I'm contemplating an application for a servo. How do I figure out how
much torque I'll need? The application is for twisting tempered steel
wire into a twisted knot. You've probably seen a knot-type wire wheel
brush. I do this with hydraulics now but I know it's WAY overkill.
Then I need to figure out if I should use a gear box or a timing belt.
Hydraulics are persnikety, the oil has to be just the right operating temp.
On Wed, 28 Dec 2011 12:59:02 -0500, Tom Gardner wrote:
Measure the torque required, by means fair or foul.
For me this would mean making some sort of fixture (or using an old
machine, if you have one) to do the job close to how you do it now, then
use an indicating torque wrench or a torque watch (possibly home-brewed)
to measure what it takes to make it happen.
Then multiply by 2, and use that as a minimum. Depending on how much
speed you want, you very likely want to gear things down considerably --
electric motors are usually at their best when they're going as fast (or
faster) than the manufacturer recommends. You really only want to direct-
drive things if it's going pretty close to the motor's maximum speed
already, or if you have a need for light, well-controlled torque.
(You should already know this, but if you don't -- take the efficiency of
the gearbox into account. Nothing's more embarrassing than leaving out
the loss of a component in your calculations, and coming up short in the
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
Why not get a torque meter and measure the torque? You can make a crude
one with an arm and a weight or fish scale. My understanding is the torque
will be about the same at any speed to perform this kind of task.
Most likely a gearbox with a relatively low ratio will be needed.
Possibly belt reduction can do it, also.
Right. That's how a Prony Brake dynamometer works. I've made several
of them for measuring torque on small electric motors.
I've used a DC motor, running as a generator, into a big variable
resistor as a brake. That's how robotics hobbyists often do it --
there's a whole book written about those things.
The first one I made, back in the '60s, was for the purpose of
measuring torque on small racing outboard motors. For that, I used the
classic Prony Brake setup, right on the prop shaft, with a flywheel
and a piece of automobile brake lining material running on the outside
of it, with a lever to adjust the braking. It was a little rough,
pulsing with the cylinder firings, but it did work. And the spray from
the cooing water outlet didn't make things any easier...
Using a motor/generator with a resistive load works much better. You
don't need to measure the output of the generator; its only purpose is
to provide a load, adjusted until the spring arm doesn't move and the
measured spring rate is stable. You need to arrange some stops for the
arm so you don't wreck your spring scale or break your knuckles. d8-)
>Most likely a gearbox with a relatively low ratio will be needed.
Ha! Here's a really nice homemade Prony Brake, a much nicer version of
the one I made in high school to measure outboard motor torque:
If you google Prony Brake you'll also see some that use DC motors as
generator/brakes. The simplest brakes are just two spring scales, a
leather belt, and a small flywheel:
Possibly a dumb question, not seeing the machine, but why is a servo
required at all? Twisting wire knots would seem to be a "rotate x times"
thing that would be simple limit switches, not requiring any sort of
The bit about the current hydraulics and temperature sensitivity imply
proportional hydraulic servo valves. Without seeing the machine I'm
having a hard time picturing why such would be needed for what would
seem to be a simple "twist it until you hit the limit switch" type of
Tom, if I were doing it, I'd build a simple test apparatus to twist the
knot, and apply a torque wrench to do the job. I think you'll find that
the twisting torque increases significantly with the amount of twist, and
that it rises asymptotically to vertical as you approach the maximum
twist the wire bundle can withstand while still "chucked" at both ends.
A torque wrench would give you close approximations of both the starting
and ending torques, and give you an idea of what portion of the duty
cycle is occupied by what torques. From that information, you can deduce
whether to get a motor/drive train that covers the whole envelope within
it's 'normal' operating torque, or whether you can over-torque it at a
certain duty cycle without harm.
"Lloyd E. Sponenburgh" <lloydspinsidemindspring.com> wrote in message
I suspect it doesn't increase much because the metal is being strained
beyond its yield point where the stress-strain curve flattens out. If it was
in the linear region it wouldn't stay twisted.
I just went down to the sekrit laboratory and twisted some 1/16" soft iron
wire with safety wire twisting pliers. The wire started to twist at ~25 Lbs
pull, rising slowly to 50 as the coils wound tight.
The twisting nozel has to realign itself perfectly with the disk at the
start of each cycle and varying the speed in different parts of the
twist would have a great advantage. I assumed a servo motor would be
the way to go and there would have to be two more servos, one for
indexing the disk and one for the pull-back. A stepper could work for
Sure sounds a like a good servo application to me. Its been too many
years but I used to work with a company, Parker Compumotor, for such
applications. Their application engineers were the best. A few phone
calls and we'd have the right equipment selected. Later, at machine
startup, they'd all but write the machine code to run their motors.
I'd at least ask if they still provide such excellent service.
Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here.
All logos and trade names are the property of their respective owners.