Thanks for the analysis, marc. It's now clear that FQP33N10 is the wrong part to use for my device, but out of curiosity, which graph were you looking at (trying to understand what you mean exactly by going linear)?
Also, do you mean I should not use any resistor between Arduino PWM output and the gate?



I looked at the specs of this FET, but accordingly to the power and temp estimations suggested by ronv, it will not survive the task

If you look at the spec sheet, there are graphs with the magnitude of current thru the device given for certain values of gate to source voltage. At 5 volts for your device you will notice that no matter how you change the voltage at the drain, the total current does not change (it is constant). In this situation if you do not change the load resistance and the total current is constant the voltage from source to drain must change because the voltage across the load will not. This increased voltage along with the current (IR drop) must result in a increased power dissipation (P=IE) in the device.

Also you may want to look up the term 'load line'

Another way to look at it is to consider a class A amplifier. It has a no ac input signal output of 1/2 of its power supply at a certain current at the point where the collector (or drain) are connected to the load.

That 1/2 of the power supply at a certain current is dissipated in the transistor in the class A (linear) amplifier, whereas in a switching mode amplifier (as in pulse width modulation) the emitter to collector (source to drain) voltage is very small, dissipating much less power in the device.

As p = i.e. and e = ir the power in terms of resistance and current is ids squared times rds. This action makes linear amplifiers work, but when you just need to either have it on or off you do not want to have a voltage drop inside the device, you want all the voltage to be developed across the load.

I also think that you may want to consider using a TO5 package and a serious heat sink. I do not think TO220 packages are going to cut it.

As far as using a resistor in series with output of the ardunio and the gate, there is no reason to, the important thing is to make sure there is a pulldown resistor from the gate to the drain. The gate to drain resistance is essentially infinite and looks like a 0.001 ufd capacitor. What you wind up with if you use a series resistance from the pwm output is a RC time delay circuit.

Look up the term 'integrator'
Marc

 

Marc,
See what you think about the one in post 34.
How long do you think it will take to switch with 150 ohms in the gate?

 

If you look at the spec sheet, there are graphs with the magnitude of current thru the device given for certain values of gate to source voltage. At 5 volts for your device you will notice that no matter how you change the voltage at the drain, the total current does not change (it is constant). In this situation if you do not change the load resistance and the total current is constant the voltage from source to drain must change because the voltage across the load will not. This increased voltage along with the current (IR drop) must result in a increased power dissipation (P=IE) in the device.

Also you may want to look up the term 'load line'

Another way to look at it is to consider a class A amplifier. It has a no ac input signal output of 1/2 of its power supply at a certain current at the point where the collector (or drain) are connected to the load.

That 1/2 of the power supply at a certain current is dissipated in the transistor in the class A (linear) amplifier, whereas in a switching mode amplifier (as in pulse width modulation) the emitter to collector (source to drain) voltage is very small, dissipating much less power in the device.

As p = i.e. and e = ir the power in terms of resistance and current is ids squared times rds. This action makes linear amplifiers work, but when you just need to either have it on or off you do not want to have a voltage drop inside the device, you want all the voltage to be developed across the load.

I also think that you may want to consider using a TO5 package and a serious heat sink. I do not think TO220 packages are going to cut it.

As far as using a resistor in series with output of the ardunio and the gate, there is no reason to, the important thing is to make sure there is a pulldown resistor from the gate to the drain. The gate to drain resistance is essentially infinite and looks like a 0.001 ufd capacitor. What you wind up with if you use a series resistance from the pwm output is a RC time delay circuit.

Look up the term 'integrator'
Marc

Marc,
See what you think about the one in post 34.
How long do you think it will take to switch with 150 ohms in the gate?

Guys, thanks a lot for the help, I will look into your post, marc, in more depth a bit later, but I have a follow up question regarding the series resistor.
As I understand it, its purpose is to limit the current through the gate and also limit the current out of the Arduino pin.
It is quoted on Arduino web site that it has 40mA per pin, isn't my job to make sure the current is within the defined bounds?

 

marc, you also said that the pulldown should be to the drain, but in my case it's source that's connected to the ground, the drain is connected to the diode and the motor "-".
Don't you mean pulldown from the gate to the source?

 

marc, you also said that the pulldown should be to the drain, but in my case it's source that's connected to the ground, the drain is connected to the diode and the motor "-".
Don't you mean pulldown from the gate to the source?

Yes, sorry. Gate to Source. It used to be from the grid to the cathode a long time ago. The cathode being the source of electrons going to the plate (drain) is probably where the source got its name.

 

TO-5 is smaller than a TO-220, or at least it was in 1973.
Maybe he meant TO-3.

Current does not flow through a gate. The gate is secretly a capacitor and that is the amount of current that needs to flow...enough to charge the gate capacitance.

 

TO-5 is smaller than a TO-220, or at least it was in 1973.
Maybe he meant TO-3.

Current does not flow through a gate. The gate is secretly a capacitor and that is the amount of current that needs to flow...enough to charge the gate capacitance.

So what is the consensus?
Is there a need for the series 150 ohm or not?

 

I think so.

 

So what is the consensus?
Is there a need for the series 150 ohm or not?

I do not think using the 150 ohm resistor would make any difference, and I meant TO3 (the bigger one). Hey I was only off by 2!

I intend to do a breadboard, get some data and share it.

I think my set up will involve an IGFET, a load and some series resistance between the input source and the gate. Any ideas on how I should do this experiment?

Of course there is no current from the gate to the source (well maybe nano amps), which just makes the capacitance more problematic if you do not deal with it.

Marc

 

I do not think using the 150 ohm resistor would make any difference, and I meant TO3 (the bigger one). Hey I was only off by 2!

I intend to do a breadboard, get some data and share it.

I think my set up will involve an IGFET, a load and some series resistance between the input source and the gate. Any ideas on how I should do this experiment?

Of course there is no current from the gate to the source (well maybe nano amps), which just makes the capacitance more problematic if you do not deal with it.

Marc

Sounds great, Marc!
Yesterday I've built the new circuit with three PSMN1R1-30PL and quick test was promising.
I put all three on a good size heat sink and also put the diode on its own small heat sink.
The temp started slowly climbing and reached about 37C in maybe about 15 mins and from the looks of it kept creeping up.
I added a thermistor to the heat sink, didn't have time to wire it in yet, but I'll try to get the temp graph later this week.
I'll do more real life testing today, hope it survives

 

Sounds great, Marc!
Yesterday I've built the new circuit with three PSMN1R1-30PL and quick test was promising.
I put all three on a good size heat sink and also put the diode on its own small heat sink.
The temp started slowly climbing and reached about 37C in maybe about 15 mins and from the looks of it kept creeping up.
I added a thermistor to the heat sink, didn't have time to wire it in yet, but I'll try to get the temp graph later this week.
I'll do more real life testing today, hope it survives

Yeah.

I did a little experiment just now and observed something I expected, but not so dramatically.

I connected a SMP50N06-25 (N channel enhancement mode MOSFET) it is rated for 60V max at 50amps.

I attached a 10 ohm 50 W resistor from +25v (I limited to about 5amps) to the drain. I then applied a 10v square wave at 32Khz (50% duty cycle) with a 150 ohm resistor between my function generator and the gate. I purposely did not heat sink the MOSFET and the case temp settled down to about 90C. When I got rid of the 150 ohm resistor the case temp fell by 20C !

By the way the resistor settled down a little north of 160C.

My function generator was kinda noisy as I was not too careful about wire routing. I suspect if I input a good clean, low distortion more 'beefy' signal and a 'real' pwm signal, the case temp would have been lower. With the 150 ohm resistor, you could clearly see the output waveform rounding, causing some linear operation. My function generator kinda sucks. Probably should fix it.

Marc

Marc

 

Your right of course that it will switch slower with the resistor, but his frequency is not nearly so high so much less time switching.

BTW, the circuit is currently running at around 300Hz

It will not be so pronounced. Or as hot.

The other thing to consider (that we don't know) is what the micro would think about driving the gate directly. It is probably ok because of the low duty cycle but...

But having said all that I think 3 is to far to go using the big FETs because of their high capacitance.. Now it does start to add up. Better with just 1.

 

I understood the signal source was PWM from a Arduino, and the default frequency for this device is 32khz, and the spec sheet for flg540's device states the Vgs for the specified Rds needs to be 10v as well.

No I did not make any change to the frequency, that was coming from the function generator. It was always 32khz. Additionally the Rds for the device I used is specified with a Vgs of 10v. Using a Vgs of 5v would not be advisable if you wish to achieve the Rds specified. Of course as it's Rds gets higher (with less Vgs), power dissipation in the device would increase. For use in a PWM application, the device I used should not be used with a Vgs of only 5v. I suspect this is the main cause of the heating problem in this thread as well.

 

Something like this device should solve the problem.

NCP5181: MOSFET / IGBT Drivers, High Voltage, High and Low Side, Dual Input

Marc

 

I think he is running at 300 Hz. as stated in post #8 and using the FET in post 70. So if he isn't he can just slow it down and all will be good.
I made a little table using a FET like his driven from the micro. At 32Khz nothing looks to good. At 320Hz it's only a few hundred mw in the FET.
No benefit from paralleling them because of the slow turn on time. Without the resistor the peak current in the micro is 120ma. The spec. says 40 ma max. I would guess it can drive more for the short time but I have seen no spec to support that.

 

Sorry I did not stop to read the part about the divisor function.

Still it would be good to know what the setting for this application was.

I could have been 62 or 32Kz.

From the Arduino site:

Base frequencies:
* o The base frequency for pins 3, 9, 10, and 11 is 31250 Hz.
* o The base frequency for pins 5 and 6 is 62500 Hz.

Marc

 

I think he is running at 300 Hz. as stated in post #8 and using the FET in post 70. So if he isn't he can just slow it down and all will be good.
I made a little table using a FET like his driven from the micro. At 32Khz nothing looks to good. At 320Hz it's only a few hundred mw in the FET.
No benefit from paralleling them because of the slow turn on time. Without the resistor the peak current in the micro is 120ma. The spec. says 40 ma max. I would guess it can drive more for the short time but I have seen no spec to support that.

I do not understand how the input to a IGJET could cause a current spike of 120ma. I could see a very short spike caused by the input capacitance, but I would think it would be very short, on the order of pico seconds.

I have learned several things form this discussion.

1. When using PWM, maintain a gate to source input voltage of 10V pk to pk for most parts to keep the Rds within specified limits. This would mean using a driver chip such as a NCP5181, especially when the max current spec is close to the operating point of the circuit. (If 5v causes a current limit of 20Amps and your circuit needs 22Amps, the device will get hot)
2. A PWM of 32Kz should probably not be used.
3. Be aware of the gate voltages that will allow constant current (linear) operation and avoid them if you are using PWM.
4. They probably do not make a lot of TO3 devices any more (so 70's!).
5. I am not even sure they even had NPN power transistors in 1972.

marc

 

It is not to long a spike, but more than pico seconds. See simulation 1
32Khz might be ok if he used a driver. Then instead of 3 usec to turn on and off it would turn on in about 600 ns. Sim 2.

5. I am not even sure they even had NPN power transistors in 1972.

Ouch! Your aging me. :=)

 

Great discussion, guys! Appreciate you taking the time and doing all these experiments. Keep it going!
This has been sort of an emergency redesign in the middle of another crisis and I haven't been able to get all of the measurements, but I'm getting there.
Today the circuit was put to the test again, this time I used the thermistor on the heat sink to log the temps. The temp varied between 40C-50C, with probably an average of 43C.
Another problem I discovered was that my reference thermistor voltage that I use to compute PWM values was not in the expected range (was indicating higher temperatures then they actually were) and as the result PWM was pegged to 100% most of the time.
So before I change anything in the MOSFET circuit I want to debug that problem first to make sure that my control signal is what I want it to be.

 

Maybe you can confirm the PWM frequency is still 300 Hz. ?