Self-Switching MOSFET: Exploring Frequency Oscillation with IRF840

[Jdo300]

Hello All,

I was playing around with a small circuit that used a MOSFET to pulse power from my DC power supply into a small coil that I wound (I was messing around with BEMF in the coil). I was driving it with a small function generator, connecting the function gen's output directly to the gate of the MOSFET.

I realized that I probably should have put a resistor between the gate and source to drain off some of the charge (it wasn't switching very well) so I put a 2 KOhm resistor there and then it didn't work at all. So I dropped the resistance to a 200 Ohm resistor and still it wasn't working. Confused and a little frustrated, I got the bright idea to connect a small capacitor (40 nF) between the gate and the source (just to see what would happen). When I turned on the function generator, the MOSFET immediately ran away switching up into the low MHz while I was still pulsing it at around 10 kHz!

I thought this phenomena was pretty interesting and was wondering how common it is to do something like this with a MOSFET? Also, I'm wondering how one would tune the frequency of oscillation of the FET (I used an IRF840). I posted a circuit diagram of my setup below. Any insights welcome.

Thanks,
Jason O

P.S. I should also note that the 300V cap on the 50 turn coil doesn't have to be there. It still works if there is just a piece of wire across the coil. It doesn't seem to work without that though. My thought is that the coil's BEMF is recycled through that loop when the MOSFET is shut off.

 

[Jdo300]

Hello All,

I drew the circuit wrong, here is the corrected one:

 

[es1]

You mean there is an oscillation at the gate of the FET, right? Where exactly were you measuring? What were you measuring with? Did the oscillation decay?

What is the function generator? Actual test equipment or another circuit?

If it is a decent function generator (as in test equipment) then the caps and resistors should of had little to no effect at the frequencies you mentioned (assuming the magnitude of the pulse wasn't >30V (a 150mA output current), and if it was you killed the FET).

I suspect a wiring issue in the setup. Or perhaps an interation between the function generator and whatever you measured the voltage with.

 

[chroot]

You've created a resonant LC tank with the inductor and capacitor in parallel, tied to the drain of the FET. The FET itself has nothing to do with the oscillations.

Your function generator briefly allows the FET to conduct, which then briefly allows current to flow through the inductor, building up a magnetic field within it. When the FET is turned back off, the energy stored in the inductor sloshes back and forth between the inductor and capacitor, and you see an oscillating voltage waveform at the drain of the FET.

If you make your function generator's pulses very far apart in time, you'll see that the oscillations decay over a period of time. The frequency of the oscillation depends on the value of your homemade inductor, which is probably something near 1 microhenry. The resulting resonant frequency is:

$f = \frac{1}{\sqrt{LC}}$

Which is in the megahertz range for your element values:

http://www.google.com/search?hl=en&q=1+/+sqrt(1+microhenry+*+0.22+microfarads)&btnG=Search

You could change the frequency by adding or removing some turns of wire from your inductor, or by using a different value of capacitor. Also, you may not need a discrete capacitor at all, because your large power MOSFET has a large drain capacitance.

In practice, no one really uses LC circuits for oscillators, for a number of reasons. Inductors are expensive, hard to make, and generally large and unwieldy. High-precision capacitors are hard to make, and the resulting circuits are also very sensitive to temperature.

- Warren

 

[Jdo300]

Hi Guys,

Thanks for the tips but I am 100% sure that the oscillations are not due to the LC tank. As a matter a fact, I can replace the capacitor with a piece of wire and it still works. Furthermore, trying different sized capacitors does not modify the frequency which the circuit oscillates at.

To start it, I simply apply a single pulse to the gate of the MOSFET (In my case, I used a kit function generator with a 10Hz square wave input @ 50% duty cycle, 3V). I only touched the lead to the gate of the MOSFET for an instant to get it started. Once going, the circuit continues to oscillate indefinitely on its own (I left it running for 30 minutes with no noticeable decay). To see the waveform, I had my scope leads across the inductor. NOTE, I did not have the ground wires of the probes attached, I used two separate probes and monitored the separate waveforms on either side of the coil. But attaching the ground wires does not affect it either.

The odd thing about this circuit is that I tried swapping different coils, caps, and wires in the circuit with no significant deviation in the frequency of oscillation. I even switched the cap on the gate of the MOSFET. This *seemed* to modify the frequency slightly, but not enough to be certain(+- 1%).

Again, the MOSFET doesn't have to be attached to a frequency source to keep oscillating once the initial pulse has been applied. Another thought I had was that my power supply was somehow oscillating with the FET so I replaced it with a couple of 9V batteries and it still worked just as well.

- Jason O