Linear motion setup

[Racer26]

I'm searching for a linear motion setup, and hoped the CD community could help me find what I'm looking for.

I'm looking for something that meets the following criteria:

Low Cost (< $200)
High max Speed (> 1 m/s)
Long-ish max stroke (10-14")
Ability to accurately (read: within 0.1") position it to anywhere within its travel
Variable speed

I've found Festo's DNCE-LAS actuators, but haven't been able to get a price for them.

[Racer26]

Got a price: nearly $4000, so DNCE-LAS is out, but something like it would be ideal.

[Tristan Lall]

Are you hoping for a self-contained solution (i.e. all those functions in a single unit), or would components be fine?

Also, what's the orientation and loading of the linear actuator? Do acceleration (and deceleration) matter? How variable does the speed need to be (fine or coarse control; large range or small range)? Any environmental issues (oil, contaminants, etc.)? Do you need absolute position feedback or relative, and do you need it in human-readable form or electronic? Is this for an FRC application?

I've used leadscrew-driven slide stages from Velmex before. They were quite nice, because they could support heavy loads travelling horizontally, but I'm not sure how fast they could reliably be driven. I believe they had ones with motors attached; we only used the hand-cranked ones. Some had graduations and vernier dials; maybe they have versions with encoders.

[DonRotolo]

Like Tristan mentioned, load is important. Moving 30 kg at 1m/s is different from moving 42 grams.

In general, for masses up to about a kilogram or so, a multi-start lead screw and a stepper motor & controller should be fine and well within your price and accuracy range.

Recent KoPs have come with such a lead screw.

[Racer26]

I don't mind creating the drive logic on my own (ie. a stepper motor with driver chip that takes in enable, direction, and step inputs)

I don't really mind assembling components, so long as they fit together with a minimum of modifications/custom work.

Orientation would be roughly horizontal in typical usage, though if vertical orientation is possible, all the better. Load is relatively low, 1/8hp is plenty to meet the requirements of the application, and 1/16hp is acceptable. Acceleration is important, in so much as the linear-speed goal is to have a reciprocating motion, between 200-400 cycles per minute.

Environmental issues should be a non-issue, and position feedback isnt really necessary, if the motor is a stepper. I merely need to know how far its going to go when I give it a particular signal. Its a non-FRC application.

[msimon785]

You may consider using a basic 12v linear actuator. They will never achieve a speed of >1m/s, but then, thats fast for pneumatics. Motorized actuators essentially use a leadscrew powered by a motor which suits the load and speed needs.
Any of these actuators should suit your needs. I filtered the search to show high-speed actuators.
Also, they also present the cost benefit. These will probably never hit $200.
To control their position, you will need a sensor kit. I believe http://automationdirect.com sells them, don't remember the price. As to speed control, basic PWM will do the trick.

[Racer26]

It would be moving a < 1kg mass, against a light resistance.

[DonRotolo]

Originally Posted by 1075guy

a reciprocating motion, between 200-400 cycles per minute.

Yikes! That's an awful lot of fast motion, assuming your average stroke of 12". A lead screw wouldn't last long (months, not years). Better use loctite on those fasteners, or it'll shake itself apart.

Oh, and doing the math, that's about 3 meters per second at 12" and 300 cycles per minute, assuming no time for acceleration and deceleration.

To put it into perspective, that's as fast as a commercial paint can shaker, faster than the average person would shake a spray paint can. And to move <1 kg at that rate is pretty demanding, mechanically.

I think you need to carefully re-think whatever you're trying to accomplish, because your design specs are somewhat harsh.

[Tristan Lall]

That's really fast...I don't think those Velmex slides are right for you. They're good for moving heavy stuff slowly and precisely.

It looks like you need to move light stuff quickly and precisely (if 47 W is the low end of acceptable, you won't be moving much mass at that speed). I can't think of any off-the-shelf parts that would work nicely for you.

Are you building a pick-and-place machine of some sort? The ones used to manufacture circuit boards can achieve approximately the sorts of speeds of operation and travel you're looking at—with relatively small loads—but they're not even remotely in your budget. (Maybe you can find pieces of one of those for sale at an industrial auction or something?)

For something that can't quite manage that sort of speed, but is at least reasonably close, check out an inkjet or dot matrix printer head. They have very accurate position control built right in, if you can figure out the control scheme.

[Al Skierkiewicz]

Linear actuators of the type you describe have been used for years in tape archive robotics. Most use a stepper motor driving a taut wire rope (aircraft cable) over pulleys in a continuous loop. One of the largest I worked on was an Ampex ACR225. http://www.lionlamb.us/quad/acr225.html
This machine could hold up to 256 D2 video cassettes and pick and inject a tape in one or two seconds. (Depends on where the arm is located, what cassette it is commanded to pick and what transport it is loading.) The robotics assembly required interlocks as the mechanism moved so quickly it would easily take your hand off if it were in the way. The linked in the article above shows the front of the machine. The black area is the tape cassette storage. The robotics needed a calibrate routine where the picking arm would need to sense the edges of each of the cassette bins. It did this by using a touch sensor and moving until the sensor touched the sides and bottom edge of each of the 256 bins. This routine took quite a while but it was fascinating to watch. The pick mechanism was an X-Y robot with grabbers that would position at a bin opening, grab the cassette, pull it out of the bin and then place it in one of four tape transports. Each cassette had a bar code label that the machine would use to identify the location of each cassette. The robotics rarely needed calibration but they did require regular maintenance. The vertical and horizontal devices moved on polished shafts with high speed linear bearings. Keep it cleaned and lubed and it didn't give us any problems. Same type of X-Y system is used in robotic inventory systems.

[Alan Anderson]

Do you just need quick, repeatable motion with consistent endpoints, or does it have to seek more than one location at each end of the travel? Tell us more about what the application is and we might be able to suggest an appropriate design space.

[Racer26]

I apologize, but I am being deliberately secretive about the application.

Tristan is correct, moving small loads, quickly, and precisely, though precision is not super-important. As I said, 0.1" is good, and really, even 0.5" would be acceptable.

The goal is to be able to stop and change direction at any arbitrary point in the travel. Max speed is something I'm willing to sacrifice if I need to. For example, 200-400 cycles/min @ 4" stroke, but only 65-130 cycles/min @ 12"

I think a stepper motor driving a multistart leadscrew with like 2 effective TPI would be the sort of solution I'm looking at.

[Ether]

Originally Posted by 1075guy

The goal is to be able to stop and change direction at any arbitrary point in the travel.

Will your controller be told ahead of time what the arbitrary point is, so that it can plan for that and start to slow down as it approaches? Does the 0.1" precision (accuracy?) apply to this arbitrary point?

Or will the controller need to respond to a "change now!" command given at an arbitrary point in time? In which case, how much overshoot is acceptable?

You can provide a lot more detail without revealing the specific application. The more detail you provide, the more likely you will get useful help.

[Racer26]

The controller will know its operation profile ahead of time. It will have a command list along the lines of

Go 8.5" forwards at a speed of 1"/sec
Go 4.5" backwards at a speed of 10"/sec
Go 1" forwards at a speed of 100"/sec
Go 5" backwards at a speed of 50"/sec

Repeat ad infinitum.

The profiles its requested to move could be completely arbitrary, within its mechanical limits of stroke and speed, but they will be known before it starts, so yes, it will be able to plan for, and begin to decelerate as it approaches the arbitrary points.

Conversely, the profile could be as simple as:

Go 12" forwards at 1000"/sec
Go 12" backwards at 1000"/sec

Repeat.

(Obviously, 1000"/sec is way faster than it would be able to go, its an example of the simplest sort of command string. It will always return to the starting point before repeating the cycle.)

I have considered a pneumatic setup, with the sort of fancy valves we're not allowed to use in FRC, and it would probably work and meet the needs, but pneumatics are much noisier than electric motors (quiet operation is preferred for this application), and speed control would be difficult.

[PAR_WIG1350]

I'm thinking something like a tapered cam on a stepper attached to a lead screw attached to a second stepper might work for adjusting the stroke in one direction, while allowing for some very high speeds. Varying the stroke in the other direction would require more complex motion and would slow it down a bit, but it would still be fast.

EDIT, It would probably be faster to use a standard cam and drop the lead screw and second stepper. The tapered cam would be better only if you had a set return point and only one end of the stroke varied.