best TORQUE-WEIGHT RATIO

[colin]

Hi,
Im considering a direct drive for an actuator.
mainly to do away with backlash and anti backlash gears etc,
presumably there will be less noise too.

I worked out I need at least 30NM of torque,
wich is a lot more than most motors, (~x100)
but only about as much as a starter motor.

but it also has to effectivly lift its own weight at the end of the arm.

pancake motors seem to be quite efficient,
but i got stuck trying to work out the optimum format,
obviously diameter as large as posible wich can be 200mm
but how many steps is optimal ?

if the force on a wire is proportional to current and field strength,
how does the number of poles affect this presuming the total wire length
remains about the same ?
yet stepper motors seem to have more torque the more steps it has.

I did quite well at electric machines at uni,
but that was ages ago.

Il probably need several layers, ive got a collection of
neodinium magnets, wich im going to try play about with.

you can also get fairly large discs,
is it practical to remagnetise these ?

Colin =^.^=

 

[colin]

The torque/copper ratio is not so much affected by number of poles for
a normal motor design. With fewer poles you will need more iron, however,
since the field lines become longer. On the other hand, at a fixed and
high speed, there is more iron loss for multipole motors, since there are
more field reversals per turn. So efficiency normally drops with nr of
poles for high speed motors at fixed weight.

But for maximum torque/weight at low speed, use as many poles as possible.

Stepper motors are made differently, but they are still extreme cases of
low speed and high torque.

thanks, I'm realy thinking of a stepper pancake type motor,
hopefully to get the benefit of both, the point about iron path length is a
good one,
and the speed is extremly low as there is no gear reduction.

I cant quite get my head around what makes a stepper motor more
weight/torque efficient
I know the construction of them is quite different as I have taken many
apart.

reluctance type motor is easier to see why increasing poles increases
torque.

maybe a similar reasoning can apply to the types of configuration of PM
steppers ?

Colin =^.^=

 

[Sven Wilhelmsson]

thanks, I'm realy thinking of a stepper pancake type motor,
hopefully to get the benefit of both, the point about iron path length is a
good one,
and the speed is extremly low as there is no gear reduction.

I cant quite get my head around what makes a stepper motor more
weight/torque efficient
I know the construction of them is quite different as I have taken many
apart.

The key point is the 'air' between the cogs. It does about the same job as
copper & current, however much better. To have one tesla over 1 mm airgap
would require about one kA, and it is hard to get such a high current
through copper in this very small volume. So a 'normal' motor with 100
poles is not practical, unless the diameter is very big.

 

[Phil Hobbs]

Hi,
Im considering a direct drive for an actuator.
mainly to do away with backlash and anti backlash gears etc,
presumably there will be less noise too.

I worked out I need at least 30NM of torque,
wich is a lot more than most motors, (~x100)
but only about as much as a starter motor.

but it also has to effectivly lift its own weight at the end of the arm.

pancake motors seem to be quite efficient,
but i got stuck trying to work out the optimum format,
obviously diameter as large as posible wich can be 200mm
but how many steps is optimal ?

if the force on a wire is proportional to current and field strength,
how does the number of poles affect this presuming the total wire length
remains about the same ?
yet stepper motors seem to have more torque the more steps it has.

I did quite well at electric machines at uni,
but that was ages ago.

Il probably need several layers, ive got a collection of
neodinium magnets, wich im going to try play about with.

you can also get fairly large discs,
is it practical to remagnetise these ?

Colin =^.^=

Given the limitations of cooling the copper and of the maximum available
B fields in permanent magnets or electromagnets, the torque of an
ungeared motor is pretty well proportional to its volume. The reason
for this is that the total force available is proportional to the area
of the armature (maximum current density times maximum B field times
area), and the torque is equal to force times perpendicular distance
from the shaft. You can get factors of 2 by going to rare-earth magnets
or water cooling or intermittent service, but otherwise this simple rule
is surprisingly accurate.

The best way to get lots of torque from a small motor is to run it at
high speed and gear it down, unless you really can't manage the noise
and slight backlash, in which case you sort of have to take what you can
get.

Cheers,

Phil Hobbs

 

[colin]

Given the limitations of cooling the copper and of the maximum available B
fields in permanent magnets or electromagnets, the torque of an ungeared
motor is pretty well proportional to its volume. The reason for this is
that the total force available is proportional to the area of the armature
(maximum current density times maximum B field times area), and the torque
is equal to force times perpendicular distance from the shaft. You can
get factors of 2 by going to rare-earth magnets or water cooling or
intermittent service, but otherwise this simple rule is surprisingly
accurate.

yes I would go along with this,
however different types of motor have very different torque/weight
ive looked at quite a few, and some have as much as 4NM/kg
wich would mean it supports its own weight at the end of a ~ 400mm arm.

The best way to get lots of torque from a small motor is to run it at high
speed and gear it down, unless you really can't manage the noise and
slight backlash, in which case you sort of have to take what you can get.

im looking for quite accurate positioning,
so any backlash would cuase problems,
especialy in a positional feedback loop,
I have had the idea of having an aditional geared motor
to take most of the weight
but with some springiness to allow the direct drive motor
to control the fine movement.

the direct drive motor would only need to overcome the inertia
the geared motor would strive to keep the direct drive motor at zero load,
so the direct drive motor could have quite a low duty cycle.

2 small motors and simple gearing compared to one probably huge and unwieldy
motor.

Colin =^.^=

 

[Rich Grise, Plainclothes Hippie]

pancake motors seem to be quite efficient,
but i got stuck trying to work out the optimum format,
obviously diameter as large as posible wich can be 200mm
but how many steps is optimal ?

Are you talking about a mounting scheme for something off-the
shelf, or are you talking about inventing a new motor?

Thanks,
Rich

 

[colin]

Are you talking about a mounting scheme for something off-the
shelf, or are you talking about inventing a new motor?

an off the shelf motor would be nice,
im not sure id actually be inventing a new motor,
that would be like inventing a new type of wheel lol,
im just not convinced the ones ive seen available so far
are as light as they could be for a certain torque.

might be a while before i go through some text books
and remind myself of the maths.

im still sure the more poles the better,
as there is less rotational distance between the poles being aligned and
not,
if the energy required to sepereate the poles is the same this means more
torque,
although the degree of seperation would also change a bit with the number of
poles.

the monting scheme is nothing to fuss over tbh.

Colin =^.^=

 

[Rich Grise]

an off the shelf motor would be nice,
im not sure id actually be inventing a new motor,
that would be like inventing a new type of wheel lol,
im just not convinced the ones ive seen available so far
are as light as they could be for a certain torque.

might be a while before i go through some text books
and remind myself of the maths.

im still sure the more poles the better,
as there is less rotational distance between the poles being aligned and
not,
if the energy required to sepereate the poles is the same this means more
torque,
although the degree of seperation would also change a bit with the number of
poles.

the monting scheme is nothing to fuss over tbh.

Well, not that I know anything about motors, but how about a disk-shaped
rotor, with supermagnets arrayed around the circumference, and a stator
that's a bunch of electromagnets that drag a magnetic field around the
circle?

Thanks!
Rich

 

[colin]

Well, not that I know anything about motors, but how about a disk-shaped
rotor, with supermagnets arrayed around the circumference, and a stator
that's a bunch of electromagnets that drag a magnetic field around the
circle?

yeah thats sounding exactly like a pancake motor,
also looked at pcb motor wich again are pancackish.

basicaly it needs large air gap area with large effective diameter,
and minimal iron path length, maximum steps etc ...
multiple discs would give good surface area for the volume.

by supermagnets I assume you mean neodymium ?

Colin =^.^=

 

[Phil Hobbs]

yes I would go along with this,
however different types of motor have very different torque/weight
ive looked at quite a few, and some have as much as 4NM/kg
wich would mean it supports its own weight at the end of a ~ 400mm arm.



im looking for quite accurate positioning,
so any backlash would cuase problems,
especialy in a positional feedback loop,
I have had the idea of having an aditional geared motor
to take most of the weight
but with some springiness to allow the direct drive motor
to control the fine movement.

the direct drive motor would only need to overcome the inertia
the geared motor would strive to keep the direct drive motor at zero load,
so the direct drive motor could have quite a low duty cycle.

2 small motors and simple gearing compared to one probably huge and unwieldy
motor.

Colin =^.^=

A coarse-fine strategy like that would probably work fine, but it'll be
more complicated to get working. Maybe try a preload spring to make
sure you're always at one end of the backlash travel?

Cheers,

Phil Hobbs

 

[colin]

A coarse-fine strategy like that would probably work fine, but it'll be
more complicated to get working. Maybe try a preload spring to make sure
you're always at one end of the backlash travel?

yes, basicaly im interested in trying to come up with new ideas,
theres quite a lot of antibacklash options :-

very precise machining lowers backlash considerably,
springs can take up the slack, but this increases the torque needed,
a strong spring would be needed to always keep at one end of the backlash,
the weight of the arm would tend to do this to some extent,
vibraton might still defeat this.

a counterweight can reduce the holding torque needed
but increases the inertia, a spring (like on an angle poise lamp)
would have to take into account the orientation.

other ideas come from an old disc drive head actuator wich had
flexible thin steel strips as belts wich were fixed to each
pulley so they cldnt slip, limit is about 1 turn of the motor.

Colin =^.^=

 

[Phil Hobbs]

yes, basicaly im interested in trying to come up with new ideas,
theres quite a lot of antibacklash options :-

very precise machining lowers backlash considerably,
springs can take up the slack, but this increases the torque needed,
a strong spring would be needed to always keep at one end of the backlash,
the weight of the arm would tend to do this to some extent,
vibraton might still defeat this.

a counterweight can reduce the holding torque needed
but increases the inertia, a spring (like on an angle poise lamp)
would have to take into account the orientation.

other ideas come from an old disc drive head actuator wich had
flexible thin steel strips as belts wich were fixed to each
pulley so they cldnt slip, limit is about 1 turn of the motor.

Taut band drive can go further than that if you do it right--stiffer
than belt drive, less backlash than gears.

Cheers,

Phil Hobbs

 

[colin]

Taut band drive can go further than that if you do it right--stiffer than
belt drive, less backlash than gears.

yes im sure it could, the most that is needed is almost one turn on the
larger pulley, the band can wind itself down the motor shaft so it doesnt
overlap wich would cuase ratio problems, a guide on the larger pully might
be
a help.

ive got some shim steel wich is thin and flexible, its only 12" long,
but might be long enough given that 2 lengths can be used.

this probably seems the most atrractive I wonder what the pitfalls
are to avoid ...

another advantage is the motor can be closer to the sationary end
of the arm.

Colin =^.^=

 

[colin]

The key point is the 'air' between the cogs. It does about the same job as
copper & current, however much better. To have one tesla over 1 mm airgap
would require about one kA, and it is hard to get such a high current
through copper in this very small volume. So a 'normal' motor with 100
poles is not practical, unless the diameter is very big.

well I dont think ive got any fresh ideas to come up with
and I dont think direct drive is going to be so advantagous,
even without doing the maths,( although i may look into it
just to satisfy myself i can stil do it lol),
except for rotation that isnt fighting against gravity.

so now im looking at existing robot arm actuators.
there doesnt seem to be much details about the construction of
commercially available ones, maybe they like to keep it secret
although you would think someone might want to boast about
how their design is better.

plenty of hobby details about though,
plenty of varations on gear reductions,
steel belt fixed to puley at the top of my list atm.

Colin =^.^=