You might consider 4 coil continuous control. A no load home position would be 2 outside coils fed to match armature poles. One can accelerate with 4 coils......but the neat thing is 2 sets of two poles. Now you have two directional variables for position and acc/braking. There is always and push and pull. The armature is always under 2 accelerations. Not the most efficient....but very find control.

You do want an analog movement, right? You want to kiss too.

 

For pool I think a linear motor not the best idea here.

 

If the axial length of a coil is greater than the thickness of a disc magnet then I think the closeness of the two magnetic poles of the disc will mean that the coil's magnetic field will be pushing one pole but pulling the other pole, effectively (almost) cancelling the force. This would be similar to using a non-magnetised steel disc instead of a permanent magnet.

Hi Alec_t. Interesting. If I could get away with using non-magnetised steel discs, that would help later. There are many magnetically sensitive sensors in the rest of the machine that I've just been planning to somehow silence when the tube of incredibly strong magnets is on the move.

Your theory and this idea will be easy to test once I learn how to control the current dump through the coils. I'm circling in on a "Power N MOSFET, with low Vgs voltage as a coil driver" to order. I'll keep you posted as parts arrive and testing resumes.

 

For pool I think a linear motor not the best idea here.

We've tried a pneumatic solution, an elastic solution, a spring solution and are now trying an electromagnetic solution. Open to suggestions.

 

For the sake of documentation. Here is a fantastic article on Arduinos controlling mosfets. I just ordered 20 IRL540 Mosfets.
https://arduinodiy.wordpress.com/2012/05/02/using-mosfets-with-ttl-levels/

 

Hi, next question, how are the poles of the PM(permanent magnets) oriented? By that I mean the N and S poles, PMs are made in many different pole orientations. Some are radial, 1/2 of the diameter N one S. Some are linear one of the flat sides N the other S.

Have you taken a stepper motor apart to see how it is made? If so, you would see that there are both PM "ribs" or poles and also electromagnet(EM) poles involved. Now let me say this up front, I've never made a linear motor, but have seen them, but never have taken one apart. But that said I don't think just having EM coils wrapped around a tube will get you the movement you are after. You are just creating a magnetic field around the tube. You need, like in a stepper or even a BLDC individual or 'salient' poles, to concentrate the EM force. This will allow the opposite EM pole to pull the PM pole forward. And by reversing the EM pole of the rear side of the PM to be the same polarity to push the PM.

There is also a geometrical problem with the current arrangement you have and also will come up if you make a new outer coil, using pole pieces. That is the width of the PM pole to the distance between EM poles. There are some thesis papers out there online that can explain it to you better than I can if you Google using the word thesis in the question.

A couple of more thoughts on what you're trying to do.

One that may help, you can't just use a micro to decide when to switch your coils on and off. You need to know where the PMs are in regards to the EMs. Doing it works in a regular stepper because it is basically in a circle, it is always self contained movement. The linear is always trying to get away from is self so it need some form of sensing to know where it is.

Second is that in the end I don't think this is going to do what you think/want it to. I'm assuming this is for a school challenge? Think about how a pool que is moved during a shot. It starts out from a stopped position and accelerates during the stroke, not in one 'sudden' move but in a controlled fashion. This I think will be an either full on or full off stroke, like from a standard solenoid. May be wrong here but don't think so.

Hi Shortbus,

Thank you again for your thoughtful consideration.
The magnets I selected are "linear", as you describe, one side N, the other S.

I have some experience with stepper motors, yes. I am also encouraged by the violent movement I observed when shorting the battery directly to the coils. The magnet moved to be aligned with the center of the activated coil. The next coil, or step, is 1/4" away.

I definitely just guessed on the geometry of this first test system, using physical design constraints of the rest of the project as my only guide posts. I suppose I'm doing the Edison, "1000 failed bulbs" design method rather than the Tesla, "plan forever, do it once, right" method. I am comfortable reading thesis papers and I have time for my new chips to come over on the boat from China. So I'll add some deeper understanding to the mix.

"You can't just use a micro to decide when to switch your coils on and off." Well, I definitely can, assuming these new chips come in and work as expected. However, I recognize what you mean... the moving part will likely be erratic and a fixed pulse arrangement will likely lose the PMs or behave differently one stroke to the next. Good point. Oh man... how to add feedback to the system...? I wonder if I could make some sort of brushed linear motor system... I'm visualizing sparks flying and a trail of flames down the side of the system reminiscent of a time traveling Delorean... I'll see what I can find in the way of linear encoders and methods of reading one into a controller.

Regarding full on or full off: I'm hoping to address that (and looking forward to experimenting with) using analogWrite commands from the arduino to vary the EM field strength from the coils, hopefully controlling the strength of the shot. Also. as the full shot stroke will be only 60 "steps" it is reasonable to change time duration of each step to create an acceleration curve for the pool cue. Again, I'm with you that I'll likely end up wanting position feedback.

Thank you again.

 

You might consider 4 coil continuous control. A no load home position would be 2 outside coils fed to match armature poles. One can accelerate with 4 coils......but the neat thing is 2 sets of two poles. Now you have two directional variables for position and acc/braking. There is always and push and pull. The armature is always under 2 accelerations. Not the most efficient....but very find control.

You do want an analog movement, right? You want to kiss too.

Yes, I want analog movement. Yes, I want to do very soft, kiss shots (highest priority) as well as break shots.

I'm not concerned with efficiency and do want very fine control... so, I'm really excited by this suggestion and am trying to understand what you are saying. But, my head is spinning.
Any clarification, or a sketch, or link to reading would be much appreciated.

In the meantime, here are some clarifying questions:
Your idea requires each coil to be able to have polarity control, correct? So an H-Bridge on each coil, yes?

For the sake of communicating, let's name the coils A, B, C and D.
"No load home position" : do you mean that A and D are on in opposite polarity to hold the magnet centered between coils B & C?

My idea is to turn on A to attract the first magnet, as the magnet approaches alignment with the A coil center, A turns off, B turns on. As the coil passes the alignment point on A, A switches polarity and repels the magnet towards B. As the magnet approaches alignment with B, B turns off and C turns on, and so on...

Your idea is similar but uses "two sets of two poles"? How does that look in terms of A, B, C and D?

Thank you for any further clarification. This idea feels like it could be one of the much simpler solutions that seem to always present themselves when I'm in over my head while designing.

 

What are the constraints in the challenge? Size limitation?

Being from more of a mechanical back round I'd do it differently. I'd use a car door lock motor unit and have that move a toggle linkage to do the actual stroke of the cue. By controlling the voltage to the motor you could also control the speed and strength of each shot. And each shot would be the same/consistent in distance of movement.

 

What are the constraints in the challenge? Size limitation?

Being from more of a mechanical back round I'd do it differently. I'd use a car door lock motor unit and have that move a toggle linkage to do the actual stroke of the cue. By controlling the voltage to the motor you could also control the speed and strength of each shot. And each shot would be the same/consistent in distance of movement.

Shortbus, thanks for your continued interest.

Interesting. I just ran a project where students built pinball machines using car door lock actuators. The stroke on those was an inch, maybe 1.25".

My hesitation in using a short stroke device like a car door actuators and most pneumatic or solenoid systems is twofold: first, I'd like the machine to have the appearance of using a cue stick so that it feels to the human opponent that the robot is competing using the same tools. (I recognize this lacks logic) Second, I would like to have the ability to follow through on shots, if desired. Short stroke devices wouldn't have that ability.

Project constraints and limitations?
Safety police and common sense say that a robotic shooting device must be entirely gone before a human can approach the table. Long ago, the decision was made to have the shooting device UNDER the table when parked (not shooting) in order to meet this criteria. The mechanism to get the shooting system out, around and over the table takes up much of the space under the table, so the shooting mechanism is better, the smaller it is. The project is modeled on a 1:6th scale here, including some previous shooter system test videos: https://www.instructables.com/id/Pool-Playing-Robot/

 

I have never done what you're doing. So...I would have to start from scratch. The first thing I would do is wind one coil. Put a resistor in series with it to limit and set current.

When you reach a armature response(with different size resistors).....does the armature travel to the end.....or lock at segments? Also feel the resistance to manual movement....it will indicate the power needed to static lock. If it locks at segments...increase current and see if armature travels to end.

You need a feel for the individual armature magnet's density(and reaction).....and the overall armature density. An armature with long distance poles is ok......but to keep it simple....only one total field for analog control. We don't want notch movement.

Once you know how the armature reacts to one coil......you might want to adjust armature spacing or etc. Adjustment of coil might be needed too.

We need to verify the static yin yang of armature and coil.....before doing anything. You can tell by pulling on armature and releasing.......the action needs to be smooth, constant and to the same orientation, and position before proceeding. After you verify the action needed.....then you can take reference measurements. One coil has a short N-S pole distance. You can lengthen it with another coil. With a separate coil...you can vary the strength of that distance....even the polarity.

But before you go to independent coil control.....counter wind(or flip over) the second coil. Connect this coil in series with first coil. Apply slow sine, close to static measured values.
You want the armature to slowly oscillate.

This will be the fun part. Counter acting the armature inertia. As you increase the test sine rate....the armature will be ejected. Now you get an idea of the speed and magnitude of magnetic field to contain armature. An armature position sensor and two large guard coils could inprison it.

The path you take is going to depend on what you find along the path.

How are you balancing the armature(stick)? What will it ride on?

 

How are you balancing the armature(stick)? What will it ride on?

The coil holder is currently designed to be slightly larger than the carbon fiber tube. There are small tabs on the inside ends of the coil that actually touch the outside of the tube. Here is a closer look:

First of all, thank you for further details and steps to take. I'm still processing everything else you said. In the short term, it sounds like I need to get and to read about a signal generator in order to "apply slow sine."
I'll dig around to see what I can find.

While I continue to process, do you have any good recs for tools to enable me to "see" the current going through the coils? Current could be up to 24A...

Thank you again.

 

I've been trying to catch a glance of current for a long time. Haven't caught it yet.

I think the closest we've come to seeing current .....is it's ion trail.

The project that you are undertaking will require a lot of study and skills.

How are you with DSP? This will require fast, sophisticated digital control.

There are many algorithms for rotational power control. I'm not familiar with linear control such as this.

There are many control and driver control/designers on this site. Real experts. I'm just thinking out-loud.

Whether you succeed or not......just investigating it will teach you.

 

I've been trying to catch a glance of current for a long time. Haven't caught it yet.

I think the closest we've come to seeing current .....is it's ion trail.

The project that you are undertaking will require a lot of study and skills.

How are you with DSP? This will require fast, sophisticated digital control.

There are many algorithms for rotational power control. I'm not familiar with linear control such as this.

There are many control and driver control/designers on this site. Real experts. I'm just thinking out-loud.

Whether you succeed or not......just investigating it will teach you.

Oh man, "See" was a poor word choice on my end. Rather than see ion trails, I'd like to be able to understand what is actually happening in the coils in terms of current flow as I'm switching the IRL540 mosfets. I feel like an oscilloscope could do this, but I haven't used one since college. Is that the correct tool for that purpose? Up to 24A? I'm happy to read manuals and whatnot, but a point towards the correct tool would be very helpful.

How am I with digital signal processing? Add it to the reading list. I've not processed any digital signals. I'm assuming that's what I'll be doing when I get proficient with the tool that answers the questions above.

No linear motion control algorithms? Writing algorithms is something I've been trying to learn. Pool ball location analysis and shot selection is how I starting in the algorithm writing realm. Perhaps I reroute my efforts towards linear motion control instead.

Yes, the teaching received while doing this project has been fascinating. "I learned a lot" feels like the consolation for every failure. That, and the fact that the world doesn't actually need a pool playing robot.

Anyway, thank you again for your thoughts and for your help.

 

Hi BobaMosfet, (nice handle)

I believe a solenoid to be a single coil and a single magnet. I'm sequentially activating four coils to move a series of spaced magnets... so I feel valid in sticking with the title "linear stepper motor." I'm not ringing doorbells with this, but hoping to have controlled motion. Here is a video of the inspiration:

Speaking of controlling motion with this thing, NO, I definitely have not calculated the time necessary to generate the magnetic field. My basic understanding of electricity told me that the magnetic field instantly appeared once the current dumped. Based on your comment, this is not the case. Uh oh.

Here are my back of the envelope calcs as of now:
The mass I am hoping to move is 8x that of a cue ball, or 1.28KG.
Pros can break at speeds of 25MPH (~11 m/s). Let's aim for that.
Initial velocity = 0 m/s
Ideal striking velocity = 11 m/s + 20% for losses during impact = ~13m/s
Distance (full back-swing to strike point): 0.4m

a= (v2 - v1)vav /d
a = (13-0)6.5/.4
a = 5m/s^2

F = ma
F = 1.28Kg*5m/2^2 = 6.4N
(assuming constant acceleration)

Soooo, how do I generate 6.4N of force between the coils and the permanent magnets...? I'm off to do reading but happy for any guideposts you may have.

How quickly a field develops is a question of how much voltage is applied. Voltage is not a quantity, so much as a force of attraction. The greater the attraction, the faster current moves. You have to remember that an electronic 'field' is composed of electrons-- they push back against one another- they don't like each other. This is what stops a magnetic field from building. Electrons push back against the incoming electrons with equal force because the voltage cannot make any more of them clump together in the field. Once that happens, excess electrons bypass the field and move through the conductor coil, making a short circuit.

Yes, electricity is fast, very fast, but it isn't instantaneous. Much of actual understanding in electronics, is understanding this small but critical aspect of how electrons behave.

So, you have 4 solenoids. Let's just stay with 4 solenoids for the moment, to simplify your problem. Together, they are the 'linear actuator', but, you have to control them individually. In order to move your rod, you must control each solenoid differently in terms of both charge, and polarity. Simply charging them fully isn't enough. If you want controlled movement, you have to charge some slightly, others at max, and still others the same, but in opposite polarity so that you control the sliding of your core rod.

I recommend, for proof of concept, you work with just one solenoid, and one core to start. Once you can move it one way and then the other, then add another and so on.

 

My hesitation in using a short stroke device like a car door actuators and most pneumatic or solenoid systems is twofold: first, I'd like the machine to have the appearance of using a cue stick so that it feels to the human opponent that the robot is competing using the same tools. (I recognize this lacks logic) Second, I would like to have the ability to follow through on shots, if desired. Short stroke devices wouldn't have that ability.

See, I thought you needed to use a solenoid, coils or linear type motor. Why not a stepper motor with a pinion gear mounted on the shaft. This pinion would mate with a rack gear. Then by the number of steps being taken from ball contact(the zero point, like in a CNC mill) the number of steps backward could be programed and the number of steps forward of zero could also be programed. The rack would be tipped on both ends by a cue segment, like used in a two piece cue, to mimic a "human" using one with the motor and rack area mimicking the 'bridge' hand of a human. By using a zero point, step limits and speed of movement for stroke, all of the normal computations that a human does when making a shot are also able to be programed into the robots movements.

 

Not sure if this would add something helpful to the OP but, for some time, I was an avid reader of catalogs listing hundreds of voice coil motors. I got the concept that the definition (?) of their movement is as fine as your electronic control could be. The other concept: going higher in power, price goes high quickly.

Every time the actuators subject comes, I recall Firgelli.

Isn't the OP saying that bypassing all the control stuff, he got a clear response?

 

Isn't the OP saying that bypassing all the control stuff, he got a clear response?

I think he did, but not the stepped response he's after, just got an all or nothing movement, like a solenoid.

 

How quickly a field develops is a question of how much voltage is applied. Voltage is not a quantity, so much as a force of attraction. The greater the attraction, the faster current moves. You have to remember that an electronic 'field' is composed of electrons-- they push back against one another- they don't like each other. This is what stops a magnetic field from building. Electrons push back against the incoming electrons with equal force because the voltage cannot make any more of them clump together in the field. Once that happens, excess electrons bypass the field and move through the conductor coil, making a short circuit.

Yes, electricity is fast, very fast, but it isn't instantaneous. Much of actual understanding in electronics, is understanding this small but critical aspect of how electrons behave.

So, you have 4 solenoids. Let's just stay with 4 solenoids for the moment, to simplify your problem. Together, they are the 'linear actuator', but, you have to control them individually. In order to move your rod, you must control each solenoid differently in terms of both charge, and polarity. Simply charging them fully isn't enough. If you want controlled movement, you have to charge some slightly, others at max, and still others the same, but in opposite polarity so that you control the sliding of your core rod.

I recommend, for proof of concept, you work with just one solenoid, and one core to start. Once you can move it one way and then the other, then add another and so on.

Hi again BobaMosfet,

Thanks for breaking this down a bit.
Fascinating. So when the magnetic field is maxed out (depending on the voltage) the "excess electrons bypass the field and move through the conductor coil, making a short circuit." So it isn't the flow of electrons that creates the magnetic field, but the build up of electrons trying to fight through the coil... when the coil is full, then current then actually flows. So, I would see a delay between the presence of the magnetic field and the detection of current flow. Fascinating.

Yes, your description of how to test is precisely what I hope to do when my new IRL540 chips arrive. All this talk has me looking for suppliers who can get them faster...

Thanks again.

 

So it isn't the flow of electrons that creates the magnetic field

Yes it is... It's just that when the coil gets "filled" (as far as dc is concerned) current starts to flow freely, unrestrained by the effects of inductance, and ohm's law takes over and begins to create heat real quickly... but the magnetic field generated stays steady, until the current is shut off.

That's why many industrial magnets have some sort of cooling system, which can range from forced air flow, to liquid cooling through a heat exchanger .

 

When you strike a pool ball straight on....the ball travels in the hit direction.

How would you like NEW pool balls, that when you hit them straight on.....the ball will fly 90 degrees to the right or to the left of the hit direction?

You will have to learn to play with these new pool balls. A new mechanics.