Controllable torque electric motor questions

"Richard Ferguson" wrote: (clip) Oh, and the application is similar to twisting square bar for decorative railings, something common for blacksmiths. Pretty easy to do if you get the bar red hot. (Obligatory metal content). (clip) ^^^^^^^^^^^^^^^^^^ Well, I don;'t know much about the electric motor part, but I am very experienced at twisting square bar for handrails, which, as you say, is the metal content of your post. It's normally done cold, for two very good reasons: 1.) 1/2" square stock, which is the most commonly used size, is not hard to twist cold. 2.) If you try to do it hot, any unevenness in the temperature will cause unevenness in the twist. If this is the type of thing you are planning to do, I wanted to be sure you knew this. BTW, the twisting should be done inside a pipe, to keep the bar from looking like a corkscrew.

I think you could do this with a VFD and an AC induction motor with a gear reducer. I'm not sure I see why controlled torque is in any way necessary. In fact, it seems like a big mistake, as different bars will take different amounts of torque to start to bend. Why not just turn the motor slowly, and stop when you get the required number of turns. Why do you need to reverse? Just to get the two bar clamps aligned so the straight bar can be inserted? No need to reverse, just go to the nearest "index" position. So, it could all be done going in one direction, it seems. When you have twisted the bar and stopped, it will be under torsional load, and may be hard to remove from the clamps. Not a problem on a manual twister, but there may need to be some mechanism to unload that tension so the finished bar can be removed.

As for motor overheating at low speed, just add a fan to force air through the motor.

Jon

The torque of a DCPM (permanent magnet) motor is linearly proportional to its current almost regardless of speed. Some if not most servomotors are designed to deliver a rated torque down to and including zero speed, indefinitely.

If your accuracy requirement is only 10%, then straight current control would probably work fine for you. If you want more accuracy, then you'd have to measure torque somewhere in the system, perhaps with a loadcell, and use that as a feedback signal in a closed-loop control.

One possible error source is the stuff between the motor and the load, e.g. gears and other power transmission components. Their drag torque may not be constant with speed, temperature or over time as things wear and/or the state of lubrication possibly changes. The most accurate and reliable approach would be to control motor current using a feedback signal derived from a torque sensor that meauses actual torque delivered to the load.

One ft-lb of torque at 1000 RPM is about 142 watts or about 0.2 HP. That'll give you a rough idea of what size motor you need -- but you'll want to select a servmotor that can deliver 1 ft-lb continuously at stall. Just one example: Baldor MT-3363-BLYCN can deliver continuous stall torque 11.25 in-lb (almost a ft-lb) when drawing 4.75 amperes. It's rated speed at 100 volts is 2400 RPM, so with slightly more gear reduction than you contemplate this motor would easily do your job. There are many other servomotors available from a number of manufacturers. Google is your friend.

Seems like DC motor is exactly what you want (and I *am* an electrical engineer). A variable- or fixed-current voltage-limited power supply will deliver any torque you care to specify from stall to some very slow top speed.

Not true of many DC motors. They are typically designed to deliver lots of torque at stall or low speed, and to dissipate heat by radiation/convection rather than by fanning themselves. Think of elevators, or industrial conveyors.

Look for a continuous stall torque rating in the specifications. Multiply by your gear ratio. (Note: gearing efficiency doesn't matter at zero rpm.) Find one that delivers the oomph you need.

Your arms only put out fractional horsepower. So a motor shouldn't present much problem replacing them to twist stuff.

His example cited 1000 RPM motor speed with 100:1 reduction.

His specified motor torque (1 lbf-ft) already assumed multiplication by a 100:1 gear ratio.

I did something very like this a long time ago, except it was one way and more continuous. I used a DC motor input to the reducer. Nowadays, I'd use a small AC induction motor and VFD. The torque part was controlled by a magnetic particle slip clutch, which are very linear and repeatable over 0-24 V on the actuating coil. The application was web tension, and the sensor was a roller mounted on load cells. Worked extremely well.

Pete Keillor

I wish I had paid more attention years ago when visiting a fellow in Mass. He had just built a device for twisting bar stock using a pneumatic cylinder to do it. He hit the air valve and instantly 4' of bar stock was uniformly twisted and perfectly straight. Any ideas??? Respectfully, Ron Moore

Unless you need very fast accelerations (which doesn't seem likely), and size the motor conservatively, a standard PM DC gearmotor fitted with an encoder would be adequate. The torque/current ratio will vary with temperature, so you don't want to limit the temp rise of the motor.

Boston Gear can supply a motor/gearhead assembly, or you can buy a C- face motor and gearhead separately. You do need to figure out a way to mount the encoder if you can't get a motor with auxiliary shaft.

I'm partial to Galil controllers and Copley amps. The Galil stuff is expensive, though I bet you can find something on ebay. For example...

I built an automated station for a video cartidge assembly machine that did almost precisely what you've described, but on a much smaller scale. It wound the leader (these were unstuffed cartridges - leader only) all the way onto one hub; applied a specified torque to test the leader attachment to the other hub; reversed and tested the attachment to the other hub; reversed again and positioned the leader for splicing the tape into the cartridge.

Ned Simmons

Ned Simmons

I am not experienced with twisting square bars, but have seen what a local " Wrought Iron " shop used. He used a ordinary induction motor connected to a worm gear reducer. The output of the gear reducer was connected to an automobile transmission. He did not do the bending inside a pipe and they still came out straight. I can't remember exactly the way the square rod connected to the machine, but as I remember it just dropped in a three sided square socket. The distance between the sockets was fixed so when the twisting was being done, the square rod would be getting slightly shorter and would be under some tension. Which is probably why the twisted rods came out straight. The speed was pretty slow, I would guess about 1 rpm. And as I remember he did not use any sophisticated limit switch. Just shut the power off after the rod was twisted the correct number of turns.

So no fancy motor, no fancy turns counter, and the gear reducer was not too expensive as the auto transmission supplied the final drive. I think he may have had a 1/2 hp motor. This is recalled from seeing it a good many years ago.

Dan

Sounds like a job for a large scale RC servo. All the gearing and feedback mechanics are built right in. 6 volts dc. available at any hobby shop for 25 bux

Gearing efficiency _does_ matter at zero RPM. A gearbox's stated efficiency loss is almost entirely due to frictional losses, which you need to take into account going from the torque at the motor to the torque at the shaft. With a 100:1 ratio it'll be hard to effectively control the torque at the output by controlling the torque on the input, unless you pay a lot of attention to the effects of the gearbox on torque.

You are correct that a universal-wound motor would be a bad idea. You probably want a DC servo motor that's sized for the service you intend. As already pointed out there are DC motors that are designed for continuous high torque slow duty.

I don't think that you'll be happy with trying to control torque through a 100:1 gearbox. I could be wrong, and the first thing you may want to do when you get hardware is to give it a whirl, but I think you'll find that the gearbox will introduce too much uncertainty in your torque, even if the motor is doing exactly what you want.

Were I doing this I would go with the suggestion for a load cell. Better yet, you might want to consider a torsion spring on the output of your gearbox. If you could stand the low speed, winding up the spring through the gearbox should give your controller time to react appropriately, and putting rotation sensors on each side of the spring should give you a good indication of torque.

sci.engr.control would be a good place to post this question. There's lots of guys there doing this sort of practical control, so the answers will be based on solid ground.

From your description, I'd look at air cylinders. Air pressure controls torque. have the cylinder drive a rack, rotation on a pinion. KISS

If you can do that mechanically I'll second it. If you only need to rotate through 90 degrees or so a mechanical linkage rather than a rack may be a better idea -- but only if it's simpler than a rack.

Good idea!

If nothing's moving then frictional loss is by definition zero.

Maybe you mean, "stiction"?

True, but if your mechanism is designed to not move you need neither motor nor gearbox. As soon as things start moving the gearing efficiency _does_ matter a _lot_. Why? because Colombic friction is not a viscous drag that is somehow proportional to velocity, it is a constant torque that must be overcome.

I suppose that while your statement that "gearing efficiency doesn't matter at zero rpm" is technically correct, you left out the "but it matters for _any_ non-zero speed"

No, I wanted to avoid that question entirely if I could. But if you have to study up on it this may help:

The point is, the *stall* torque developed at slow speeds isn't affected by friction. As a linear instead of rotational example, air cylinders develop force = area x pressure, without regard to friction.