Multiple gear motors per output shaft

[ErikY]

Please forgive me if any of my terminology is off, or I word anything incorrectly, in advance.

As I look around for DC motors it is becoming pretty clear to me that I am not going to get the type of torque and speed from a 12V motor with the size requirements that I want.

My question is about using multiple gear motors to generate higher torque without sacrificing as much speed, for one output shaft.

I know that each gear you add costs you about 10% efficiency for one motor per output shaft, so I am wondering if the same holds true for multiple motors per output shaft?

Also, is there a proper way to line up multiple motors per shaft?

I would love to use some more advanced gears, like planetary, but I suspect the cost is going to be outrageous so most likely dealing with spur gears.

[jwatte]

Yes, multiple motors can drive a single shaft, and the efficiency should be the same as a single motor driving. But I don't think I've ever seen this done in reality; I think you always end up with a better solution by just specifying a bigger single motor.

[ErikY]

Yes, multiple motors can drive a single shaft, and the efficiency should be the same as a single motor driving. But I don't think I've ever seen this done in reality; I think you always end up with a better solution by just specifying a bigger single motor.

I would strongly prefer a single motor, the problem I am finding is to get the power I want, I need a much bigger motor than the space  requirements allow for.

Simply as a benchmark, the MX106 has 1,189 oz/in of torque, with 45RPM no load speed. The MX106 motor is approximately 7/8" diameter, by around 1" long not including the shaft.

The gear ratio is 225:1

By comparison, the 37D mm gear motors are far less powerful. Take for example the 100:1

This motor is  almost 2x larger than the MX106 motor

220 oz/in torque, 100rpm no load speed.

If I were to apply a 2.25:1 gear ratio on this motor, I would get the following:

450 oz/in torque at a comparable 45RPM speed.

I assume the Mx106 is probably only using 2 gears, where my calculations are using 4, leading to ~10% loss in efficiency, but the power difference is still about 1/2, for a far bigger motor.

Now, the pololu motor is only $25 so this is not unexpected at all.

Given the price of an Mx106, I would have no problem spending $100 on a motor to get that type of power/size ratio, I just am not finding them commercially available to purchase.

Note that the 106 is not my power requirements by any means, I am really just using that as a benchmark.

So this is why I started thinking about using multiple smaller motors to achieve more torque, although I REALLY don't want to do that.

[jwatte]

Given the price of an Mx106, I would have no problem spending $100 on a motor to get that type of power/size ratio, I just am not finding them commercially available to purchase.

A few more clarifying questions:
1) How come you can you fit multiple smaller motors, but not one bigger motor?
2) What are your actual requirements for cost and power?
3) Have you considered over-volting some of the motors you can find? 1.4* voltage means 2* power.
4) Have you considered all motor kinds: brushless, coreless, etc?
5) What makes you think Robotis can purchase those motors for less than $100? The brand of the motor is actually published in the specs; have you looked for a supplier for that motor?
The Maxon RE-MAX (used in the MX-106) is about $206 in singles straight from Maxon. It's also a 24V motor, so perhaps they are under-volting it, or getting a custom build. When buying in volume, I expect they pay less, but it's quite possible they pay more than $100 even in volume.

I looked into building my own servos, and it just isn't worth it IMO. I'm not going to save a lot of money, and I'm going to waste more than I save in the cost of failed experiments, not even counting the value of my time. Thus, I came to the conclusion that I either find the money to buy the servos I need, or I live with the fact that I can't get the level of power I would like.

[ErikY]

A few more clarifying questions:
1) How come you can you fit multiple smaller motors, but not one bigger motor?
2) What are your actual requirements for cost and power?
3) Have you considered over-volting some of the motors you can find? 1.4* voltage means 2* power.
4) Have you considered all motor kinds: brushless, coreless, etc?
5) What makes you think Robotis can purchase those motors for less than $100? The brand of the motor is actually published in the specs; have you looked for a supplier for that motor?
The Maxon RE-MAX (used in the MX-106) is about $206 in singles straight from Maxon. It's also a 24V motor, so perhaps they are under-volting it, or getting a custom build. When buying in volume, I expect they pay less, but it's quite possible they pay more than $100 even in volume.

I looked into building my own servos, and it just isn't worth it IMO. I'm not going to save a lot of money, and I'm going to waste more than I save in the cost of failed experiments, not even counting the value of my time. Thus, I came to the conclusion that I either find the money to buy the servos I need, or I live with the fact that I can't get the level of power I would like.

1) My main goal is to ultimately build big (4ft+) humanoid robots, design and looks are important to me, and if you have a very long motor, it will make for very wide, unattractive joints (IMO). Ideally, the smaller, the fewer the better, but if given my choice from a design perspective, I would rather have more realistic width joints, and have the motors be running up the limb and covered by connecting materials over making a wide joint, hopefully I am explaining myself properly

2)I would be lying if I threw out any actual numbers right now, so here is my thinking:
Charli is using Mx106's, and it took engineering far better than I could accomplish to support a biped of that size with those servos. Those servos are probably at the absolute high end on the cost spectrum to what I can envision spending in order to buy enough to build the type of bots I want to build. So, given that those servos are at least at the top of my price range, yet not powerful enough to do what I need, I know I need to set my sights higher than seem to be available. The pro line dynamixels would probably do it, but I envision those will cost 3x + what the Mx106's will cost, making them unattainable.

3) No, to be honest, I never even heard of that before as a method, I was under the impression that you would seize up the motors if you do that consistently, and I did not realize it doubled the power. I don't know how you know all this stuff

4) To be honest, my main sources of searching for motors have been: trossen, pololu, robotshop, servo city, amazon, ebay, mcMaster, granger, and from the first 4, the main options are DC brushed motors with or without gears, so my thought was if those are "robot" type shops, those would be the most viable options for motors, and that is more of what I stuck with for searches, and never stumbled upon the other types.

5) I had no preconceived notions on price, other than that I should be able to get it a decent amount cheaper than what it would cost for a Mx106, being that it does not have the encoder, the case, the logic, etc. I had no idea you could even find out the motors used, again, I don't know how you know all this stuff I will try to find those bad boys. Definitely more $$ than I had hoped, but significantly cheaper than an Mx106 itself.

I may come to the same conclusion about being worthwhile, but you kind of got me steered in this direction, when you said usually the process is to understand power requirements, pick the motor, understand accuracy, pick the encoder, and I realized that if I am limited to off the shelf servos, I will always be at the mercy of the market so to speak in terms of what is available.

My thought is that if I can build a repeatable concept, starting relatively cheap, small cheap motors, plastic gears, and if it works, the logic and the methods can be carried out to real projects.

This may be a pointless exercise, but I feel like I need to at least explore it.

[jwatte]

Overvolting: Yes, overvolting will reduce the lifetime of the motor, approximately by an exponential function of the amount of overvolting.

1.4x voltage == double power: Ohm's law says power equals voltage squared over resistance. Resistance doesn't change, so multiply voltage by 1.4 means multiply power by 1.4*1.4 which is approximately 2. (You really want square-root-of-two rather than 1.4)

How to find the internals about the servos: It's listed right on the front page of the product listing on the Trossen Robotics site. It's also available in various Robotis marketing literature.

How do I know things? I read, and remember :-)

[ErikY]

Makes sense on the overvaulting, thanks for the explanation.

I need to be more disciplined when I read. I read a lot, and usually remember what I actually read, but my brain is always onto the next thing. I have been on that page a million times, and never read the motor, and in turn spent rediculous amounts of time looking for motors!

But I digress, that problem is not one that I think anyone on this forum can help me with!

I am sure this will come back to bite me, but how do you know exactly which Re-max it is? I was on there and there were a million of them, did you use the 225:1 gear ratio from the servo info and the final torque/speed to figure it out, or did I miss this somewhere too?

[jwatte]

how do you know exactly which Re-max it is?

I don't know exactly! In fact, all the listed Re-max motors are 24V motors, which makes me wonder whether they got a custom wound version, or whether they use the 24V version and undervolts it for longevity.
You could call Maxon and ask them what they think about your particular application. They probably have sales engineers who answer that precise kind of question!

[ErikY]

how do you know exactly which Re-max it is?

I don't know exactly! In fact, all the listed Re-max motors are 24V motors, which makes me wonder whether they got a custom wound version, or whether they use the 24V version and undervolts it for longevity.
You could call Maxon and ask them what they think about your particular application. They probably have sales engineers who answer that precise kind of question!

Gotcha!

I have been playing with some calculations, and something is clearly off, would you mind taking a look at my calculations and telling e what I am missing?

I am looking at a motor with the following characteristics:

12V
nominal torque: 12.4 mNm
stall torque:  207 mNm
no load speed: 9130 rpm
nominal speed: 8170 rpm

I am using a gear ratio calculator, which asks for the following:

input torque (in/lb's)
input rpm 
Gear Ratio

And spits out an adjusted torque and adjusted RPM

When I look at stall torque, I am converting .207 Nm (from motor specs) to be 1.832104309 in/lb's from a separate converter

So I entered 1.832 in/lb's
9130 RPM's
225:1

And I get the following output:

~ 412 lb's/in stall torque or 6,592 Oz/in
~40 rpm's

This has to be wrong, that is about 6 timex stronger than an Mx106 for $207 at 12V and only 160grams with a 29mm diameter.

What am I doing wrong?

This is the maxon part number if anyone is interested: 226802

[jwatte]

that is about 6 timex stronger than an Mx106

One difference is that the MX-206 is run at 12V, not 24V -- this is a 4x difference assuming it's the same motor.

If you also use a different gear box ratio to get to 60 rpm instead of 40 rpm, you get another 1.5x difference.
4x * 1.5x == 6x, so that would almost exactly explain the difference you suggest.

[ErikY]

that is about 6 timex stronger than an Mx106

One difference is that the MX-206 is run at 12V, not 24V -- this is a 4x difference assuming it's the same motor.

If you also use a different gear box ratio to get to 60 rpm instead of 40 rpm, you get another 1.5x difference.
4x * 1.5x == 6x, so that would almost exactly explain the difference you suggest.

Well, I was doing everything at 12V, using the Mx106 specs at 12V which is 1,189 oz·in stall torque and 45 rpm No load speed.


The motor I was basing this off of was a 12V maxon motor, not a 24, so am I missing something here?

I will be honest, reading the specs from maxon is not like reading specs from pololu or trossen.

[jwatte]

The motor I was basing this off of was a 12V maxon motor, not a 24, so am I missing something here?

It's quite possible that the MX-106 uses a smaller one, then.
You could have MOAR POWAR!!!

not like reading specs from pololu or trossen

Hence, why places like Pololu and Trossen exist. They use their expertise to figure out things to try. Then they try things, some of which work. Then, the things that they try and that work, they actually sell, with packaging/descriptions/support for the hobbyist. They charge a mark-up for this service, but for someone who's not a professional and doesn't necessarily know how to decipher a data sheet or speak to an applications engineer, that mark-up is totally worth it!

[ErikY]

The motor I was basing this off of was a 12V maxon motor, not a 24, so am I missing something here?

It's quite possible that the MX-106 uses a smaller one, then.
You could have MOAR POWAR!!!

not like reading specs from pololu or trossen

Hence, why places like Pololu and Trossen exist. They use their expertise to figure out things to try. Then they try things, some of which work. Then, the things that they try and that work, they actually sell, with packaging/descriptions/support for the hobbyist. They charge a mark-up for this service, but for someone who's not a professional and doesn't necessarily know how to decipher a data sheet or speak to an applications engineer, that mark-up is totally worth it!

Can never have enough powar!!!

Very true, never thought of it that way, but makes a lot of sense.

[DrBwts]

I know that each gear you add costs you about 10% efficiency for one motor per output shaft, so I am wondering if the same holds true for multiple motors per output shaft?

I've never heard this before, where did you get this stat from?

[ErikY]

I know that each gear you add costs you about 10% efficiency for one motor per output shaft, so I am wondering if the same holds true for multiple motors per output shaft?

I've never heard this before, where did you get this stat from?

Actually gear efficiency varies based on the type of gear.

10%  is approximate for spur gears, and others can be less efficient.

I'm pretty sure gear efficiency is a pretty well known topic, but I first read about it on the main SoR site tutorial about gears.

[ROBOT420]

Make any gears you want with this. http://woodgears.ca/gear/index.html    Just make a simulation that you like, print the template, and cut the gear out of whatever material you like. Hope that this was helpful, good luck!