What type of Motor should I use? (Discrete Position + Vel/Acc Control)

Greetings,

Going back to the drawing board on an Engineering problem I've been having. We are trying to create scaled representations of Earthquakes. As such, we need to discretely control position and speed or position and acceleration at all times. This has proven to be an extremely difficult task. We have sought the following avenues, but all have been dead ends for us, either because our research is wrong, or we haven't pursued the right 'type' of research. We will be using either a Rack & Pinion or Scotch-yoke style system..

First, our specs:

Angular Precision: Minimum 1 Degree accuracy. Preferred Half a degree or higher.
Velocity: Discretely controllable between 0 and 300 RPM. Preferred up to 600 RPM.
Torque: Minimum 100 oz-in, may need as much as 180 oz-in
Acceleration: As High as possible
Component Budget: $160, prefer under $100

Stepper:
Great position and speed control, but can only stop or switch direction at full steps (Usually .9 or 1.8 degrees).

• Position Control not accurate enough
• Requires very high voltage and current (30V+ , 2A+)

Standard Servo:

• Can control position, but not speed (sweep does not control speed, only uses position incrementing that masks velocity/acceleration profile from the human eye)

Continuous Servo:

• Can control speed, but is not discrete. (No guarantee of speed accuracy)
• Cannot control position

DC+Encoder:

• Have yet to see any such system be in our specification range.
• Not sure Arduino Uno has processing power to handle controlling each device at 50Hz.

We're in this funky area where we're somewhere between hobby electronics and industrial electronics, trying to use the top end of the former to accomplish the latter.

Anyone with thoughts, advice, or product ideas is greatly appreciated.

Post a drawing of what you are building. Your wasting your time listing motor specs without the drawing of the proposed system.There are too many factors to boil it down to motor specs. You can do that after we see what it is you are proposing to build.
Hint: Did you remember to mention WHAT SCALE FACTOR ?
Have you figured out how to generate the earthquake ?
Have you ever seen an Industrial strength Voice coil, AKA VIBRATION TABLE (10' X10') ? (The coil is about a meter across)
Post the drawing.

This is a very early mockup of the design from 8 months ago.

This is the design as it stands now:

As I said initially, we are using a Scotch-Yoke system. This provides the advantage of having easily-generated harmonic motion, but at the disadvantage that you lose power-transmission from rotary to linear motion as the cam approaches a perpendicular (bind) state to the yoke.

We have considered a rack and pinion system as well to eliminate this problem with power loss at +/- 90 degree angles, but have not implemented it yet, as the actuator system is the more imminent problem.

Table02.JPG800×451 24.8 KB

raschemmel:
Post a drawing of what you are building. Your wasting your time listing motor specs without the drawing of the proposed system.There are too many factors to boil it down to motor specs. You can do that after we see what it is you are proposing to build.
Hint: Did you remember to mention WHAT SCALE FACTOR ?
Have you figured out how to generate the earthquake ?
Have you ever seen an Industrial strength Voice coil, AKA VIBRATION TABLE (10' X10') ? (The coil is about a meter across)
Post the drawing.

"Scale Factor:" Our Specifications regarding displacement, velocity, and torque are specifically designed regarding our system load. The scale is variable depending on what type of seismic event is being reproduced.

"Have you figured out how to generate the earthquake" - That's what we're trying to solve via actuation issues. We will use either a Scotch-Yoke or Rack & Pinion system for simplicity of design, affordability, and ease of manufacture.

"Vibration Table" - Unfortunately these are industrial test devices not suitable for the type of motion we are trying to generate. We are looking solely to reproduce motion lateral to the surface of the Earth, and hoping via Fourier transforms to mimic multiple-frequency waveforms seen in Earthquakes.

I can't see any dimensions

Could you use an offset weight driven by a motor in a piston like motion?
How about a slide driven back and forth by a pneumatic solenoid?

Second picture device is approximately 23" x 17" x 10". Motion surface is 18" x 18". Next iteration of the device will be shrunk to 12" x 12" motion surface to reduce materials cost and device load. Device is designed for 1" of linear displacement each direction, or two inches of total displacement.

Both of the drive ideas you mentioned, however, unfortunately don't address our pressing need to control both displacement and speed or displacement and acceleration to the discreteness that our project is seeking.

Unrelated note, while the device is capable of continuous rotation, it is not designed for that. It is meant to be able to change direction and speed rapidly.

Could you drive the table with a motor driving a screw (a ball screw for low friction and low backlash, perhaps). If you used a stepper motor the screw would "multiply" its precision. You could even have reduction gears to drive the screw. Or you could use a DC motor and a rotary encoder to drive the screw if a stepper is too slow.

Have you considered whether the sort of servo drives that are used on machine tools are within your price range? (They may be horrendously expensive, I have no idea).

...R

Sounds like very small linear drive. ( I don't think you could afford one)

Look at the voice coil actuator about fourth from the top.

http://www.tpa-us.com/linear-motors.html?gclid=CJv5u46MmL4CFQmSfgodwZUADQ

Peak Forces from 28N (6 lbs) to 131N (29.5 lbs)
Direct Drive with zero backlash and Zero cogging
Low moving mass with very fast response and high bandwidth
These are very high precision products and pricing starts at approximately $2,000.00

https://www.google.com/search?q=linear+drive+motor&rlz=1C1GPCK_enUS504US504&es_sm=93&tbm=isch&tbo=u&source=univ&sa=X&ei=SkBpU_ykM8jtoATb3YHQBw&ved=0CF4QsAQ&biw=960&bih=468&dpr=2

http://www.ebay.com/itm/High-speed-Linear-actuator-motor-Stroke-100mm-Force-120N-24-VDC-/221231297437

voice coil linear actuator.jpg762×600 38.7 KB

I still don't see any Force specs in your proposal (Nm)(dNm/dt)

raschemmel:
I still don't see any Force specs in your proposal (Nm)(dNm/dt)

Priest:
Torque: Minimum 100 oz-in, may need as much as 180 oz-in

@Robin2
We inspected a ball screw system early on, but concluded that it couldn't give us the linear velocity necessary without exceeding our budget of $150. We've seen Servo Drives, but are not really sure what they 'do' beyond ensuring command accuracy.

Torque: Minimum 100 oz-in, may need as much as 180 oz-in

100to 180 ozin = 0.706 -1.28 Nm

After doing an afternoon of research, I can't find any research to support what was told to me regarding Stepper Motors and Changing directions.

I was told that a stepper motor can only change directions at a full-step increment (commonly .9 or 1.8 degrees). This does not seem to hold true after reading Dr. Jones' work at Iowa.

The lowest amount of torque a stepper motor will provide is halfway between steps. This torque is defined as h/(2^.5), where h is the holding torque or two-phase torque.

What this seems to imply is that as long as friction and load do not exceed that number, I'm golden and won't get slippage. Moreover, this seems to say that I can hold a micro-stepped position (IE: 2.7 degrees on a 1.8 degree/step motor) as long as I do not exceed this torque.

Anyone have any knowledge to verify / refute?

Edit: Note that this is in an Ideal motor, and ignores Detent and Quantization effects.

Priest:
What this seems to imply is that as long as friction and load do not exceed that number, I'm golden and won't get slippage.

That is my understanding. The key things seems to be a motor with ample holding torque.

Moreover, this seems to say that I can hold a micro-stepped position (IE: 2.7 degrees on a 1.8 degree/step motor) as long as I do not exceed this torque.

From bits and pieces I have read I'm not sure how accurately the motor will move through the microsteps if it is moving a load. My sense is that it will "recover" its accuracy at each full step position (when the torque is greatest). Again, this may only be an issue with motors that don't have a sufficient margin of torque over load.

...R

A DC or BLDC motor configured as a servo-motor is exactly what you need.
Unfortunately commercial servo-motors are expensive.

The potential performance of a good servo-motor is impressive, a few minutes
of arc position error at full speed and acceleration, ability to go from full
forwards speed to full reverse in a matter of milliseconds under load. All
high-end positioning systems use servo motors because they perform much better
than steppers (for speed, power, accuracy and noise).

However on a budget there is a route, to construct your own servo-loop
around a motor and encoder. This is basically what a hobby servo does,
but hobby servos are low power and limited to 180 degree rotation. A home-brew
won't be able to match commercial units for ultimate performance but once
you have closed-loop control you gain a lot of ability. Once you have that
position control automatically controls speed and acceleration.

You need a motor with adequate torque (remember most motors can go
to 3 or 4 times their continuous torque rating for brief periods, so base
the desired torque more on the rms power of your waveforms than the peaks).

Add an encoder with adequate resolution - this is going to be the hardest part to
source since you want fairly high resolution, something like 1000 ppr is needed.

Perhaps some reduction gearing, 8:1 or so seems reasonable - thus the motor
should be a budget gear-motor with an encoder or at least a second shaft for
mounting an encoder.

Then you need control electronics - this implements a PID loop where the error
is the difference between current position and desired position, and the output
is PWM drive to an H-bridge adequate for the motor - some of the beefier MOSFET
H-bridges at pololu are plausible for this. A battery supply is probably mandatory
as your set up is inherently 4-quadrant.

You also need a way of feeding the desired position waveform to the controller
(which I'm assuming is some flavour of Arduino board) - this is potentially
high bandwidth, comparable to audio? SD-drive?

But you first need to work out the torque and power requirements. How much
mass are you moving and at what accelerations?

If you go the stepper motor route you will need something fairly chunky run
from a chopper driver, but you still need to source your waveform data and
convert to stepper pulses.