My friend and I have been working on our Senior Project for about six months now, and are just about out of ideas. This is our project:

We are making a model Power Transmission system, to show what the power company needs to do to be able to keep a constant 120VAC at your home. To do this, we are using a 24V, 22A, 2200 RPM, DC Motor to belt drive an alternator we removed from a Grand Prix. The alternator normally puts out ~14V to power your car battery, but we removed the diode pack, and are able to get a three phase voltage on the output (roughly 30VAC phase to phase, once a field current has been applied). When the load on the alternator varries (shown by turning on and off lightbulbs), we can adjust the voltage on the DC Motor to keep the voltage at the homes constant. We know this is a rough model, but this is the general idea.

Using this principle, we decided to automate the process. We thought this would be a difficult circuit to make, but once we sat down and designed the circuit, it ended up being far simpiler than we could have imagined. The circuit can be found here:

http://i122.photobucket.com/albums/o271/Farringa/Circuit1.jpg

The Control Circuit works. We are able to turn on and off the motor using the potentiometer, and we are able to adjust the AC voltage, and turn on and off the motor. The trouble comes when we connect the motor to the alternator... it seems to be a current issue. But we have tried the following components in place of the IRF250:

FDH44N50
HGTG10N120BND
NTE2389

They all are able to control the motor, and all of them fail when we begin to drive the alternator with the motor. Here are the things we have tried.

HEAT - At first we thought it was a heat issue. And it was, because when we were driving the motor alone, they did get hot and fail. We have now mounted the transistor on a heatsink, with a 60 CM fan blowing directly on it. We do not believe this is the issue any more.

PULSE CURRENT - This was the next thing we thought might the problem. It seemed that they were failing when we shut off the motor. At low voltages (less than 6 volts), we were able to turn the motor without issues. But once the voltage got too high, and we tried shutting it off, the motor would start to slow down, and then speed back up and stay on, leaving us with no control. We felt this was due to the motor acting as an inductor, and forcing current through the transistor's internal diode, and destroying it. We then put in a flyback diode in parallel with the motor, and this took care of the problem. At this point, we could continuously turn on and off the motor for long periods of time, as well as leave it on and turn it off. But once we started to drive the alternator with the motor, the transistors began failing again.

So, we started measuring the current and pulse current. When driving the laternator with the motor, we were drawing roughly 14A continuous, and roughly 72A pulse. These values are under what our equipment was rated for. We are not sure why they are failing... We figured it was still a pulse current issue, so we ordered the FDH44N50... Today we tried this, and fried two of them, though it appears to be a seperate issue. They smell hot, and the motor runs slowly, before speeding up. Once it speeds up, we know that the MOSFET has been destroyed. Again, it controls the motor just fine, but when we drive the alternator with the motor, it fails...

Trying to limit the pulse current, we placed a 470 uF capacitor in parallel with the 24V source. When did this, we actually had some success. We had have one of the NTE2389 working for roughly 20 minutes, driving the alternator. We were able to turn it on and off repeatedly, and leave it running for 4 - 5 minutes, and still have complete control. At this point, we started to apply a field current to the alternator. Encountering a different problem (now resolved), we stopped everything to ask one of our professors. We decided to show him the working circuit, and the first time we turned it on, we lost control of the transistor again.

So, after this long and drawn out post, I must apologize for boring you all. If you managed to get all the way through it, thank you. It is a difficult problem to explain, and I have a hard time puting down all of the troubleshooting steps. Please post questions, and I will answer them the best that I can. I appreciate any ideas and/or direction at all.

Thanks in advance for the help.


EDIT: See post #21 for pictures of the project.

 

That fact that you can run and control the motor fine without the load of the alternator attached tells me it has to be a current / power dissipation problem with the FET driver. You appear to be trying to operate the FETs in a linear manner, that is acting as a variable resistance. Large power FETs are really designed and spec'd for switching service. While it will be a little more complex driver circuitry, you should really search and study up on PWM. It's much easier on the FETs as they spend most of the time either totally off or totally saturated on, so are much more efficient then when trying to control current in a linear manner. I'm sure this is the heart of the problem you are having.

Good luck, and it sounds like a cool project you are working on.

Lefty

 

Make sure that the voltage on the gate of the MOS with respect to ground is 10 V at least as to fully turn on the MOS. If it is somewhere between 10V then the MOS does not fully conduct and it gets hot. Another think to consider is that when the motor is working with no load on its axis, it draws less current than when you load it (connect the alternator). Maybe this current is more than the current limit of the MOS. What is more, note that if for some reason the voltage difference between the inputs of the op amp is very low then its output will drop somewhere between the power supply voltage. This will cause the MOS to go off from full conduction and overheat it.

 

Thank you both for the quick responses:

Leftyretro: You are right, we will be operating it close to the linear region, when the circuit is working properly. PWM is something else we are looking at, and had originally planned on doing. However, we weren't sure how to make this automated (we would have to manually adjust a potentiometer). But, this is moot, because we arn't able to get to the point where it is switching very quickly yet (only as fast as we can turn a thumb screw... 2 or 3 Hz tops), so I don't think this is the cause. We may still go with PWM, if the linear operation causes heating issues, but we need to get the transistor working before we can go to PWM.

Mik3: We are using 12V to turn the MOS on and off. The unloaded current of the motor is around 2A, the loaded current is around 14A... but the ratings for the transistors are all above 35A. We know about the linear region of the transistors, but they still fail when not operating in this region.

Thanks again for the ideas, please keep them coming.

 

Good luck, and it sounds like a cool project you are working on.

Lefty

It is a good project but one of its drawbacks is that because the alternator's speed is varied to keep the voltage constant the frequency of the voltage will vary. This is not good because reactive devices designed to work better with one frequency won't work well. A better solution that is currently used in power stations is to use a synchronous machine driver at constant speed and vary its field (rotor) current as to keep the output voltage constant.

Anyway, a good link for motor control is this:

http://www.nxp.com/acrobat_download/applicationnotes/APPCHP3.pdf

 

but the ratings for the transistors are all above 35A.

They are above 35A for an ambient temperature of 25 degrees Celcious. As the ambient temperature increases you have to reduce the maximum current value by the derating factor noted in the datasheet.

 

It is a good project but one of its drawbacks is that because the alternator's speed is varied to keep the voltage constant the frequency of the voltage will vary. This is not good because reactive devices designed to work better with one frequency won't work well. A better solution that is currently used in power stations is to use a synchronous machine driver at constant speed and vary its field (rotor) current as to keep the output voltage constant.

Anyway, a good link for motor control is this:

http://www.nxp.com/acrobat_download/applicationnotes/APPCHP3.pdf

We thought about trying to vary the field current, but we decided against it. The reason we decided against it, was we wanted to show the affects of different power factors, and we didn't believe it was possible to do both.

 

They are above 35A for an ambient temperature of 25 degrees Celcious. As the ambient temperature increases you have to reduce the maximum current value by the derating factor noted in the datasheet.

Yup, we know this, but they are failing before they have a chance to heat up (usually within less than 15 seconds). Even the time we had it running for roughly 20 minutes, we had shut it off, come back when it was cooled (we had left the fan running), and turned it on, and it failed immediately.

I am not saying this isn't the problem, but is there any way we could keep the transistor cooler? We have it mounted on a heatsink (taken from a computer CPU), on the inside with thermal paste, with a fan blowing directly on it. The transistors are always warm to the during use, but never hot...

 

Are you sure the diode use to absorb back EMF is good?
Use two of them in parallel in case one fails. Also, they have to be fast switching diodes. Use a low lead inductance capacitor in parallel with the diodes too.

 

Are you sure the diode use to absorb back EMF is good?
Use two of them in parallel in case one fails. Also, they have to be fast switching diodes. Use a low lead inductance capacitor in parallel with the diodes too.

The diode to absorb back EMF is good. It is way oversized, and should not fail (rated for something like 80A). The switching time, I am not sure about, and I will find a part number and post it.

What would the capacitor be used for?


EDIT: Diode is an NTE5991. Rated for 40A. However, it does not say what the recovery time is, but in the datasheet, they use test frequencies of 60 Hz. There is no way we have approached that yet.

 

I feel like I should clarify the failure characteristics of the two trials we had today:

Today we got the FDH44N50 MOSFETs, and wanted to try them. We ordered them because of their high pulse current ratings (176A), because we felt that was the problem.

We connected the first one, and turned our potentiometer to turn on the MOSFET. The motor began to spin, but at a slower speed that we are used to seeing. It was as if the MOSFET had a high resistance, and was limiting the current through the motor. Then, the motor quickly sped up, as if the resistance had been removed. We tried turning the MOSFET back off, and nothing happened. We smelled something hot, and couldn't believe it was the MOSFET, as we had bought them because of the high current rating.

So, we took out the next one, and tested it. Everything was working properly. We removed the belt, and connected the circuit. We turned the motor on and off just fine. We left it on, and had no heating issues.

So, we put the belt back on, and tried driving the alternator again. This time, to try and surpress the surge current, we turned the MOSFET on, and slowly increased the voltage to the 24V. As we were increasing it, the motor was again running at a slower speed than normal. We smelled something hot again. Then the motor sped up, and we lost control of the MOSFET.

These MOSFETs are different in that, they seem to have a high resistance (even though the 'on resistance' we measured was .2 ohms), when our others didn't. We arn't sure why this is. The FDH44N50 seem to be failing under the continuous current load, while the others seemed to fail under the pulse current.

Why would this be?

 

I am a bit sleepy now so i will post tomorrow my thoughts if someone else does not catch me. Make sure that there is no possibility of the voltage on the gate to go above the specified limit even for a short period of time. This destroys the silicon dioxide insulator inside the gate and the MOS is destroyed.

 

I think there is something wrong with the motor and it creates momentary short circuits internally and large currents flow. Another option is that voltage spikes on the gate destroy the MOS. However, I doubt if you drive the MOS into full conduction and that is why it gets destroyed (overheating).

 

I think there is something wrong with the motor and it creates momentary short circuits internally and large currents flow. Another option is that voltage spikes on the gate destroy the MOS. However, I doubt if you drive the MOS into full conduction and that is why it gets destroyed (overheating).

The motor creating short circuits with large current flows could be a possibility. We did monitor the current with a clamp-on ampmeter, and we read a very stable 14A when driving the alternator (without a field current on the alternator). We then turned the motor on and off, and the amp-meter has a memory setting, so we could check the highest point, and it was 72A. Both of these ratings are under what the transistors are rated for, especially the ones we are using now.

As far as voltage spikes on the gate of the MOS, we could try limiting this by placing a 12V zener from the gate to ground, but I don't think we are getting spikes. We are running the control circuit with a seperate power supply, and have monitored the output of the comparator for extended periods of time, and we are confident that it is preforming as designed. As well, the voltage could never get higher than VCC, which is +12V, and the maximum gate voltage on the transistors we are using are all around 20 - 30V max.

Making sure the transistor is in full conduction is something we considered, when we saw how the two we have now failed (see post #11). To test this, we bypassed the control circuit, and connected the gate directly to a voltage source. We then adjusted the voltage source, to turn on and off the motor. Everything worked fine. We measured the drain to source resistance, and measured .2 ohms (the datasheet says the on-resistance is .12 ohms). We applied a higher voltage, and the resistance didn't drop. We then put the transistor back into the control circuit, and we could turn the motor on and off, the same as before. The control circuit applies a Vgs of +12V when on, and the Vgs(th) of the MOSFET is 5V. Yet, again, when we began driving the alternator, the transistor acted as if it had a large resistance, and was limiting the current through the motor. Luckily, we were expecting this, and we were able to turn the power supply off before the transistor was damaged.

 

The motor creating short circuits with large current flows could be a possibility. We did monitor the current with a clamp-on ampmeter, and we read a very stable 14A when driving the alternator (without a field current on the alternator). We then turned the motor on and off, and the amp-meter has a memory setting, so we could check the highest point, and it was 72A. Both of these ratings are under what the transistors are rated for, especially the ones we are using now.

You measure the 14 Amps current when the motor runs at no load. If you load it the current will double or even more thus the MOS can't handle it. Also, when you say you measure the maximum current to be 72 Amps, you measure the rms value with a clamp meter. The current surge may be more than 72 Amps for a short duration but the clamp meter can't respond to such fast spikes and it displays only the rms value.

 

You measure the 14 Amps current when the motor runs at no load. If you load it the current will double or even more thus the MOS can't handle it. Also, when you say you measure the maximum current to be 72 Amps, you measure the rms value with a clamp meter. The current surge may be more than 72 Amps for a short duration but the clamp meter can't respond to such fast spikes and it displays only the rms value.

We measure the 14A when the motor is driving the alternator. When the motor is unloaded, we measure 2A.

The motor is a DC motor, so there is no RMS current, right? I understand that the meter might not be reading what we are actually spiking at, which is why we way oversized. The MOSFETs we are using now are rated for 176A spike...

Also, to try and eliminate current spikes as a problem, we turned the FET all the way on, and then slowly increased the voltage across the motor, and we had the same problem (Motor ran slowly, as if the MOS had a large resistance).

Could it be that the MOSFETs we are using right now (the FDH44N50) have an internal zener? The other transistors we tried had internal diodes, but I don't believe they were zener diodes... If this is a problem, why would they rate the device for 500V and 44A continuous?

 

No, its not the zener diode. You have tried many MOS, I don't think they were all problematic. Something is wrong with the circuit or the motor itself. Replace the motor with something else and check the MOS again.

 

The FDH44N50 Mosfet is high voltage, low current. Of course it gets hot and burns out when it drives your high current motor. It has a high on-resistance.

Your LM311 circuit does not make sense. It needs a pullup resistor at its output for the output to go high. The pullup resistor will turn on the Mosfet very slowly which makes the Mosfet get too hot.

 

The FDH44N50 Mosfet is high voltage, low current. Of course it gets hot and burns out when it drives your high current motor. It has a high on-resistance.

Your LM311 circuit does not make sense. It needs a pullup resistor at its output for the output to go high. The pullup resistor will turn on the Mosfet very slowly which makes the Mosfet get too hot.

The FDH44N50 is rated for 44A, and we are drawing 14A... it has a low on-resistance (.12 ohms), as a matter of fact, it is the lowest of any of the components we have tested so far.

We do have a pull up resistor on the LM311 circuit, but in MultiSim, you don't need one, so it is not depicted in the circuit I posted.

I don't believe it has to do with the switching time. When we leave the MOSFET on, and slowly increase the voltage, why would it get too hot? The MOSFET is fully on, has a resistance of .12 ohms, and then we increase the voltage across the motor, and it reacts the same way.

However, if it is the on-resistance of the MOSFET, do you have a better MOSFET that you would recommend?

I am heading down to the lab now, and we are going to try using the circuit to adjust the field current, and see if that gives us our desired control. I will post the results.

 

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.