Electromagnet driven by capacitor

[Elledan]

I'm working on a design involving an electromagnet which receives its power from a capacitor, meaning that the electromagnet is only powered in pulses (when the capacitor discharges).

Since the current going through the coil of the electromagnet is limited only by the resistance of the coil, I need to know the maximum frequency at which the capacitor can discharge in order to keep the coil and surrounding structures from melting (or worse).

I realize that the produced heat and the rate at which this heat is removed depends on the design of the coil and the environment, but all I need are some references to sites or other with information which will allow me to calculate the maximum frequency.

From a site on railguns or similar I got the figure of maximum 0.5 Watt/square inch as a reasonable treshold for the heat buildup inside the coils. How realistic is this?

Also, what is the typical time a capacitor requires to fully charge itself?

Any help, tips or requests for more information are welcome

 

[gee]

Originally posted by Elledan
I'm working on a design involving an electromagnet which receives its power from a capacitor, meaning that the electromagnet is only powered in pulses (when the capacitor discharges).

Since the current going through the coil of the electromagnet is limited only by the resistance of the coil, I need to know the maximum frequency at which the capacitor can discharge in order to keep the coil and surrounding structures from melting (or worse).

I realize that the produced heat and the rate at which this heat is removed depends on the design of the coil and the environment, but all I need are some references to sites or other with information which will allow me to calculate the maximum frequency.

From a site on railguns or similar I got the figure of maximum 0.5 Watt/square inch as a reasonable treshold for the heat buildup inside the coils. How realistic is this?

Also, what is the typical time a capacitor requires to fully charge itself?

Any help, tips or requests for more information are welcome

building a coilgun or railgun? fun times, I've always wanted to build one tell more! tell more!

Anyway, the amount of energy stored in a capacitor is 0.5*C*V^2, in joules. If you assume that the capacitor is being fully discharged with each cycle and that no power from the capacitor charging circuit is making its way to the coil, then just multiply this by Hz and you have the amount of wattage that's being driven into the coil.

Is there any particular reason for pulse-driving the coil?

 

[groggory]

Q = Q f (1 - e - t / R C),

http://www.rwc.uc.edu/koehler/biophys/4g.html

that's how caps charge

 

[Elledan]

Originally posted by gee
building a coilgun or railgun? fun times, I've always wanted to build one tell more! tell more!

Heh, hasn't everyone and their dog made a rail/coilgun by now?

Well, I hate to disappoint you, but what I'm working on isn't going to shoot projectiles through brick walls or pets unfortunate enough to be in the wrong spot at the wrong time.
The best way to describe it, I guess, is as an artificial muscle, i.e. a structure capable of altering its length even when working against another force. A picture would help a lot, but in short it consists out of a string of identical units, each unit basically consisting out of a coil (the electromagnet) with an iron core which fits inside the coil. A piece of string (I have yet to determine which material I'll use for this) is attached to one side of the iron core and the closed side of the coil of the next unit.

Now, when the electromagnets are activated, the iron cores are pulled inside the coils, effectively shortening the length of the 'muscle'.

Is there any particular reason for pulse-driving the coil?

Two reasons:

- Muscles are pulse-driven as well, so it makes sense to stick to that design.
- With a capacitor there's no need for a big, fat powersupply to deliver lots of current. Due to the nature of capacitors (either charge or discharge) pulse-driving is inevitable.

Anyway, thanks for the information, gee and groggory!

 

[cpemma]

Originally posted by Elledan
I'm working on a design involving an electromagnet which receives its power from a capacitor, meaning that the electromagnet is only powered in pulses (when the capacitor discharges).

Since the current going through the coil of the electromagnet is limited only by the resistance of the coil...

Not quite true, the coil has inductance that resists any sudden changes in current flow, so the pulse is blunted/current reduced and the capacitor will take a bit longer to discharge than a simple RC time constant (tau) would forecast. Best seen on circuit simulator software, with the coil as a resistance and some guesstimate (or calculated) inductance values in series.

http://mgc314.home.comcast.net/inductance1.htm might help, shows the blunting effect.

 

[mattg2k4]

http://www.oz.net/~coilgun/mark2/rlcsim.htm This is a nice RLC simulator which will let you see how various parameters affect the current flow. There are also some pages with theory and an inductance calculator for a coil w/ a certain guage wire, ID, OD, and length, so you might want to check out the site map.

 

[Elledan]

cpemma, mattg2k4: thanks for those links and information!

This should be enough to get me started, I hope

I'll be back later to bother you guys with more questions

 

[plot]

coils take some time to respond to current changes. anyways, for testing different circuit perameters, i recommend using a spice simulator.

here's a good free one that tells everything and more about basic circuit simulations (power dissapated, peak current, etc.)

http://www.linear.com/software/

first one on there, LTSpice Switchercad. pretty intuitive so it's easy to use. only 4-5megabytes to dl, takes like 15megs HD space max (i have it installed on my pen drive...)

 

[Elledan]

Originally posted by plot
coils take some time to respond to current changes. anyways, for testing different circuit perameters, i recommend using a spice simulator.

here's a good free one that tells everything and more about basic circuit simulations (power dissapated, peak current, etc.)

http://www.linear.com/software/

first one on there, LTSpice Switchercad. pretty intuitive so it's easy to use. only 4-5megabytes to dl, takes like 15megs HD space max (i have it installed on my pen drive...)

Looks like a great program for simple circuit simulations. Thanks for the link

 

[Aristarchus]

I'll second the notion that you have to factor in coil inductance in your calculations. If you get into any substantial drive frequency at all, it will very quickly become the dominant factor in your equations. Getting a good idea of the coil inductance you have to work with will be a very important step. Also, the coil inductance will almost certainly change with the position of the servo rod, i.e. inductance will be a function of extension.

Also, I'm interested in why you're choosing to go with a capacitor driven design. You mentioned that a cap-driven design wouldn't require a beefy, high current power supply. I'd agree with that principle, with the caveat that your final, high current output from the cap would have a limited duty cycle, that's proportional to the ration of PSU output current to cap output current, in an averaged sense. Please keep in mind that there's nothing magical about capacitors; they only store energy, they can't create it. ALso, keep in mind that when you start working with reactive components (inductors/capacitors) that your resistive losses in your supply leads increase, due to voltage & current being out of phase, and as a result your overall system becomes less efficient.

Please allow me to suggest a different tack to your design that may be beneficial; Rather than using caps to store & release energy, build a switching array that can power each individual relay at full power in sequence. By incorporating catch diodes to allow the stored energy in each coil to recirculate through the coil when turned off, and cycling through the coils fast enough, you should be able to achieve the effect you're looking for without a big capacitor array.

On the other hand, depending yon your design objectives, you might just want to bite the bullet, and build a supply large enough to drive all the coils at full power simultaneously. You'll get a lot more motive force out of your whole assembly, have a much simpler drive system, and by incorporating some simple pwm, you'll be able to achive variable force quite simply. Also, let me suggest connecting the coils in series, and letting the voltage do the work, rather than the current. Doing so decreases your resistive losses by an exponent of 2. (rememper, P=I^2xR)


Have fun, and good luck. Oh, and BTW, DigiKey has some nifty logic level MOSFETS from fairchild semi that cost about $.80 and are rated to 60V and 50A or so, that might be just what the doctor ordered for your application.

 

[Elledan]

Originally posted by Aristarchus
Also, I'm interested in why you're choosing to go with a capacitor driven design.

Blame my lack of experience with electronics for not choosing the most efficient design

Basically, the reason why I went with this design was because I knew it would work, but not how well.

Please allow me to suggest a different tack to your design that may be beneficial; Rather than using caps to store & release energy, build a switching array that can power each individual relay at full power in sequence. By incorporating catch diodes to allow the stored energy in each coil to recirculate through the coil when turned off, and cycling through the coils fast enough, you should be able to achieve the effect you're looking for without a big capacitor array.

With a switching array I assume you mean a switching powersupply or something similar?

I had catch diodes incorporated in the original design as well, since I used transistors to drive each string of units (relays) with a low-voltage signal.

On the other hand, depending yon your design objectives, you might just want to bite the bullet, and build a supply large enough to drive all the coils at full power simultaneously. You'll get a lot more motive force out of your whole assembly, have a much simpler drive system, and by incorporating some simple pwm, you'll be able to achive variable force quite simply.

How does this differ from the previous design with a switching array?

Also, let me suggest connecting the coils in series, and letting the voltage do the work, rather than the current. Doing so decreases your resistive losses by an exponent of 2. (rememper, P=I^2xR)

Unfortunately I won't be able to do this, since each coil is driven independently from the other coils.

Have fun, and good luck. Oh, and BTW, DigiKey has some nifty logic level MOSFETS from fairchild semi that cost about $.80 and are rated to 60V and 50A or so, that might be just what the doctor ordered for your application.

I would say that 3 kW is plenty to drive those coils. If it doesn't cause them to melt, of course

 

[Aristarchus]

What I meant by a "switching array" is having a dedicated power switch (transistor, MOSFET, whatever) to turn each coil on and off. Sounds like you may already be doing this, so you might be in a position to just play around with control strategies, and see what works best.

As far as the catch diodes-- good call. A lot of total noobs would run a large coil on a 12v line or something with a 30v rated transistor, and wonder why the thing quit working after the second time...

As far as my suggestion of driving all the coils at once, this would mean having a single power switch running everything, rather than 1 for each coil. At that point, power control can be as simple as a 555 with a pot, as opposed to something requiring more complex logic circuits.

And as far as connecting the coils in series, you shouldn't let your existing design dictate what your final design should be If you have a low voltage power supply, then you might have a bit more of a design problem on your hands, but if you're running off an AC line, and using 12v solenoids or some such, there's nothing to keep you from connecting the 10 in series. (Just make sure to have things properly insulated if you do!) I'm guessing that you're trying to end up being able to control position, when all is said and done. If you connect all the coils in series and use a PWM approach, what you end up controlling is force. With a little creativity, you can develop a control system that will allow feedback from a position sensor, and set the PWM duty cycle accordingly to exert the required amount of force to maintain the "arm" at the desired position... But perhaps that's beyond the scope of your project.


By the way, what's the purpose of this device? Science project of some sort? I'm curious

 

[Elledan]

Originally posted by Aristarchus
What I meant by a "switching array" is having a dedicated power switch (transistor, MOSFET, whatever) to turn each coil on and off. Sounds like you may already be doing this, so you might be in a position to just play around with control strategies, and see what works best.

Indeed, that's pretty much what I had in mind when I created this thread: a capacitor charged by a power source, which discharges through a coil w/ a catch diode in parallel once the transistor receives the low-voltage signal.

As far as the catch diodes-- good call. A lot of total noobs would run a large coil on a 12v line or something with a 30v rated transistor, and wonder why the thing quit working after the second time...

Doing something like that can give interesting effects, though

If you have a low voltage power supply, then you might have a bit more of a design problem on your hands, but if you're running off an AC line, and using 12v solenoids or some such, there's nothing to keep you from connecting the 10 in series.

I was planning to use a laboratory power supply, capable of delivering up to 40 V/5 A, at least for the initial testing.

I'm guessing that you're trying to end up being able to control position, when all is said and done. If you connect all the coils in series and use a PWM approach, what you end up controlling is force. With a little creativity, you can develop a control system that will allow feedback from a position sensor, and set the PWM duty cycle accordingly to exert the required amount of force to maintain the "arm" at the desired position... But perhaps that's beyond the scope of your project.

This is a very interesting thing you're proposing, but I fear that it won't work for the application I'm creating this design for.

Being able to control the exerted force in such an elaborate manner, aside from being quite complex and a potential nightmare to develop and maintain, is basically overkill.

Although, it's resolution is far higher than that of the design I proposed, especially when two of such 'muscles' are used in an antagonistic pair.

I'll put your proposal in the 'things to consider'-file for now

By the way, what's the purpose of this device? Science project of some sort? I'm curious

A personal science project, more or less. Basically what I'm trying to do is to construct an artificial muscle, which I might end up using in a robot or something like that. Regular robotics just isn't extreme enough for me, I guess

 

[Aristarchus]

One other thing I just thought of: how do you intend to control the inrush currents to the capacitors during the charging cycle? charging a cap without a respectable series resistance is going to give some nasty peak currents, so you may do well to add some heavy inductive filtering at the main power input.

Also, depending on your mechanical setup, getting a feedback controller together wouldn't need to be all that difficult. If you can stick a good quality potentiometer in a place where it's linearly related to position, (or even nonlinearly related, if you don't mind having nonlinear controls...) you should be able to put together an analog controller for a few bucks that could get you what you needed. Power supply concerns aside, if you stick a few connectors in the right places, you should be able to swap out your simple board with a feedback control board in a matter of minutes. Given that you're working with a "slow" mechanical system, you might be able to get by with a simple proportional-integral controller... All of two op amps with a few resistors and trim pots to set the gains to suit you.

Anyhow, my room still isn't getting cleaned...<sigh>

have fun eh. playing is the best way to learn this stuff

 

[Elledan]

Originally posted by Aristarchus
One other thing I just thought of: how do you intend to control the inrush currents to the capacitors during the charging cycle? charging a cap without a respectable series resistance is going to give some nasty peak currents, so you may do well to add some heavy inductive filtering at the main power input.

To be honest, I depend on you guys to think of such things, since I lack the experience to foresee all potential problems

Could you elaborate on this inductive filtering?


Anyway, as you mentioned, the control circuit doesn't appear to be that difficult to design. It'll be fairly straightforward as it's pretty much isolated from the coils and high-current/inductive stuff through the transistor which drives the contraption.

If you connect all the coils in series and use a PWM approach, what you end up controlling is force. With a little creativity, you can develop a control system that will allow feedback from a position sensor, and set the PWM duty cycle accordingly to exert the required amount of force to maintain the "arm" at the desired position... But perhaps that's beyond the scope of your project.

I've given this proposal some more thought and I think that it is the way to go if the resulting product is going to be usable. With an antagonistic pair of 'muscles', each muscle can exert a certain amount of force using PWM, resulting in the arm (or whatever) having the capability to be positioned in pretty much any position.

It might be a lot harder than a simple pulse-driven design like my original design, but the resulting product will be more usable (I hope).

Anyhow, my room still isn't getting cleaned...<sigh>

have fun eh. playing is the best way to learn this stuff

Heh, it's all fun and games until someone fries himself during an experiment involving high currents

 

[Aristarchus]

Glad to be of service

If I might pry a bit, where are you at in your schooling? Guessing upper high school, but then again I didn't know much my first couple of years on college either. (Right now I'm working on year #7, on a MS in electrical engineering.... Betcha wouldn't have guessed that....=)

Along those lines also, so you have any background in calculus? I'll assume you do, if you don't, read some more.

Anyhow, the voltage across a cap is proportional to the integral of the current through it over time. Starting from zero charge, a cap connected to a DC voltage source is virtually a short circuit. As time goes on, current flows into the cap, the value of the integral increases, (the cap charges) and the voltage across it increases, until at some point the voltage across the cap is nearly equal to the DC source voltage.

Inductors, on the other hand, work the other way: the current through an inductor is proportional to the integral of the voltage across it over time. Connect an inductor to a DC voltage source, and it initially behaves as an open circuit. As time goes on however, the integral of voltage with time increases, and the inductor eventually begins to act more like a short circuit.

Cliffs Notes:
~Capacitors "like" a constant voltage; they'll supply whatever current is necessary to maintain the voltage across them, to the degree that the remaining energy stored in them permits

~inductors "like" a constant current; they'll supply whatever voltage is necessary* to maintain the current flow through them, to the degree that the remaining energy stored in them permits.

*This can be a fun, annoying, or terrifying thing, depending on the circumstances.


OK, hope you hadn't heard that before, and are tired of it already. People seem to have a hard time getting a handle on inductors, much more so than capacitors. Anyhow, on to inductive filtering. With the PWM setup you're looking at, you're going to have an ugly, jagged current draw. This is bad. Referencing the above, we find that inductors tend to smooth out current waveforms. Therefore, placing an inductor in series with your power supply lead will tend to smooth out this ugliness, and reduce the peak current demand on your power supply. However, this can work to your disadvantage, because the inductor doesn't like the current being abruptly shut off either, and will cause nasty voltage spikes if you don't provide some means for a current flow when you turn your switches off. (catch diodes...) You can also help this problem by putting an appropriately sized cap between the supply lead and ground AFTER the inductor. The two together form what's known as a lowpass LC filter, and when sized right, will make life a lot better.

so my suggestion for you system would be to find a nice big toroidal inductor (www.BGmicro.com had a nice one last I bought there.) and a big cap that can tolerate high frequency charge/discharge, and a diode rated for at least 5 amps continuous. Put the diode between supply and ground first thing, then the coil in series with the supply lead, and the cap between supply and ground after the coil. If you don't put the doide in the wrong way (might be worth buying another diode to prevent this from happening accidentally) you now have a filtered power supply.

Other side notes:
You're using transistors, which are a current driven device, so it isn't nearly as much trouble as it is with FET's, but you should be careful not to run your drive signal lines too close to the power supply lines, or else you can get coupling between them, which could suck very, very badly.

If you are going with an antagonistic pair idea, let me suggest this to you: make up your control circuit with a 555 or some such, and have one solenoid bank be active when the 555 output is high, and the other active when it's low. By varying the duty cycle, you'll have control over the force imbalance, with a single pot.

Solenoids tend not to like being shut on an off rapidly, i.e. thousands of times per second. This is because of magnetic hysteresis effects within the coil material. Long story short, if you apply a high frequency AC voltage to a cheap DC solenoid, it's going to get hot and cook itself to death. Solution: modify my filtered power supply advice above, and make two supply filters, one supplying each "muscle" in the pair. The higher your drive frequency, the smaller the coil and cap you need to get a decently smooth waveform to supply to your muscles. Offhand, I'd suggest a drive frequency around 20KHz. Why? #1, because this thing is going to vibrate at the drive frequency, and it's going to get annoying really damn quick if that frequency happens to be in the audio range #2, this frequency is likely low enough to be within reason for your switches, and filter caps and coils. (Most everything is affected by hysteresis problems, just some things more than others...)

Anyhow, hope you haven't spend a fortune buying 20 some odd transistors and caps yet... If you have, I'd suggest hitching them up to an LM3915 and making a wall-sized audio level meter. Mine has 10 strings of christmas lights stuck on it, and it's really quite entertaining

Have fun, eh.

 

[Elledan]

Originally posted by Aristarchus

If I might pry a bit, where are you at in your schooling? Guessing upper high school, but then again I didn't know much my first couple of years on college either. (Right now I'm working on year #7, on a MS in electrical engineering.... Betcha wouldn't have guessed that....=)

You certainly have a lot more knowledge (and experience) of electronics than I have (who would have guessed ).

I'm currently in the final (6th) year of 'high school'. Thing is that I'm trapped in a different kind of education system than you are, considering the fact that I live in the Netherlands, so it's a bit hard to make a direct comparison.

A big disadvantage of having everything taught in Dutch is that all terms and such are also in Dutch, so I basically have to learn everything all over again (biology, mathematics, (bio)chemistry, physics, etc.) if I actually want to do anything with this knowledge (like discussing a design for an artificial muscle with a stranger on the internet =) ).

Along those lines also, so you have any background in calculus? I'll assume you do, if you don't, read some more.

Enough to follow along, I guess, although I may have to look up some terms

so my suggestion for you system would be to find a nice big toroidal inductor (www.BGmicro.com had a nice one last I bought there.) and a big cap that can tolerate high frequency charge/discharge, and a diode rated for at least 5 amps continuous. Put the diode between supply and ground first thing, then the coil in series with the supply lead, and the cap between supply and ground after the coil. If you don't put the doide in the wrong way (might be worth buying another diode to prevent this from happening accidentally) you now have a filtered power supply.

What orientation does the diode have to be in? Or do I just have to RTFG(oogled)P(age)?

Other side notes:
You're using transistors, which are a current driven device, so it isn't nearly as much trouble as it is with FET's, but you should be careful not to run your drive signal lines too close to the power supply lines, or else you can get coupling between them, which could suck very, very badly.

Ah yes, the classic 'high/low-voltage lines in parallel' issue. If in the physical layout the drive and supply lines are at a 90 degree angle, this should not be an issue, right?

If you are going with an antagonistic pair idea, let me suggest this to you: make up your control circuit with a 555 or some such, and have one solenoid bank be active when the 555 output is high, and the other active when it's low. By varying the duty cycle, you'll have control over the force imbalance, with a single pot.

Sounds like a simple, n00b-friendly solution =)

Are solenoids the only option in this design? That is, if they don't like being switched off and on rapidly, isn't there an alternative, or are the supply filters you suggested enough to keep them from 'letting out the magic smoke', so to speak?

Anyhow, hope you haven't spend a fortune buying 20 some odd transistors and caps yet... If you have, I'd suggest hitching them up to an LM3915 and making a wall-sized audio level meter. Mine has 10 strings of christmas lights stuck on it, and it's really quite entertaining

Don't worry, I'm in fact one of those n00bs who don't immediately rush out to buy lots of (expensive) parts the moment they get a 'great' (at that moment) idea for some wonderful device


On a sidenote, have you ever noticed how your notes are always less readable than you think right after jotting them down? That's the first thing I thought while reading back the notes I took while dissecting your post

 

[Aristarchus]

DOH! Sorry if I'm a bit US centric. Being stuck smack dab in the middle of North America, I often forget that the 'net is global. But yeah, I can see how learning everything in Dutch could be a bummer. (By the way, you wouldn't happen to have any great or great-great grandparents by the name of Hoekman, would you?)

As far as the diode is concerned, just think about it for a sec. A diode between DC supply and ground. The "wrong" way would be the orientation that causes a virtual short, and turns the diode into an incandescent light bulb.

As far as board layout is concerned, putting the leads at 90° will help reduce crosstalk commensurably, but if you're talking about a MOSFET with an input capacitance of a picofarad, it doesn't take much. That's why transistors are a bit more forgiving in this respect- being a current driven device, a picoamp-level induced current isn't going to make a bit of difference in a switching application.

As far as other design options, wire wrapped around a magnet is pretty much the only cheap option for transforming current into linear motion. The supply filtering I described, if properly designed, should be able to take care of things in this respect. Besides, solenoids are pretty rugged devices, since after all, they
re just a bunch of wire. If they start getting uncomfortable to touch, you probably need better filtering...

Glad to hear that you didn't buy a whole bunch of stuff you didn't need...

Yes, I very often do find that my notes are incoherent when I try to read them. Often times, this is due to the fact that I'm writing them without looking, in order to keep up with the guy writing on the whiteboard in front of the class...

...I find using the quote feature rather unwieldy... Sorry if it makes reading my posts harder

Good luck, have fun eh. Feel free to ask if you have more questions

 

[Elledan]

Originally posted by Aristarchus
DOH! Sorry if I'm a bit US centric. Being stuck smack dab in the middle of North America, I often forget that the 'net is global.

Shocking, isn't it?

(By the way, you wouldn't happen to have any great or great-great grandparents by the name of Hoekman, would you?)

Not as far as I know.

As far as the diode is concerned, just think about it for a sec. A diode between DC supply and ground. The "wrong" way would be the orientation that causes a virtual short, and turns the diode into an incandescent light bulb.

So you meant the diode between supply and ground in your earlier post. I thought you were talking about a second diode. And yes, having a diode 'light up' is a bad thing =)

As far as board layout is concerned, putting the leads at 90° will help reduce crosstalk commensurably, but if you're talking about a MOSFET with an input capacitance of a picofarad, it doesn't take much. That's why transistors are a bit more forgiving in this respect- being a current driven device, a picoamp-level induced current isn't going to make a bit of difference in a switching application.

That's another reason why I like using transistors: they're quite forgiving, as long as you don't abuse them too much.

...I find using the quote feature rather unwieldy... Sorry if it makes reading my posts harder

I mostly use the quote feature to make it easier to reply to longer posts without having to scroll between the post I'm typing and the one I'm responding to. It keeps one from missing questions and things like that as well.

Good luck, have fun eh. Feel free to ask if you have more questions

Sure thing