MaSu on Machines, Meteors & Mozzies

has anybody ever tried setting up a magnetic levitation system based on a microcontroller and if so what were the end results?

I'm sorry I havn't ... but it sure as hell sounds fun and V interesting. I'd have thought that by pulsing the electromagnet you could stop it running away with itself, basically pwm it?

Obviously try it with a paper cube first not the precious model.

Del

G'day Dell,

· I'm sorry I havn't ... but it sure as hell sounds fun and V interesting. I'd have thought that by pulsing the electromagnet you could stop it running away with itself, basically pwm it?

I came to pretty much the same conclusion on the use of PWM especially since I plan to utilize a microcontroller. The only thing I'm a little worried about is the RFI that you can get by driving thumping great coils with what is for all intense and purposes a square wave.

However, part of the model is constructed from cardboard that is coated with aluminium so if I do have problems with RFI I can build a small enclosure out of it and earth the metallic finish to form a shield.

I finished building the ultrasonic distance measuring unit last night and surprise, surprise it worked perfectly measuring the distance to the model on the first attempt. Currently it utilizes a LED that varies in intensity as the position of the model changes. Hopefully all I will need to do is pick up the voltage it uses to drive the LED and feed it into the microcontroller's analogue to digital converter.

Regards, masu.

One thing you may want to look into is using a Digital to Analog converter to drive a voltage to current circuit. You could then use this to drive the electromagnet portion of your levitation system. The DA converter is an easily accessed circuit and may even be built into the microcontroller you are using. The Voltage to Current circuit is also very simple and can be built with an Op Amp and possibly a transistor to drive the higher currents you may need in your coil.

The major advantage of this approach as opposed to using PWM is that you would have no induction effects.

It sounds like a great project. I will be very interested in seeing how it progresses. I would be especially interested in hearing you insights into implementing the levitation control in software.

Good luck. I wish you the best on this and hope you can keep us updated as you go.

-Doug

G'day Doug & Dances with Trees

• One thing you may want to look into is using a Digital to Analog converter to drive a voltage to current circuit. You could then use this to drive the electromagnet portion of your levitation system. The DA converter is an easily accessed circuit and may even be built into the microcontroller you are using. The Voltage to Current circuit is also very simple and can be built with an Op Amp and possibly a transistor to drive the higher currents you may need in your coil.</quote>

As you suspected the microcontroller has built in analogue to digital and digital to analogue converters built in.

There are two ways I see that I can go about this with each having its advantages and disadvantages:

1. The first it direct current control. This lends itself to using bipolar transistors in the final driving/power amplifier stage but it prone to building up a considerable amount of heat in both the driving transistors and main coil.
2. The second is pulse width modulation or PWM where you regulate the time the coil is energized at a fixed maximum current. This works well with field effect transistors and because they are either fully on or off doesn't have anything like the same heat build up in the transistors and to a lesser extent coil. It also has the added advantage that the varying magnetic filed that the coil creates can be used to transfer a certain amount of energy to a coil that is mounted inside the model, but more on that in the next paragraphs.

Dances with Trees wrote in post number 5

• You might add a red LED to the main thruster to simulate it firing, just for fun.

This is starting to get a little scary as it seems us CR4 type ginger beer really do think in very similar ways.

I actually added a LED where the combustion chamber of the main engines would normally be and it looked quiet good so it's definitely something I will try.

Ok, this is where the PWM rather than constant current control comes in. If I use PWN then by mounting a small coil inside the model I should be able to pick up enough energy from the varying magnetic field to drive a couple of LEDs and perhaps enough to drive a small circuit that can be used to mimic the strobe lights they used for visual navigation when carrying out the lunar orbit rendezvous after the landing.

It will be a while before I get this going but I promise to post pictures and maybe even a short video of the final result. However, I am at the moment short of a descent digital single lens reflex camera and need to convince the accounts department (in other words wife) that it's a must have item. Any hinting, grovelling, begging, pleading etcetera that she should buy me that Canon EOS 450D DSLR with the twin lens kit should be directed via the link to Sue hairdresser extraordinaire.

I will also start a thread on the model as soon as I get everything finished and properly documented. The model is primarily constructed from 200 gm^-2 cardboard which it the stuff they make business cards out of and can be purchased in A4 sheets from stationary suppliers. Ultimately the plan is to have a kit that you can download and then with a decent inkjet printer print out the various sheets with the components printed on them. It's then a matter of cutting out the shapes and assembling them to give you the final model.

Thanks for the posts, your thoughts are greatly appreciated and stay tuned folks, there's a lot more to come.

Sounds like a very cool project.

A few notes on the direct current vs PWM. In order to levitate the object at a point in space it will take X amount of force/energy. This amount won't change whether you use PWM or direct current control. If you use PWM then the power of the pulses will be proportionately higher than the direct current you are using. Therefore the heat dissipation issues won't go away with PWM and the transistors/FETs you are using will need to be able to handle higher current pulses. In addition using PWM in a coil will cause extra heat/energy loss due to induction and eddy currents generated.

If you use the PWM pulses to power the lights in your model through some form of induction then realize that while you are drawing power from the coil it will create a bit of back EMF (see Lenz Law) and it will affect your management of the levitation. This probably won't be a big issue if you are using some form of basic feedback loop but keep it in mind if you see weird effects when you are testing. A basic watch battery setup may be easier to manage even if it is a far less elegant solution.

PWM will create an audible hum in the coil if the frequency is less than roughly 15kHz.

Direct current control will require a more complex circuit and more parts. I think in some ways the PWM solution is actually simpler to implement. Either way you go be careful to protect your microcontroller from any back EMF from the coil. Completely isolating the circuits would be best but some good diodes and caps may be good enough.

Have fun with this.

-Doug

Funny how so many of us think alike, scary as well.

I will be watching for the future post, sounds like a really FUN project, I wish i had more time to play with something like this one. Most of my time is now taken up building big nasty off road trucks, which reminds me I really have to get back to my off road camper planning.

have a great weekend.

Kevin

Agreed, tuning an analog PID is a bear and very time consuming even with a good mathematical model of the process. But you don't need full PID, just a little lead compensation. See this paper:

http://web.mit.edu/kumpf/www/MagLev/Kumpf-Writeup.pdf

G'day stevem

Thanks for the post.

I had already seen the article and it's definitely got plenty of the good stuff.

However, since I ultimately intend to animate the system rather than just suspending it means that it would more than likely be easier to achieve using a PID rather than the concept in the paper.

PIDs are also fairly easy to achieve utilizing microcontrollers. The simplest way is to take a series of samples at regular intervals and storing a second or so worth of values in a FIFO table. The derivative can be obtained by looking at the difference between each of the samples while the integral can be obtained by summing an appropriate number of the values.

Regards, masu.

I have never tried to build what you have done so far, and it sounds fascinating.

I can picture the concept very clearly, thanks for that, now please correct me if I got this wrong; the processor is going to control the voltage to the coil to control assent and decent speed.

the consept is sound, and I don't see why it would not work.

You might add a red LED to the main thruster to simulate it firing, just for fun.

On a side note I have the models of the service module, and the lander around here someplace, I built them way back when the lunar landing was big news.

You could also try fuzzy logic with the microcontroller.

Hey

Since you got me thinking I started browsing. I came across this page that had lots of relevant information;

http://bea.st/sight/levitation/

He is doing the levitation on a microprocessor based system also, as well as using coupled resonance coils to transfer power.

If he is willing to share any of the code or more schematics I would love to see them also.

-Doug

May I give a suggestion regarding using continuous power to the coils rather than PWM?

The heat generated by the coils will require a cooling fan. Normally, a fan would be undesireable for a showpiece like your project but you can put it to good use.

You can put some strips of paper or cloth underneath the rocket motor and allow it to extend a few inches. The air flow from the fan will make the things flutter simulating the smoke and flame trail. The light from the LED will create the flicker effect nicely.

regards,

Vulcan

G'day gals, guys & gurus,

Thanks for the posts, they are truly appreciated and have provided me with some exceptional lines of thought that I doubt I would have come up with on my own.

In post #9 Doug wrote:

• A few notes on the direct current vs PWM. In order to levitate the object at a point in space it will take X amount of force/energy. This amount won't change whether you use PWM or direct current control. If you use PWM then the power of the pulses will be proportionately higher than the direct current you are using. Therefore the heat dissipation issues won't go away with PWM and the transistors/FETs you are using will need to be able to handle higher current pulses. In addition using PWM in a coil will cause extra heat/energy loss due to induction and eddy currents generated.

You're absolutely correct in that the RMS power going to the coil is the same regardless of the way it's regulated so there's no real difference in the heat build up in the coil.

However, when it comes to the power amplifier or driving circuit things are a little different. For the moment lets assume we're utilizing a 12 Vdc supply and that the coil requires 5 A at 5 Vdc to produce the appropriate magnetic field.

• Constant Current: The power going to the coil is 5 Vdc x 5 A giving 25 Watts. For the transistor the current is still 5 A but the voltage drop is 12 Vdc -5 Vdc or 7 Vdc which then give us a power dissipation of 7 Vdc x 5 A or 35 Watts.
• Pulse Width Modulation: This is a totally different equation entirely as we have to take into account the duty cycle or time base. Basically when you have power going to the coil you give it the whole 12 Vdc less a small voltage drop across the transistor, which we will for this example give a value of 1 Vdc. That means when the coil has power going to it, it receives 11 Vdc rather than the 5 Vdc of the constant current system. As a result the current is proportionally higher and comes in at 11 A which then give us an instantaneous power going to the coil of 11 Vdc x 11 A or 121 Watts. Now this is far more power than the coil needs so the circuitry limits the RMS value of the power by turning the coil on and off very quickly. In this case to achieve the 25 Watts required means that the coil will be on for 25 ÷ 121 or 20.66%. In other words the power will be going to the coil for around 20% of the time and be off for the remaining 80% of the time. Ok, there's really no difference at the coil but when you look at the transistor it's completely different. First off the voltage drop across the transistor has dropped from 7 Vdc to 1 Vdc while the current has increased from 5 A to 11 A. That give us an instantaneous power of 11 Watts but when we take into account the 20% duty cycle the effective power that the transistor needs to dissipate becomes 11 x 20% or 2.3 Watts.

So for a constant current drive system you need transistors that are capable of supplying 5 A and a cooling system that is capable of dissipating at least 35 Watts. On the other hand with a PWM system you need a transistor that can accommodate 11 A but only needs a cooling system that can dissipate 2.3 Watts which is less than one tenth of the power with the constant current drive.

Now this is a considerable simplification and working with electromagnetic fields can be one of the most complex things an electrical engineer ever comes across but it demonstrates the point that PWM circuits are usually more efficient that current regulation.

• If you use the PWM pulses to power the lights in your model through some form of induction then realize that while you are drawing power from the coil it will create a bit of back EMF (see Lenz Law) and it will affect your management of the levitation.

Absolutely correct. When I do get around to building it I will almost certainly get it going without the pick up system in the model, actually as Dell suggested I will use a simple ballast weight made of the same materials and having the same mass as the model.

However, utilizing a microcontroller gives you the ability to rehash the feedback by simply changing the software. Not only does this make it simpler to get it working in the first instance but makes it a whole lot simpler to accommodate different models that have different shapes, centres of gravity, masses, materials and endless list of other variables. It's also the only way that you could even possibly consider the animation concept.

By the way, thanks very much for the link. I've only had a glance at it but it sounds almost exactly what I am trying to achieve.

Secondary Coils

Something else I noticed in one of the sites was the reference to the use of additional small secondary coils that are used to rotate the item being levitated. The idea was that over a minute or so it would rotate the item so you can see it from different aspects and get a better overall view.

It got me thinking that by adding a series of secondary coils it would be possible to modify the magnetic field so that it results in a magnetic field that is intrinsically more stable than the field from a single coil. I haven't thought this through and it's just something that jumped into my deranged mind so it may be total crap, but has anybody got any thoughts on the use of secondary coils?

In post #12 Vulcan wrote;

• The heat generated by the coils will require a cooling fan. Normally, a fan would be undesireable for a showpiece like your project but you can put it to good use.

I had assumed that the system would need some fans but I do like your ideas of adding something that enhances the display. It may also be possible to utilize the exhaust to rotate the model rather than messing about with secondary coils.

I have done a reasonable amount of research into the use of cooling fans. For some reason designers prefer to utilize fans that suck hot air out of systems rather than blow cool air in. It is a little easier to design cooling systems that suck rather than blow but they have the phenomenally bad side effect of turning whatever it is into a vacuum cleaner that clogs everything up with dust.

Going back over 15 years I carried out a series of experiments that replaced the sucking fans with blowing fans that had directing vanes/ducts, filters and flow sensors. The end result was slightly more complicated but not only did it keep kept the inside clean and dust free, but the electronics ran between 2°C and 5°C cooler than the sucking fans.

I also built a home automation system that ran Windows 3.1 and a special software package that was based on a 386 processor board and 30 Mb hard disk drive. I mounted it in an old 19" lowboy modem cabinet I had lying about and installed a series of fans that kept everything inside clean, cool and dust free. The result was a phenomenal increase in reliability and over the 13 plus years the system operated for 24 hours a day 7 days a week it never crashed even once, never had a disk error or any other error. The only unplanned or unwanted downtime was for a couple of hours when an extended power outage ran the batteries in the uninterruptible power supply flat.

It also kept the other half happy as the filter collected all the dust in the air and kept the room almost totally dust free. All you had to do was run a vacuum cleaner over the filter once a month or so and bingo, it was all clean.

Anyway, time to get back to the model so thankyou all, your comments and thoughts are greatly appreciated.

Regards,

masu.

Not sure I can help, but the concept was intriguing enough that I found this article:

http://www.dip.ee.uct.ac.za/eee4022/upload/2005/wlllan004-ElectromagneticLevitationThesis.pdf

One thing it brought up was a deficiency inherent in using an ultrasonic sensor; and some alternatives. I see no reason why one of those would not work, with the microcontroller taking the place of their control circuit.

Hey Masu

I came across another great link for you. This time at instructables.com.

http://www.instructables.com/id/Electromagnetic-Floater/

This one even has the code available. Please keep us updated as you make progress.

-Doug

G'day Doug,

Thanks for the link.

I'm not sure that I will use the code as it would involve a considerable amount of patching to get it to do what I have in mind. Personally in situations like this I have found that it is nearly always better to start from scratch rather than modify existing code.

What I think happens is that the way people organize their thoughts and flow of logic within their minds is reflected in the way their programs are organized and flow. Since there are for all intense and purposes an infinite way of organizing the human mind there is a myriad of different ways to write code that produces a given final result.

Anyway, the important thing is that the concept of utilizing a microprocessor rather than a bucket load of analogue circuits and feedback loops is fundamentally sound and is definitely the way to go.

Things are also progressing at this end even if they are at a snails pace. Fortunately I managed to convince the financial director (wife) that investing in another lab power supply was a good idea. I already had a 3-30 Vdc 30 A lab power supply but while it was overload protected there was no way to use it as a constant current source rather than constant voltage source. While the new power supply can supply 3-30 Vdc it has additional circuitry that allows you to set it so it pumps out a variable voltage and constant current between 0-5 A. This will not only mean that I can isolate the coil and driver electronics from the other electronics but limit the damage that can be done during the design and testing stage.

So, I had better stop drivelling on and get back in the workshop.

Regards, masu

Nice work Masu! Great detail.

Keep us informed as this proceeds. I am especially interested in your work with the levitation electronics.

-Doug

There's no way to tell the scale though. How big is that thing? Can't be too big, that's for sure.

But, big or small, the detail is astounding! Nice piece of work, Masu!

regards,

Vulcan

G'day Vulcan & Doug,

· How big is that thing?

Sorry about not including a scale, it just didn't occur to me at the time but in future I will include at least one image like the one on the right that either has a major dimension or something that will give you the idea of the size.

In this case the model is ¹/₃₂ scale which makes the distance between the centres of opposite foot pads a little over 266 mm (10.47 inches).

The LM model is actually the first of a series that will ultimately include not only the Apollo Command & Service Modules but the Saturn V launch vehicle as well. Originally I was going to make the whole thing ¹/₄₈ scale but I found that the level of detail I was aiming for was difficult and some times impossible to achieve so I increased the size of the model to ¹/₃₂ scale. That will mean that the full stack of the Apollo Spacecraft and Saturn V 3.366 m or a tad over 11 feet long. That might sound large for a model but the Saturn V is to say the least an extremely large launch vehicle that the smaller scales just don't do it justice.

Ultimately I would like to develop a version of the model that shows the interior detail of the Saturn V which would be very difficult with the smaller ¹/₄₈ scale.

Regards, masu