Is your concern about overheating of the electromagnet? At this frequency, you have left the world of DC and entered the world of AC. And it is not clear if the applied signal is sinusoidal or a square wave. This information could make a difference.

It may be possible to cool the magnet using some liquid but without more information, the suggestions may not meet your application needs.

By the way, many people interpret the use of all capital letters as YELLING! Please feel free to turn the "Caps Lock" off.

dear sir

thanks for telling me about caps lock.

actually its not a sinusoidal. i have to change the polarity of electromagnet and in my application i have dc source only. thats why i need to change polarity of dc source to change magnetic field direction of electromagnet and switching is with the frequency of 25 hz (its mean switching of dc source polarity from + to - and from - to + 25 times in a sec.).

If your concern is for the electromagnet, I am supposing it would be hard to rewind or have very thin wire.

My first question is, "What is the ampacity of the wire in the electromagnet?"

Without knowing more about your device, I can't tell much about it. There is a trade off between the number of turns and the current for a fixed amount of magnetic flux. So, if the issue is too much current, then you may be able to solve the problem with more turns (loops) of magnetic wire. But I don't know what constraints you have. More turns would generate more resistance and then thereby reduce the current. At the same time, the switching makes your electromagnetic take on some AC impedance properties that is frequency dependent. Consequently part of your total current limit at 1.5 Amps is a function of resistance(R) and inductance(L). Your total impedance Z = R + L.

Actually, because your reversal is more of a square wave, the actual impedance is more complicated. The square wave contains many harmonics of 25 Hz and the resultant impedance Z = R + L(at 25 Hz) + L(at 50 Hz) + L(at 75 Hz) and so on to some point where is is considered to be unimportant.

With that in mind, the geometry of your electromagnetic could be important. For instance, if you have a coil wrapped around an iron core with square corners, then the radius of the bend in the wire on that corner may become significant.

It may also be important to take into consideration what methods of heat dissipation you can use. For instance, must this electromagnetic be powered up at all times? Or is there a reason why you can't increase the frequency?

As you can see, for something so simple as a coil of wire, there can be many considerations. And then, you must think about the device doing the switching because it is possible to zap the electronics or switching device when the magnetic field is removed or reversed rapidly. Perhaps you can describe your electromagnet in some more detail? And, throw in a description of the temperature problem in more detail as well.

It's not damage to the electromagnet you should be worried about, it's the inductive kick flyback voltage damaging your reversing switch. If possible, it'd be good to evaluate your coil's inductance and winding resistance. It has an energy E = ½ L I² to get rid of, which it does with a flyback voltage spike if you open the circuit.

Reversing Relay. We had an application like that, and our first version, which we threw together in a few hours, was just a DPDT relay with a transistor driver.

Our coil, actually an electromagnet we had made, had an inductance of 655uH and a resistance of 0.33 ohms. When we ran it at 10A DC to get the field we wanted, it dissipated P = I² R = 33 watts.

Snubber. But we wanted a rapidly-reversing field, with the reversing relay. We didn't want the relay contacts to wear out from arcing, so we added a snubber, with

C > L I² / V²,

where V is our flyback voltage limit (ignoring the series loss resistance). For I =10A and V = 100V, we get C = 6.6 uF, so used a bipolar 10uF 160V cap.

Natural time constant. When the relay contacts close, the field decay time is stretched to τ = L / R = 2.2 ms, so we wouldn't want to run the relay any faster than say 100Hz. Whew, what a clatter it made, so we didn't run anywhere nearly that fast.

H-bridge reverser. To go fast, we needed an electronic reversing switch. The most common form is called an H-bridge. This has four MOSFET switches and directs the current through the coil (or it could be a transformer) first one way, then the other. The bottom two switches are wired together and provide a convenient place to add a current-monitoring resistor. If you like, the power supply can be a constant-current type. It'll adjust the voltage until the average current matches its setting.

High flyback voltage. If we want to go faster than the L/R time constant, we need to let the coil flyback to a higher voltage, where its current decays faster, given by dI/dt = V/L.

Zener flyback clamp. The drawing shows four added diodes to allow the flyback, and a zener diode that limits V to about 100 volts. We'll need to use 200-volt 10A diodes and bottom MOSFETs.

Gate driver IC. The drawing is greatly simplified, lots of pieces are required, but one thing that helps us greatly is to use an H-bridge driver IC. The part shown, an Intersil HIP4080A (datasheet link), part of one of my favorite MOSFET driver families, is available from DigiKey for $5.82 each.

I can provide more details if you like. Ask for file RIS-626 for a complete schematic.

that is schematic drawing of my project. its a magnetic driven piston. polarity reversal i will take by rotary mechanism with help of some mechanical arrangement. i just need how to make electromagnet and supply source save during reversal of polarity. and how to make cooling arrangements.

When your contacts, or your switch reverses, your coil's inductance keeps the current flowing in the same direction in the coil. Immediately after switching it's unchanged. It does this by creating a flyback voltage, which can be damaging to your switch's contacts if it's not dealt with as I described.

If you have instant switch reversal, you'll see something like the top waveform in the drawing. At the new voltage the current starts dropping to zero and then changing to the new polarity, with a time constant τ = L/R, as shown in the drawing's second waveform.

You asked about protecting your power supply. During the first part of that time constant, current will be flowing into your source, rather than out, as shown in the 3rd waveform. If it's a conventional power supply, it's designed to only source, and not sink current. But it'll have an output capacitor, which will have to absorb the reverse current. During this time the output voltage will rise as the capacitor is charged, as shown in the bottom waveform.

(In addition to the voltage hump from the reverse-charging current, the last waveform also shows step and spike artifacts from the capacitor's equivalent series resistance and inductance. The hump is actually a half-cycle** of the resonant frequency of your coil's inductance and the power supply's output capacitance.)

We can find the amount of capacitor voltage rise by equating the energy in the coil and the energy in the capacitor, E = ½ L I² = ½ C V². We can crank this relationship, and get a voltage rise of V = I sqrt ( L / C ).

Or we can rearrange to get C = L (I / V)², which lets us determine a minimum necessary output capacitance, given an acceptable voltage rise.

For example, with our 655uH coil running at 10A, if we want to limit the supply rise to say 1 volt, we'd need at least of 65500uF output capacitance! Oops, that's quite a lot (pictured, a UCC 68,000 uF, 25V cap), so if we compromise and abuse the power supply by allowing say ΔV = +3 volts (ouch, but probably OK), we can get by with (only) 7280 uF of external capacitance. Some power supplies might come with 7500 uF or more of built-in capacitance on their output terminals, but watch out, it might not!

So, as I said earlier, you really do need to know the inductance of your coil, and pay attention to the energy stored there.

** If you know (and are sure) that your power supply can handle the reverse charging and the associated voltage-rise, you may prefer to let this happen, and use it as a quick way to recover the energy from the coil and re-insert it in the opposite direction.

dear win thank u for such up date. one more question if i change my power supply with a 12 V dc battery of 64 Ah. then i think i don't need to use capacitor. am i right???

coil will quickly charge battery again ???

that is the mechanism for changing polarity. half circle and full circles are of copper at fixed plates while centre rod is rotary having carbon brushes. left and right side of image will assembled at each other at a common axis. right side is connected with battery and left side connected with coil.rotating rod is common.

Yes, a battery can happily absorb the reversing current. As for quickly, unless you allow flyback and resonant storage, the current change will look like a simple R-C charging, except it'll be R-L, and with the time constant tau, τ = L / R. Here τ is the time to get to 63% of the reversal, and about five time constants is needed to get to within 1% of the reversed state. This will be much slower than you can achieve in the other ways I described.

If you have a diode that lets the coil current fly back into a capacitor, creating a higher reversing voltage, the time to reverse is a half cycle of the resonant frequency,

t = pi √ L C,

so if you select a small enough C you can get a very fast reversal time, much faster than just L/R.

Perhaps you can try some values, using the formulas in my previous post.

How fast do you want to go? It's time for you to calculate how much field you want, and then calculate further and see how much inductance your coil will have, and then see what the time delays will be.

If a simple τ = L/R speed is fast enough, then life will be simple for you, because you can purchase a motor-control PCB, with a set of four H-bridge switches, and work with their low voltage capabilities. You can add some capacitance to the power-supply output to protect it if necessary.

But if you want faster speed, and want to use my circuit above with an isolating diode and a resonant capacitor, you can start by examining the circuit of a commercial motor-control H-bridge to see if it can handle such treatment, or you may have to make your own H-bridge, using a driver IC and four FETs. My favorite HIP4080A is rated at 80 volts, so you can get quite an improvement over a simple 12V circuit.

Inexpensive H-bridge or half-bridge (use two of them) driver ICs are available with 600V ratings (IRS2104, IRS2108) and even all the way up to 1200 volts (IR2213), so you can play this fast reversal game to get quite an impressive speedup, or work at high reversing frequencies.

The H-bridge is a great solution. If you choose to use it, just remember to include a little time between reversals for the "switches" to turn off. In the example below, you turn on T1 and T2 at the same time for the positive polarity. You must turn T1 and T2 off (maybe 1 or 2 mSec depending on the transistor type) else when you turn on T3 and T4 you will short out the power supply.

dear sir ,

i have no option to use electronic switching so i cant use h bridge. my possible switches are shown under it a rotary switches rotating with the double speed of electromagnet. as shown in above answer. schematic

of magnetic piston. by these switches i can get exact time and distance at which i have to change polarity otherwise i got another challenge to calculate time and distance and that is very short. any ideas????? thanks.

A pair of DPDT switches or relays are equivalent to an H-bridge, and as I showed in the drawing, you can use the resonance flyback trick to get very fast current reversal, and save most of your E = ½ L I² energy while you're at it. But to protect the switches from arcs you'd need really quick double-throw switching. You said you wanted to run at 25Hz, that's very much on the fast side for mechanical switches dealing with high currents and inductive kicks. But there's no reason you can't use a mechanical or optical sensing technique to get the logic signal to run a fast transistor H-bridge.

And like I said, if you keep your voltages low (e.g., 10V or less, use fewer turns, bigger wire and higher currents) you can probably play the resonance-flyback game with one of the motor-driver H-bridge PCBs available from the huge cottage industry that's grown up around them. Some of the lower-voltage H-bridges have built-in switching transistors. I like types that use MOSFETs, because they usually dissipate (waste) less switch power. The LMD18200 or LMD18201 (links) are good ones that're are rated at 55 volts, and 3 amps, and cost $17 at DigiKey. You can also get the '201 for half-price on eBay. Sparkfun sells a "breakout board" for these parts, for the amazing low price of $1.95, so you can easily make you own.

This would allow for a 5x reversing-speed improvement if you use 10V coil drive.

dear sir i am going to use as you told h bridge in configuration of FIGURE 9. Locked Anti-Phase Control Regulates Torque at page # 11, that is given in LMD18200. i have some question about that.

1: Is it control magnetic field strength of electromagnet ????? by increasing torque or current.

2: its not mention how much fast pulse we can give at direction in put.

3: Kindly describe PMW

thanks.