If you are leaving it on, are you certain that the gate voltage is enough to guarantee full conduction? 120 milliohms and 14 amps is only 1.68 watts. Make sure the gate is at 12 volts, but not over 20. Any resistance in the source circuit will raise the source voltage and may cause partial turn off, as the gate has to be at least 10 volts above the source.

We have verified that we are using +12V to turn on the MOSFET. We have turned it up to +14V to make sure, with the same result. The threshold voltage for the MOSFET we are using right now is only ~5V.

I have some pictures of the entire project posted here, so you can get an idea of what the thing looks like. I know it looks like hell right now, but we'll fix it up once we get the circuit working properly.

I went down to verify that we could adjust the output voltage using the control circuit to control the field current on the alternator. This was successful. I was able to push 15A through the field circuit, without trouble. This was all the testing I did though, as we will have to do a lot of rewiring to make the circuit work this way permanently.

I just don't understand why we can push 15A through the field circuit, but cannot push 14A through the motor without trouble. In addition, during this test, I didn't have the MOSFET attached to a heatsink, or have a fan blowing on it. I only tested it for about a minute, but there were no heat issues. The two we fried yesterday were gone within 10 - 15 seconds, while inside our heatsink. We will have to get the clamp on ampmeter again (borrowed it from a professor), and check the current through the motor with the new MOSFET (though last time we checked it, we checked the current without a transistor in the circuit, to get a worst-case-scenario of ~14A continuous).

We are nervous about using the control circuit to control the field current... Will the field collapse cause us problems? When we use the control circuit to turn the motor on and off, the motor continues to spin, which allows the field to remain, taking less of a pulse to get it going again (which is why PWM works). I'm not sure using PWM on the field circuit will work in the same way.

Anyway, please let me know your thoughts about what the latest tests indicate... I am really scratching my head here.

 

You keep saying that you turn up the voltage to turn on the Mosfet.
Isn't the comparator using a triangle-wave at its inverting input and is producing PWM pulses when it compares the triangle's voltage to the voltage from the pot?
Then the circuit is a PWM DC motor speed controller because the very slow generator slowly charges and discharges the 1uF capacitor making a triangle-wave.

Hi Beenthere,
If the high voltage Mosfet is turned on continuously with a 14A load then its dissipation is 14 (squared) x 0.12 ohms= 23.5W but it doesn't have a perfect heatsink and gets hot so its on-resistance rises to maybe 0.18 ohms and it dissipates 35.3W.
Its heatsink must be pretty big to dissipate 35W.
I think its current is higher than only 14A.

We still don't know what is the value of the missing pullup resistor that slowly turns on the Mosfet. Maybe it is too slow and the Mosfet is ramping and the motor is drawing very high current pulses to keep spinning.

 

Seeing the pics I notice no heatsink thermal grease around the Mosfet on the heatsink.
also with so many long wires all over the place the Mosfet has an easy chance to oscillate at a VHF frequency and melt. The series gate resistor should be mounted directly on the gate pin with a very short wire to prevent the oscillation.

 

Seeing the pics I notice no heatsink thermal grease around the Mosfet on the heatsink.
also with so many long wires all over the place the Mosfet has an easy chance to oscillate at a VHF frequency and melt. The series gate resistor should be mounted directly on the gate pin with a very short wire to prevent the oscillation.

The AC input (the feedback) goes through a rectifier, and comes out DC, so its not a triangle wave going into the comparator. Right now, we are just using a function generator for the AC input, so there is no feedback. The voltage out of the rectifier is a constant DC voltage. We are adjusting a pot to turn on and off the motor, so we can see what happens. When we turn the motor on and off, all we are doing is using a thumb screw to turn the pot up and down quickly.

I am not sure what the pull-up resistor I used had for a value. I would check my lab notebook, but I left it down in the lab. If memory serves, it was something fairly high (100kohm range). I will check on that tommorow and get back to you.

There is thermal grease under the MOSFET.

Shortening the lead to the gate is something we could try, but we measured the resistance as .2 ohms, and it didn't vary. It is something we can look at and hopefully eliminate from the list of possible culprits though. I'm not sure why it would drive the motor as we expect, but not the motor when the motor is attached to the alternator. I guess it would be because the current is so much higher, that heat would be an issue, but I would think the motor would run slower because of the oscilating MOSFET.

 

Yes, I^2*R works better. That 100K pullup is way too resistive - something like 120 ohms would be better. That combination of high resistance, long leads, and the gate capacitance just wants to oscillate.

 

The AC input (the feedback) goes through a rectifier, and comes out DC, so its not a triangle wave going into the comparator. Right now, we are just using a function generator for the AC input, so there is no feedback. The voltage out of the rectifier is a constant DC voltage.

You said the frequency of the input is only 3Hz so of course the voltage of the small filter capacitor is a triangle-wave, not smooth DC.

I am not sure what the pull-up resistor I used had for a value. I would check my lab notebook, but I left it down in the lab. If memory serves, it was something fairly high (100kohm range).

Then it will take all day to charge the very high gate capacitance of the gate of a Mosfet and the Mosfet will be linear the entire time and it will melt.

Shortening the lead to the gate is something we could try, but we measured the resistance as .2 ohms, and it didn't vary. It is something we can look at and hopefully eliminate from the list of possible culprits though.

Its length causes it to be inductive, not resistive. The inductance causes it to have a high impedance and is a part of a high frequency oscillator. Then the Mosfet is free to oscillate.

I'm not sure why it would drive the motor as we expect, but not the motor when the motor is attached to the alternator. I guess it would be because the current is so much higher, that heat would be an issue, but I would think the motor would run slower because of the oscilating MOSFET.

An oscillating Mosfet's chip melts long before its outside gets warm.

 

You said the frequency of the input is only 3Hz so of course the voltage of the small filter capacitor is a triangle-wave, not smooth DC.


Then it will take all day to charge the very high gate capacitance of the gate of a Mosfet and the Mosfet will be linear the entire time and it will melt.


Its length causes it to be inductive, not resistive. The inductance causes it to have a high impedance and is a part of a high frequency oscillator. Then the Mosfet is free to oscillate.


An oscillating Mosfet's chip melts long before its outside gets warm.

The 3 Hz was in reference to us physically turning the thumb screw. The frequency out of the function generator is like 20 kHz, so once it is rectified and smoothed by the cap, it should be pretty close to DC.

I know that the resistance wasn't a factor in the length, I meant that, if the output were changing rapidly, because of a flicker on the input of the gate, you would think that the resistance seen from the drain to source would be greater than .2 ohms, or fluctuate, or do something.

The part about the chip melting is good to know... We will keep that in mind. And we will definately try a smaller pull-up resistance tommorow, and try to rule out a semi-on state for the MOSFET. I will report my results tommorow night.

Thanks again for all the help everyone. Your input has been enlightening.

 

I was mistaken, the pull-up resistor we were using was 1k. I tried a 100 ohm resistor, and we couldn't turn the motor off. Tried a 100k, and we couldn't turn the motor on. Tried a 510 ohm, and that worked, but it didn't change anything, we fried another MOSFET.

We checked the gate voltage, and verified it is 12V - again.

We put in a snubber circuit in parallel with the MOSFET, using a power resistor, and a capacitor to try and limit the initial start up current. The values: R=2100 ohms, C=13000 uF. It did limit our initial start up current (cut the spike current by about 20%). We then decided we'd try driving the alternator. We used an initial prime mover (a sheave connected to a battery powered drill), and began the motion before turning the MOSFET on, to try and limit that current even further. But this still wasn't enough, we fried another MOSFET.... our last one actually.

So we have 4 IGBT's that we ordered, and we will try them tommorow. We have some concerns about our 24V power supply being limited in its output, so we are going to hook up 2 car batteries in series and try that. We are also going to put two of the IGBT's in parallel, so that the two will share the load...

Any other ideas for limiting the initial current through the transistors?

 

Well, we figured out what the problem was.

It was a current spike after all. The problem was with our power supply. The power supply that we had apparently wasn't able to put out enough power to start the motor from a dead stop with the alternator attached. The voltage dropped, and the current increased to account for the drop in voltage.

The two car batteries in series work splendidly for us. They can put out exactly what the motor needs. We have two IGBT's in parallel right now, but we don't believe we will need them, and will probably go back down to one next semester. The reason we wait is because we have to have the project working and checked off by the end of next week to receive a grade, but we don't have to present the project until the end of next semester.

Here is a video of the project working: http://www.youtube.com/watch?v=XHtLlj4CYnM

We don't have automatic voltage control yet, but will be attempting that tommorow. The basics for the project are on display though. We turn on the lights, and the voltage dips a little, and the feedback will go into our control circuit, and increase the amount of on-time of the IGBT, which should balance out the voltage.

The only thing we are concerned about is running the IGBTs in the linear region, but we're not certain this will happen. If it does, we will probably go to pulse width modulation, and adjust the voltage on the output manually with a potentiometer.

We are also concerned by the current draw through the motor... we will measure it tommorow, but the motor was getting very hot, under limited use. It shouldn't be as much of a concern when we aren't applying the 24V constantly, but we are also considering getting a little heftier DC motor, or finding a way to cool it.

Thank you everyone for your help, and for giving us things to test. We welcome any constructive criticism or suggestions going forward. I will post the results of our full test tommorow.

 

This is the same when we were saying that the problem was to more current than the rated flowing but you was saying that the current flowing is between the limits.

 

Well, the project is working 100% now. Voltage corrects just as we intended it to, and corrects within 2 seconds. Motor runs smoothly, and everything is fine.

We are considering a fan to cool the motor, because it runs kind of hot, but beyond that, our project is just about complete. We are now going to get inductive and capacitive loads, to show the affects of different power factors, but everything runs smoothly.

Thank you everyone again for all your help. We appreciate it.