Very slow rotation speeds

[Erdbau]

We have an old (35+ yr) system that spins a cone gear at very slow speeds: input rotation needs to be as slow as about 2.5e-4 rpm, i.e. 4000 minutes per rev. The old system we had involved an AC synchronous motor running at 15 rpm, a 4:1 reducer, a multigear reducer (ranging from 2:1 to 1000:1, and a 16:1 reducer to the cone gear. The input torque on the cone gear needs to be about 100 in-lb. The manufacturer hasn't made the multigear reducer in over 20 years (pull and turn shifter). The multigear is malfunctioning (I think it is slipping, but we don't have the time or expertise to take apart and diagnose).

I have two questions: 1. Do these multigear reducers still exist, and if so, can anyone suggest manufacturers?

2. An alternate solution (need slow, constant rotation, and need to be able to repeat a speed (similar to the way I used to be able to when I set the dial to a particular gear ratio)) that I was thinking of was using a stepping motor, with the appropriate reducers to get to an acceptable torque. Would using microstepping to rotate the motor very slowly be a reasonable use of a stepping motor?

Any help would be appreciated. I am completely out of my element on this one.

 

[davidbeach]

A vector drive with a sensor ought to be able to give you what ever speed you want, if you can keep the motor cool.

 

[MikeHalloran]

Steppers don't do 'slow'.
They change mechanical state within a few ms of the electrical state changing, no matter how infrequently a step is executed. The resulting micro- hammer- blows can do an amazing amount of damage to a geartrain.



Mike Halloran
Pembroke Pines, FL, USA

 

[jpts]

If the speed of the motor is really that low, I don't think that vector-drive (or any converter-drive) will perform very well. According to my experience, close to 0 Hz, frequency converter is in trouble when constant speed is needed. It is very difficult for converter to determine the speed of the motor close to 0 Hz, and the control cannot operate very well. In practice this results that the speed is far from being constant, and the motor rotates more in steps. Even though you would have a speed-sensor with lets say 4096 pulses per revolution, this means 1 minute interval between pulses @2.5e-4 rpm!. Clearly there is no way the control can determine the speed from this. With open-loop control I think the situation is even worse. You would need extremely high sampling frequency in current measurement to see the frequency (i.e. speed).

There is also another problem related to frequency converters at low frequencies. That is the thermal behaviour of power transistors. Now you have almost DC current, which means that on- and off- periods for power transistors become very long, and this means that the temperatures of transistors go up and down (at normal frequencies they are quite constant), which can break the transistors mechanically quite fast (thermal expansion). This is quite well-known problem related to converters at very low frequencies.

 

[Erdbau]

So would a better solution be to run the stepper at a large number of microsteps per rev so the motion is "smoother", at a much higher speed, and then use a large ratio speed reducer? I wouldn't be able to feed the stepper straight to the cone gear anyway, the torque would be way too high.

Is the speed on a stepper that hard to control? My understanding was that as long as the pulses from the driver are constant, the motion will be predictable. I can understand how things may fall apart once the speed approaches zero, but what about at somewhat higher speeds (lets say around 4 to 10 rpm?

 

[MikeHalloran]

Stepping motor motion is >step<wise, hence the name. Microstepping is smoother, or less rough, but still stepwise. Speed control in either case is determined by the scheduling of state change in the driver, and is insensitive to, e.g., load. Constant, repeatable rotational speed is therefore trivial to achieve, on a macro scale.

On a micro scale, the step motor is stationary most of the time, and rings at every step. It's not clear from your specification so far if that's a problem or not.



Mike Halloran
Pembroke Pines, FL, USA

 

[LionelHutz]

Unless I missed something, this would require a motor that would run from say 15rpm to 7500rpm to replace the gear changing box. I'm not sure this is practical, both on the motor side and especially the gearbox side.

The gearbox you have sounds similar to what is on a typical milling machine or lathe.

 

[Erdbau]

Mike, the cone gear has a 40:1 reduction, and lets say I use a 1.8deg stepper with 10X microsteping. That's 80,000 pulses per revolution at the business end, which is acceptable for the purpose (rotating a ring of soil with a 5 inch OD). With the additional reducers I'd need anyway, that would mean even more pulses per rotation. I understand the discrete nature of the motion, but it seems like I'm dividing the motion up into enough steps that it seems like it would mimic continuous motion. But I guess it's more of a question of whether I can run the stepper slowly enough (4-10 rpm)

Lionel, you're right the gearbox reduced from 2:1 down to 1000:1, a 500-fold range. The biggest gearmotor range I've been able to find is about 150-fold. Has anyone seen different?

 

[Skogsgurra]

Microstepping will work quite well. One has to forget about the 'stepper' part in the name and understand that the steps are replaced by two continuous sinewaves(usually 256 levels) and that results in a very smooth rotation. I have used the technique to rotate mirrors for triangulation.

A similar solution with a PM servo will probably work even better. Lots of suppliers in Germany (handle "Bergbau" says you may be from that part of the World).

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...