Guys:

I would like to drive some small coils with AC signals at a specific frequency using the outputs of an FPGA (pulsed DC square wave). There will be a lot of these coils in the device (over 200) that are independently driven at different frequencies. I realize that I could drive them with individual H-bridges, but I'm trying to keep the cost down. I had an idea (rough schematic below) to drive them. The idea is to alternately fire a PNP and NPN transistor. When the PNP is "on" and the NPN "off" the capacitor charges through the coil. When the transistors are reversed, the cap discharges back through the coil in the opposite direction, through the NPN to ground.

Would this produce an AC signal on the coil?

Don

 

Yes, the capacitor blocks the DC average component so the current through the coil will be ±AC, but note that the signal is still a square-wave, not a sine-wave (except as possibly modified by the coil inductance).

 

With your transistors configured as shown there will be shoot-through during the rise and fall times of the input pulse edges.

 

You need resistors to limit the current to the bases.

 

Well, I know this. This was intended to be a rough schematic just to show you my concept. And actually, the transistors will be MOSFET's...the BJT's were easier to sketch quickly.

Don

 

Well, I know this. This was intended to be a rough schematic just to show you my concept. And actually, the transistors will be MOSFET's...the BJT's were easier to sketch quickly.

Don

You cannot just substitute MOSFETS into that sketch and expect them to work. You can make an inverter from a single polarity DC source, but the actual circuit is just a bit more complicated. In particular you need to provide for deadtime in the switching waveforms. Oh, also component values would greatly enhance your credibility

 

I prototyped this circuit last night. 0.47 uF seemed to work best for the cap. The inductor is hand-wound and it's difficult to determine its inductance without experimenting. I get a clean waveform on the scope with no spiking or ringing, just using an ordinary square wave input. A frequency of 19 kHz seems to give me the most flux density. And yes, I'm using MOSFET's with no gate resistors.

Don

 

The pulsed DC signal is already an AC signal, but not a sine wave. If the duty cycle of the signal is close to 50% then driving the coils directly can work out quite well, if the drive circuit has a reasonably constant resistance, both on and off.
I doubt that you are driving the coils directly from the FPGA, so please let us know what the actual coil drive circuit will be, presuming that part of the design already exists.
Depending on the required current it might be as simple as using a CD4050 hex driver IC to drive them, if the drive is less than 15 milliamps and less than 15 volts.
So we need information about the coils and the drivers. For a bit more power one of the 5-lead TO-220 size audio amplifiers could work.

 

Would this produce an AC signal on the coil?

That appears to be pulsating DC, rather than AC which has very different characteristics when used with a device made for AC operation??
AC = the current/polarity Reverses direction every cycle.

 

When the PNP is off and the NPN is on, the cap should discharge back through the coil, reversing the current direction.

Don

 

When the PNP is off and the NPN is on, the cap should discharge back through the coil, reversing the current direction.

Don

Maybe, but the inductor is going to counter that, it does not like changes in current. The inductor has a voltage of L*(di/dt). It all depends on the inductance of the coil, the frequency you are dealing with and the capacitance. Being you want different frequencies this complicates the situation.

 

Sure it's going to counter it. That's what it does when AC is applied. I don't understand how this would be any different than reversing the polarity with an H-bridge, other than the fact that the current will be transient.

Don

 

Really, just putting a capacitor in series can allow a reversal of current IF the driver has asimilar impedance at both voltages, high and low. That is why I suggested the driver device.

 

I rigged this up with LTSpice, just an example (attached). What I found makes sense if you think of it in the frequency domain. The input square wave has little to do with the frequency of the AC current. It appears the AC current is where you would expect it, at the resonant frequency of the LC circuit. If you increase the frequency of the square wave, you get a smoother version of the AC resonant frequency through the circuit. That makes sense being all a square wave is, is the sum of a bunch of odd harmonics starting at the fundamental frequency and going upward. The LC circuit just filters out the one that it is resonant at.

You cannot change the natural frequency of the AC and is has nothing to do with the frequency of the driving square wave.

In other words it is just a band pass filter circuit.

Of course this implies a 'perfect' square wave driver source, which does not exist.

 

OK, so tuning each coil to resonance, either series or parallel, will allow the coils to operate at a resonant mode. If each coil always gets the same frequency .

 

OK, so tuning each coil to resonance, either series or parallel, will allow the coils to operate at a resonant mode. If each coil always gets the same frequency .

yes, you can set the resonant frequency by just changing the capacitor value per C=(1/sqrt(L))*1(/2pi)*1/(sqrt(f)).
But make sure the square wave frequency is >10 times the resonant frequency for a nice sine wave, the higher the square wave frequency the better.
But this assumes a perfect driver circuit that can 'instantly' go from sourcing current to sinking current, which does not exist.
Also the series resistance of the coil must be low for a high Q. And the current at resonance would be inversely proportional to the coil resistance, you could easily reduce it with a series resistor if it is to high but at the cost of efficiency of course.

 

OK, so tuning each coil to resonance, either series or parallel, will allow the coils to operate at a resonant mode. If each coil always gets the same frequency .

Would not recommend parallel, the capacitor needs to be in series with the inductor to limit changes in current.

 

I wonder why the TS wanted to drive them with AC to begin with?

 

This is another instance of not enough information. MY GUESS is some sort of musical device with the electromagnets driving some musical device. Or a piece of art, no clue.

 

It's amazing how far a thread can go with
no one having the slightest clue as to what the application is !!!
.
.
.