Replacing servos with brushed motors. Some questions!

Another option is to use industrial brushless DC geared motors. They come in every conceivable torque and shaft rpm and they are cheap and strong. With in built hall sensors positioning is no problem. At 3 pulses/rev x 10k motor rpm x again the gearbox ratio even small rotations at the output shaft are possible. They have many advantages for shock applications. They are freewheeling when not engaged but can be locked magnetically for short periods.

I'm using 24vdc motors on a current project so the trick seems to be to control the motor controller rather than the motor itself. Most controllers use 5v for signal purposes so it is possible to drive them straight from the board but I usually use the board to drive an interface to interact with the external system. Ie drive a relay to switch an Enable F/R command.

I haven't looked into reading the hall sensor into the board yet but I'm hoping to be able to jumper from the motor controller to a hall sensor shield into the board. This will then give position control.

Kim

Have you ever heard of something called a “servo saver”?

This is a special control horn that actually has a spring inside, so that if an external force is applied (or the servo overdrives an output that can move any further) - the horn will rotate against the spring, instead of breaking the teeth (or causing other damage). Hence the name.

They are commonly used in the RC hobby for the steering servos on RC cars. You’ll likely only find ones for standard sized and larger servos. They aren’t very expensive (but they do cost more than a standard servo horn).

The concept of a spring as part of a shock absorbing system is common (you still need a way of dampening the spring - ie, in an automobile, you have springs on the axles - but you also have shock absorbers, which are really dampening devices); you might want to research this area, and also research what others have done for such a need on similar research legged robots (there’s tons of stuff out there dating back 50 years or more - all you have to do is look for it).

You might find that you can incorporate a spring system into your leg design that will still act in such a manner to protect the servo, but be more compatible with the layout of your current leg architecture and placement.

@KimLong:
mmm seems to be what I need…

With in built hall sensors positioning is no problem

A) do you mean these motors have a feedback position inside?

B) Do you know some of these motors with size similar to servo size with minimum 13kg of torque?

C) have you some link about Arduino + these motors?

D) What shield are you using? And, very important, do exist some shield that can handle 12 of these motors?

@cr0sh:
These servo are damn cool! and not, never listen about it!
And I think if I used these spring-servo + another spring applied to the robot foot so… probably I would get a pretty cool shock absorber system!
This is a great news!

Hi Aldoz,

A) Yes. The hall effect sensor is built into the motor, not the gearbox so it gives 3 pulses per rev of motor so you can end up with thousands of pulses per rev of output shaft of gearbox.

B) They come in a multitude of sizes/torque/voltage etc. I'm currently using 24Vdc motors that output 2.1Nm @ 750rpm. For your project you would probably have more like 10rpm so your output torque would be aprox 2.1 x 75 would give you 157.5 Nm

C) The best place to find them is on Alibaba. Just search for Brushless DC motor and gearbox

D) I am not driving the motors directly from the board so don't need a shield, other than to control relays to switch the Enable and F/R circuits and to control the pot input for speed control, but rather using the board to control the motor controller. Some motor controllers can control more than one motor however if you want independent control of each motor you would need a separate controller for each.

Regarding positioning you will need to 'jumper' the Hall sensors from the motor controller onto a shield which will then send the info to your board.

I'm only a beginner with this stuff however doing it this way, to my mind, gives you industrial grade control and reliability without the need to stress your board

Kim

KimLong: Hi Aldoz,

A) Yes. The hall effect sensor is built into the motor, not the gearbox so it gives 3 pulses per rev of motor so you can end up with thousands of pulses per rev of output shaft of gearbox.

B) They come in a multitude of sizes/torque/voltage etc. I'm currently using 24Vdc motors that output 2.1Nm @ 750rpm. For your project you would probably have more like 10rpm so your output torque would be aprox 2.1 x 75 would give you 157.5 Nm

C) The best place to find them is on Alibaba. Just search for Brushless DC motor and gearbox

D) I am not driving the motors directly from the board so don't need a shield, other than to control relays to switch the Enable and F/R circuits and to control the pot input for speed control, but rather using the board to control the motor controller. Some motor controllers can control more than one motor however if you want independent control of each motor you would need a separate controller for each.

Regarding positioning you will need to 'jumper' the Hall sensors from the motor controller onto a shield which will then send the info to your board.

I'm only a beginner with this stuff however doing it this way, to my mind, gives you industrial grade control and reliability without the need to stress your board

Kim

Thank you so mutch for these info KimLong, these motors are really interesting. I need just to make some experiments about :)

@cr0sh: Servo Saver. How cool. Did not know these thingies either. Very usefull information. Thank you.

i found another possible alternative: DC Encoder Motors

That seems to be the combination you are looking for: DC Motors with encoders directly attached to them in a small and cheap package. Found them on this page here: http://www.makeblock.cc/dc-encoder-motor-25-6v-185rpm/

Btw., i like their robotics hardware gears and beams and stuff a lot. Looks real nice and gives you a large range of possibly needed parts, so you don't have to worry about this aspect if you dont want to / just can't drill this stuff yourself. And it is even open source! So you can actually produce it yourself if you are into this things, or connect to their system easily: http://www.makeblock.cc/

@cr0sh: The servo saver must be applied between servo horn and the servo motor shaft?

@TOM-d: Really cool motors too! "1A maximum continous current per motor": I am curious to know what torque can be obtained with 1A of current without external gears

KimLong: Another option is to use industrial brushless DC geared motors. They come in every conceivable torque and shaft rpm and they are cheap and strong. With in built hall sensors positioning is no problem. At 3 pulses/rev x 10k motor rpm x again the gearbox ratio even small rotations at the output shaft are possible.

The number of pulses per revolution depends on the number motor poles and you can sense all 3 hall signals to get finer resolution (then there are 6 states per pole-pair per revolution). I've used this before to implement a crude servo-motor system - position accuracy 1/12 or a revolution for 4-pole motor.

The servo saver must be applied between servo horn and the servo motor shaft?

it acts as a replacement, as it takes/blocks the gear/shaft of the servo, so no further joint for another servo horn to attach to. but you can always attach a longer beam to it using the given screw holes on top. i just they give the false impression to be too short because they are so broad and there are surely longer versions somewhere.

TOM-d:
it acts as a replacement, as it takes/blocks the gear/shaft of the servo, so no further joint for another servo horn to attach to.
but you can always attach a longer beam to it using the given screw holes on top. i just they give the false impression to be too short because they are so broad and there are surely longer versions somewhere.

Yes, I find this image showing how to mount a servo saver.
It is interesting how they were conceived with these 3 spring mounted one over the other.
I really would to try. Just a bit difficulty understand what is the right dimension servo saver for my servo…
There are various versions…

Can you (or all you guys) know what is the right servo saver for my HD-1201MG servos from this link?
https://www.servocity.com/html/servo_savers.html

I think futaba standard just cause my futaba S-3000 servo has same dimension of my HD-1201MG but I am not too sure…

The fact is that until servo using toothed gears inside so it's very difficulty to make a decent shock absorber system. And in a quadruped robot, a decent shock absorber system is a really important matter.

Have you tried reducing the shock to the servos? You might try moving the legs at slower speeds, putting shock absorbing pads on the bot feet, making the legs out of more flexible materials, or detecting when the leg contacts the ground and stopping leg motion at that time.

MarkT: The number of pulses per revolution depends on the number motor poles and you can sense all 3 hall signals to get finer resolution (then there are 6 states per pole-pair per revolution). I've used this before to implement a crude servo-motor system - position accuracy 1/12 or a revolution for 4-pole motor.

thanks for clearing that up Tom. My knowledge is very basic on the actual workings. When I get these things to do what I need of them I don't usually delv too much further. Yes, the actual rotational accuracy will depend on the particular motor (number of poles, rpm etc) and the ratio of the final output shaft

aldoz: Thank you so mutch for these info KimLong, these motors are really interesting. I need just to make some experiments about :)

That's exactly what I thought when I came across them!

Thinking of your quad project it would probably mean a re-design, perhaps placing the motors in the body then driving the leg-sections via toothed belt. This would lessen the 'unsprung' weight and reduce the inertia in the legs during motion. Just thoughts, I don't know the design intent so this may not be practical.

@zoomkat:

Have you tried reducing the shock to the servos? You might try moving the legs at slower speeds, putting shock absorbing pads on the bot feet, making the legs out of more flexible materials, or detecting when the leg contacts the ground and stopping leg motion at that time.

Hi, I made a lot of experiments to optimize the shock to the servos; The speed of the legs, in some cases, must to be fast, or better, I prefer fast; My quadruped has various walk modes, including a fast walk that need fast legs movements; or trot mode that need just max speed legs going up and down for half of trot steps movement.

As you advising, the feet of my robot already has a shock abosorb system, in fact the foot is mounted in a side of a piston and that piston can travel inside a oiled cylinder for 1cm. That is a basic but good shock absorber tecnique but I need something over that, something can help ALL the servos of the single leg (3 servos).

Using flexible materials is a test I already made but the weight of the robot force me to use a solid iron arm and an alluminium forearm. The forearm in fact is just just just ok about solidity. If I should choose a softer material so the weight of the robot will deform the forearm joint point (the point where the forearm is attached to the forearm servo). About the detecting of the ground yeah! already made using 4 tactile sensors :) So the sensor (mounted on the foot) will stop (or better, will modify the movement steps) the walk/run/trot movement when a contact with the ground happen! But jumps, or little falls (due to bad/rocks terrain shape or for a lot of possible circumstances) need a shock abosorb system more advanced. So the servo saver seems to be something I can't avoid to use for sure! Thank you for your advice! If you have other advices PLEASE tell me!

@KimLong:

That's exactly what I thought when I came across them!

Thinking of your quad project it would probably mean a re-design, perhaps placing the motors in the body then driving the leg-sections via toothed belt. This would lessen the 'unsprung' weight and reduce the inertia in the legs during motion. Just thoughts, I don't know the design intent so this may not be practical.

yes, if I will to use these motors so I must to change all the shoulder/armpit/basin/groin/arm/forearm configuration! This long restyle time needed to make this work start to make me consider the way to continue to use servos+ servo savers. There are thousand of things I would to make/try but I need to consider my free time/money :(

But REALLY thanks for your advices and considerations! before or later I will try these motors!

You might consider servos with titanium gears like below.

http://www.hobbyking.com/hobbyking/store/RC_PRODUCT_SEARCH.asp?searchType=10&strSearch=titanium+servo&location=US&idCategory=&sortBy=Relevant&NumPerPage=40¤tPage=

zoomkat: You might consider servos with titanium gears like below.

http://www.hobbyking.com/hobbyking/store/RC_PRODUCT_SEARCH.asp?searchType=10&strSearch=titanium+servo&location=US&idCategory=&sortBy=Relevant&NumPerPage=40¤tPage=

Ahh these are pure monster servo! These titanium gears servo + servo saver + actual foot shock absorb system and sure a lot of protection. I think we can't go more far for the servo health :)