# Quickly switching electromagnet polarity. Inductance issues?

1. Sep 24, 2014

### HeZ

Hello all,

There are a million threads and a million Google hits out there on the topic of switching the polarity of an electromagnet. Very simple stuff. But none seem to touch on the issue of inductance.

I am building a small linear actuator that will be a coil wrapped around a thin plastic tube with a cylindrical neodymium magnet in the center. So by changing the polarity of the coil I can move the magnet between two positions quickly.

I want to do this at about 100Hz and with a very fast response time, but as voltage is proportional to time derivative of current - when trying to switch the current quickly, I am expecting some very high - potentially damaging - voltage spikes.

I don't think I will just be able to hook the magnet up to a h-bridge and switch the current backwards and forwards will I? I am hoping that the field can collapse fast enough to switch it like this. I do not have any figures available to try and calculate the time constant or how much emf will be produced. But I will not need very high current. I was planning to use 6-12 volts and I only require to move a 1/4" diamteter x 1/2" length cylinder about 20mm with very little force (just quickly)

Am I thinking about this correct? Am I likely to need some extra circuitry to dissipate this emf?

Thanks,
Chris

2. Sep 24, 2014

### Baluncore

An H-bridge will clamp the coil voltage and have flyback diodes across the switches.
There should be no significant voltage spike while the current is reversing.
The rate of change of current will be determined by the voltage difference. V = L * di/dt, so di/dt = V / L.

The inductance effect will be low because you have no powdered iron or laminated core. Skin effect works for magnetic fields in cores as well as for currents in wires. The internal magnet will probably have a conductive metal coating such as nickel. That surface conductance will reflect the solenoid's magnetic field so the magnet will not be a core that increases the inductance.

Your solenoid has basically an air core so it will change current direction and switch field quite quickly. The solenoid field will react with the permanent magnet's field. You can expect performance similar to a loudspeaker, well beyond 100 Hz.

3. Sep 24, 2014

### HeZ

Exactly the information I was looking for, Thank you!

4. Sep 24, 2014

### skeptic2

You didn't say what the inductance of the coil is but for example if your inductance is 2.5 mH, you could put a 1000uF capacitor (non-polarized) capacitor in series with the coil and the result would be series resonant at 100 Hz. Larger inductances would require smaller capacitors. The result would be no voltage spikes at all.

5. Sep 24, 2014

### Baluncore

I assumed that meant the field required was close to a square wave. An LC sine wave oscillator would not achieve that quick transition.

I would suggest that the fastest transition would be achieved by using a high voltage capacitor momentarily connected across the coil, then charged and reversed for the next transition. The momentary voltage spike would change the current and field at the maximum rate without high power dissipation in the coil resistance between transitions.

6. Sep 24, 2014

### Staff: Mentor

The mechanical power to move the magnet could be an issue. 6mm diameter, 12mm length would suggest a magnet with a mass of about 1g. Moving it by 2cm in 5ms requires a velocity of 4m/s or an energy of 8mJ delivered in less than 5ms (~3W). This would just allow moving back an forth with 100 Hz, however, no "fast response time". The required power scales with $\frac{1}{t^3}$ - reducing the time from 5ms to 2ms increases the power from ~3W to ~50W. This could be a challenge both for the power delivery system and for the coil.

7. Sep 24, 2014

### HeZ

Skeptic:
I have not wound a coil or decided on any parameters yet. When I said I do not have any figures available to try and calculate the time constant, I was kinda implying that I do not know inductance or capacitance. But I am after more of a digital coil that can be activated via a uC.

Baluncore:
Yes I am after a square wave.
I guess it would help if I explain the application. I am building a robot to satisfy two purposes. 1) Develop my understanding of electromagnets. 2) Win a bet with my friend that I can build a robot to beat him at an android game - Piano Tiles. Essentially four columns of black rectangles scroll down the screen at increasing speed and you need to touch them before they are off the screen. I am trying to build a set of four linear actuators to touch the screen with capacities styluses (very light, made from carbon fiber and some sponge).

Now yes, there are probably a million better ways to do this. Everything from buying a linear actuator to meet my needs, using custom ROM to scrape pixels and inject screen presses, building something from the million hobby servos that I have laying around, the list goes on. I have a Mechatronic Engineering degree and love to build robots - but my knowledge of using magnets in this manner is lacking. So that is the point of this endeavor.

mfb:
Ahh looks like I calculated my power incorrectly, Thanks for pointing that out. They are 6.35x12.7 grade N42 - So they are just under 3g each. I had done similar maths, 3g at 4m/s I got 24mJ. However working out the power I simply used P = E / t = 12W. I am planning on using a 5A 12V supply, so I thought this would be sufficient - but I guess I worked it out wrong.
Why is it 1 / t^3?

The parameters I chose to give when posting were based on this math - 1A and 12V worked out nicely to 2cm and 100Hz - and were based on a worst case. My actual application is such that motion will be strictly vertical and I only require a fast response time when dropping. So the weight is not a huge issue, only inertia. When rising, as long as they lift about 5mm before they are dropped again then it will function correctly. So the power requirement is considerable less.

Thanks
Chris

8. Sep 25, 2014

### skeptic2

I understand your wanting to use a square wave to energize the coil and if you want to activate the coil at irregular intervals, it makes sense to use a microcontroller to activate it. However before we close this thread I'd like to offer a different perspective.

1. If your magnet is 12.8 mm long and you want to move it 20 mm, you may want to make the coil at least 46.4 mm long so that the magnet doesn't extend outside the coil.

2. Depending on the inductance and resistance in your circuit, the L/R time constant could be significant relative to the 5 ms travel time. In order to reduce the time constant to get a better square wave through the coil you would need to reduce L or increase R - both of which will reduce the magnetic flux of the coil. Increasing L or decreasing R may decrease the current through the coil by increasing the L/R time constant. Maximum current through the coil occurs at the end of the travel, at the time you want to reverse polarity.

3. Whereas a higher Q (L/R) tends to work against you as in #2, in a series resonant circuit it works in your favor. A Q of 10 will increase the current through the coil by a factor of 10 over just reversing the polarity across the coil. Furthermore, the peak current is reached in the middle of the 5 ms travel time instead of at the end, accelerating the magnet earlier in its travel time. Far from an LC sine wave not being as quick of a transition, the fact that it permits a much higher current to pass through the coil and that maximum acceleration occurs earlier, will likely result in a faster transition.

9. Sep 25, 2014

### jim hardy

Have you taken apart a disk drive?
The voice coil head positioning actuator seems suited to your distance and speed needs.

There's a lot of them junked, available nearly free.
That actuator contains a great 'fridge magnet that'll actually hold something.

10. Sep 26, 2014

### HeZ

skeptic:

Great, I will be sure to allow full motion without leaving coil. However 2cm just just a worst-case scenario, The actual travel distance is not too important, as long as it lifts a good 5mm.

I will be rolling the coil around a 3D printed tube, I do not know if i can design for inductance. However my plan was to put in a pot so I can easily tune the transience.

I am a little unsure what you mean. How can I use a resonant circuit to move digitally between two positions?

Jim Hardy:

Hi Jim, That is actually where I got the idea from, or at least the proof of concept. I have scrapped many old hard drives and have dozens of those "fridge magnets" (note to any other reader - Don't put them on your fridge without wrapping them in something soft).

I could go and try to find another couple hard drives, pull out the heads and try to arrange them in a manner that will suit my proposes. But to be honest, I am not trying to build something that will just work by hacking things that already work and that I (think I) have a decent understanding of the principals of operation of.
I am trying to apply my knowledge and in the process, develop my understanding. I am trying to build something from first principals that may or may not work work, that way I can figure out why it doesn't work, what I have done wrong and how I cam improve on my understanding.

My first attempt was to wrap a coil around a nail and see if I could repel the magnets that I have lying around - of course I didn't for a second expect this to work because the core attracts the neodymium with much more force than the coil repels with.
Whilst I am on the topic - what are some materials with decent magnetic permeability that aren't ferromagnetic? I ran out of wire to wrap around things so I have not been able to experiment to much - but I thought an aluminium rod might do the trick. It's permeability is not too high so it will not conduct the field too amazingly; but perhaps it will be strong enough to lift a couple of grams - especially since the neodymium will be attracted to the poles also?

Chris

11. Sep 26, 2014

### jim hardy

I truly applaud that spirit !

hmmm i think decent permeability is a property of ferromagnetism..
http://hyperphysics.phy-astr.gsu.edu/hbase/solids/ferro.html

Iron in a changing magnetic field develops eddy currents that retard the process of magnetization, it proceeds from outside toward center. That's why transformer cores are laminated. Of course the higher the resistance of your material the less the eddy curents. Carpenter Steel offers a line of stainless solenoid grade steels with fairly high resistivity for good AC performance , see fig 2 here:
http://www.cartech.com/techarticles.aspx?id=1562

and introduction to their magnetic stainless steels here:
http://www.cartech.com/techarticles.aspx?id=1562

I'd wager you could find some little parts made from it in automobile fuel injectors. Once again, a metal recycle yard item... a complete GM Throttle Body last month cost me just one dollar.

12. Sep 26, 2014

### Staff: Mentor

Power scales with 1/t^3 as the velocity scales with 1/t, the kinetic energy corresponding to it scales with 1/t^2 and the time you have to transfer this scales with t - and power is energy per time, so 1/t^3. Note that a constant power will lead to a roughly constant increase in kinetic energy over time, so distance will be less than the final velocity multiplied by the time. I added a safety factor of 2 to account for that.

If that game is meant for humans, then you won't need a 100 Hz rate, so it should be fine.