The ideal servo for robotics
Thomas Arundel
robotics hobby market, with 360 degree rotation and feedback provided via an
optical encoder - target prices about $20-25 each using miniature precision
motor / gearbox combos similar to these:
That we're now starting to import from Asia. As well as traditional PWM
encoding, we could also consider encoding schemes like i2c. We're planning
on designing some even smaller models based around mobile phone / pager
vibrator motors.
It would seem that many people use RC servos, though there seem to be some
compromises, and I think we can improve on weight and torque as well as
offering arbitrary positioning.
I'd appreciate some thoughts, about the shortcomings of RC servos, and on
which areas we should focus to design a servo that's better for building
robots...
Also, does anyone know of any linear servos? Do you think those would be a
popular product, and if so in what sort of size / configuration?
Cheers,
Tom.
--
Thomas Arundel
TL: +44 (0) 1932 252482
FX: +44 (0) 870 137 6005
EM: t...@fastcomponents.co.uk
-----------------------------------------
Fast Components Limited
www.fastcomponents.co.uk
steamer
some robust stuff for "real world" bots please! :-)
--
"Steamboat Ed" Haas : A greasy donut, a cup of
Hacking the Trailing Edge! : coffee and thou...
http://www.nmpproducts.com/intro.htm
---Decks a-wash in a sea of words---
Gordon McComb
bit too slow; the exporter ought to be able to supply some other gear
ratios. I know there is a market for small motors for micro and nano
robots, but I think it's a less significant market than desktop robots
that use standard servos. If anything, people tend to ask for servos
that are beefier and faster that the typical Futaba S-148 (or knockoff).
That makes the target >1 rps, and >56 oz-in (at 6vdc).
If you can manage the motor and optical encoder for $20-25 that would
place it at a competitive level with a $10 servo and a $10-15 (per side)
encoder setup, but not a jaw-dropping price. Remember that an R/C servo
doesn't need an H-bridge, so while control is not as precise as regular
PWM, there's less to design, and overall less cost. RC servos are
popular not so much because of their features, but because of their low
cost, plus ease of use.
Linear servos: I've seen retrofits for taking a standard servo and
making it linear, but unless you need the more precise movement, most
people just use a bell crank, and live with the (somewhat) non-linear
positioning.
-- Gordon
Thomas Arundel
news:Lbv3f.2272$Aw.4...@typhoon.sonic.net...
> --Small is cute, but spare a little of your talent to design
> some robust stuff for "real world" bots please! :-)
Hmm I'm starting to think that a retrofit board that converts geared motor
units into precision servos might be more marketable and useful product,
than a unit reliant on one particular size of motor / geartrain...
What sort of sizes and specs are you thinking of?- we can source much larger
gearboxes too - high quality metal geartrains... :-)
--
Thomas Arundel
EM: tom.SPA...@fastcomponents.co.uk
-----------------------------------------
Fast Component sets - the best way to buy components!
Thomas Arundel
> Speed-wise I think you need to go no less than 1 rps. These motors are a
> bit too slow; the exporter ought to be able to supply some other gear
> ratios. I know there is a market for small motors for micro and nano
> robots, but I think it's a less significant market than desktop robots
> that use standard servos. If anything, people tend to ask for servos
> that are beefier and faster that the typical Futaba S-148 (or knockoff).
> That makes the target >1 rps, and >56 oz-in (at 6vdc).
So we're getting about 7-8 other types with different speed outputs in the
next 14 days. We're also going to custom make some of these with higher
speed outputs. Someone wrote to me asking for 250 RPM - if we were to do say
80, 160 and 250 RPM would that suffice, or is that a bit overkill? I
figured these would be great for hexapods and the like...
Interesting what you say about futaba servos. Any idea how fast they are /
torque output?
What about mechanical specs like volume - is the RC servo an optimal design,
or do you think it could be improved on for robot applications?
>
> If you can manage the motor and optical encoder for $20-25 that would
> place it at a competitive level with a $10 servo and a $10-15 (per side)
> encoder setup, but not a jaw-dropping price. Remember that an R/C servo
> doesn't need an H-bridge, so while control is not as precise as regular
> PWM, there's less to design, and overall less cost. RC servos are
> popular not so much because of their features, but because of their low
> cost, plus ease of use.
This is true. Have to be honest that most servos I've seen here in Europe
are more like $15 for a standard size. I think that we might be better off
developing an after-market system to turn a wide range of geared motors into
servos, rather than limiting the market and tying the two together from day
1.
steamer
: Hmm I'm starting to think that a retrofit board that converts geared motor
: than a unit reliant on one particular size of motor / geartrain...
: What sort of sizes and specs are you thinking of?- we can source much larger
built in gearbox, just cuz they've got lotsa torque and they're easily
obtainable from several surplus dealers..
Thomas Arundel
> A 360 degree servo sounds like something I would buy. Especially at the
> price point you suggest. Positional feedback is a nice feature too, but
> I think the resolution would need to be very high. An I2C interface is
> also a great feature. All in all I think you've got a good idea if you
> can meet that price point.
I think it's do able - I think we can also get such a system to report back
on output torque and motor currents for 'free' too. As for resolution, any
ideas what sort of angular position accuracy / resolution one can expect
from a RC servo unit?
If we were to consider a 100 rpm motor/gearbox, I don't think setting the
output shaft to within 10 degrees of a set position would be overwhelmingly
hard, which isn't bad given that you'd probably want to gear that down again
by a factor of at least ten if you were doing something like robot arm
positioning work. That means one could achieve an accuracy of 1 degree on a
10 rpm shaft - do you think that would be enough?
Another feature would be that one could instruct the unit to exert a
specified torque at its output. do you think that would be useful?
Gordon McComb
> > That makes the target >1 rps, and >56 oz-in (at 6vdc).
> Interesting what you say about futaba servos. Any idea how fast they are /
> torque output?
The specs I noted are those for a Futaba S-148, operated at 6vdc. Servos
based on the S-148 (Parallax's pre-modified servo, for example), or
modelled after it (GWS S03N, for instance) have the same specs.
>
> What about mechanical specs like volume - is the RC servo an optimal design,
> or do you think it could be improved on for robot applications?
The flat body is a plus for down-and-dirty mounting using glue,
double-sided tape, or Velco or Dual Lock. Remember that for the US we
don't have lots of metrix precision screws laying around, and I bet any
motor from Asia will have metric pretapped threads. RC servos just have
two or four holes for mounting with any hardware. Consider all mounting
slop. With a standard motor ou must fairly accurately match up the
mounting holes on the face, or else come up with some type of
alternative case-mounting system, like the Solutions Cubed EZ Roller
motor.
Finally, consider mounting things to the shaft of the motor. You can
attach gears, hubs, bell cranks, or whatever to the servo by fashioning
it against a common (and often, included) servo disc. For a motor with a
shaft you need a hub of some type, typically at an additional cost.
Sources here charge about $5 for a pair of aluminum (al-u-minium in the
UK <g>) hubs for 3 or 6mm shafts.
-- Gordon
Thomas Arundel
> based on the S-148 (Parallax's pre-modified servo, for example), or
> modelled after it (GWS S03N, for instance) have the same specs.
Yes I took a look at those. I'm confident that we can design some gearbox
motor configs that are as strong by 1/3 or 1/2 the weight. How important is
a plastic case - without it things will be cheaper and lighter. Don't
suppose too many people do mud wrestelling with their robots... yet.
> The flat body is a plus for down-and-dirty mounting using glue,
> double-sided tape, or Velco or Dual Lock. Remember that for the US we
> don't have lots of metrix precision screws laying around, and I bet any
> motor from Asia will have metric pretapped threads.
You guys will just have to buy one of our forthcoming metric screw sets then
;-)
> RC servos just have
> two or four holes for mounting with any hardware. Consider all mounting
> slop. With a standard motor ou must fairly accurately match up the
> mounting holes on the face, or else come up with some type of
> alternative case-mounting system, like the Solutions Cubed EZ Roller
> motor.
Yes, we'll come up with something suitable. Someone mentioned to me that a
servo with dual shafts would be very useful. I think that's doable too.
> Finally, consider mounting things to the shaft of the motor. You can
> attach gears, hubs, bell cranks, or whatever to the servo by fashioning
> it against a common (and often, included) servo disc. For a motor with a
> shaft you need a hub of some type, typically at an additional cost.
> Sources here charge about $5 for a pair of aluminum (al-u-minium in the
> UK <g>) hubs for 3 or 6mm shafts.
Yes - we'll start to look at puting together a mounting set for our current
gearboxes straight away.
John Mianowski
<t...@fastcomponents.co.uk> wrote:
>> The specs I noted are those for a Futaba S-148, operated at 6vdc. Servos
>> based on the S-148 (Parallax's pre-modified servo, for example), or
>> modelled after it (GWS S03N, for instance) have the same specs.
>
>Yes I took a look at those. I'm confident that we can design some gearbox
>motor configs that are as strong by 1/3 or 1/2 the weight. How important is
>a plastic case - without it things will be cheaper and lighter. Don't
>suppose too many people do mud wrestelling with their robots... yet.
>
Some of us abuse the heck out of servos & other equipment:
http://www.ntxbg.org
http://www.rcwarships.com
JM