future. They don't give you a complete 3D freedom really, but probably can be very useful to cut down the costs.
não vejo a hora de montar o meu ... continue nos atualizando tks
Great information, I probably would have gone with something closer to my estimates.
cool video thanks :)
That's true, I've lost like half a year trying to find the problem in code, power supply and schema, until I found out that claimed torque is just a big lie. Now I'm always checking the real torque before making the project
Also big issue is that servos accuracy drops dramatically when torque comes closer to its limits, so I'm going to attach angle sensors to work closer to servos limits.
Btw, how much ur robot weights? It looks like 1.5 kilos maximum
It is 2.4kg total
Nice ! :) as usual ^^
Ok, so I see that you're calculating total weight to mean 'total weight of the assembled foot', but in practice isn't the load required by each joint closer to the end that much less? Meaning that you could use weaker (and thereby cheaper) servos near the feet?
Yes, you are right. The Tibia servo is under less load compared to the Femur servo. That's why I am consider the Femur servo (max load) here. The Coxa servos rotate orthogonal to the load so they can have much less torque too. So the load from less to most is like: Coxa < Tibia < Femur. I actually mention this in one of my older videos. You can use cheaper servos for Coxa, but usually it is not worth it for the Tibia. But then you have to deal with different sizes and potentially different voltages that complicates the whole design.
Total is Total weight of the assembled robot, and in tripod gait 3 legs need to hold the body.
Hi. thanks for sharing your hexapod; can you tell me if "chipo" can accept other servos than the mg996R? thank you
Спасибо за совет
I am new to electronics, but my understanding from this video is that any kind of 35kg Servos will work with this project. But in the video part 1 of 3, testing the cheap servos of 12kg without supporting any major weight on them, they were failing and restarting. I am assuming that those cheap servos didn't work for another issue rather than their supported stall torque weight. right? do we know the reason? I found some cheap digital 35kg servos on Aliexpress (around 18euros) and I would like to know if they are suitable for this or not.
The 12kg servos failed for other reasons. I didn't dig too much into it, but probably because of a wide dead band and low res potentiometer. I was going to make shorter legs and use smaller step sizes to cover for the low torque. Also those were so cheap so were easily replaceable and I wasn't that worried about overloading them.
kg-cm, not kg/cm!
Wow, you are right! Thank you for pointing that out!