Mosfet Pull Up Resistor

Thread Starter

whatisgoingon

Joined Jan 30, 2013
9
Hello,

I am using a mosfet. I wanted to PWM the gate to switch the mosfet on and off. I wanted to find the appropriate pull up resistor to use at the gate. According to the data sheet

turn on delay time: 18 ns
turn off delay time: 50 ns

The input capacitance Ciss = 1620 pf.

How do I determine the right resistor (value of resistance) to use if I want to switch at a high frequency
(since it takes about 50 + 18 = 68 ns to turn off and on, frequency of 1/100 ns. should be okay?).
 

#12

Joined Nov 30, 2010
18,224
Lots to question here. The main purpose of PWM is efficiency by keeping the switch fully on or fully off most of the time. Your approach seems to be, "Run it so fast that it's always in transition". That seems like the wrong way to do it.

You need to know the voltage change to calculate the resistor, but using a MOSFET driver chip would be a good thing to investigate. Meanwhile, look up my blog about RC time constants. Something about: time = -RC Ln Vo/dV
http://forum.allaboutcircuits.com/blog.php?b=485
 

Attachments

Thread Starter

whatisgoingon

Joined Jan 30, 2013
9
I guess driver is useful but I also needed to isolate my microcontroller pwm signal from the switched side.

I was thinking: PWM to optoisolator to Mosfet.

But the output current max from my microcontroller is only 2 mA. From simulations I found that to make it work, I have to put a very high value resistor as a pull up on the mosfet gate. I have been told that a high pull up will impact my ability to PWM with high frequency (due to higher RC constant). Any tips??
 

Jaguarjoe

Joined Apr 7, 2010
767
You ought to be able to find or make a push-pull gate driver.
Current into the gate will determine how fast you switch because of the gate charge that must be overcome. Gate Q should be in the data sheet.
Use Q = IxT to determine speed or current, or both.
If you are not fully off or fully on, the MOSFET will suffer greatly as it spends time in between the two.
 

#12

Joined Nov 30, 2010
18,224
How about PWM to opto to driver to MOSFET? ps, I just checked that driver chip I posted, and I think there are better driver chips.

Can you do the math to find a frequency that will work for PWM and still be less than 10 megahertz? Just as ballpark figures, if the MOSFET can't go any faster than 68ns (14.7 MHz) you shouldn't try to operate it at more than one tenth of that speed. That way it spends 90% of its time in saturation or off.

You could tell us what you're trying to do and we won't have to guess what the goal is. There are some serious math guys here.
 

Thread Starter

whatisgoingon

Joined Jan 30, 2013
9
I have a DC voltage at around 800 V. I have a power resistor connected to the drain of the mosfet (STW25N95K3). I will vary the resistance across the 800 V voltage by switching the mosfet on and off.

I am not sure what the minimum frequency should be to switch on and off.
If I keep the mosfet on (100% duty), the resistance will be whatever the resistance is of the power resistor. If the mosfet is off (0% duty), it will be high impedance. Anything inbetween should change the resistance. But not sure what is a good frequency.

My microcontroller with an output of 5 V and max 2 mA will generate the PWM signal.

So my goal is to find an optocoupler that I can use to turn the mosfet on and off. Or a proper driver circuit that is somehow isolated as well.
 

Ron H

Joined Apr 14, 2005
7,063
Why do you want to vary the load on your 800V supply? The supply will almost certainly respond differently to PWM than it will to a variable resistance.
 

#12

Joined Nov 30, 2010
18,224
That would be true as long as there is no capacitance in the 800V supply and the "real" load is not also a PWM load.
 

#12

Joined Nov 30, 2010
18,224
This is more than I've been able to drag out of the OP in 6 hours. It represents the idea that a 1 KHz PWM load of 800 ua would cause dips in the 800 volt supply of less than 1%.
 

Attachments

GopherT

Joined Nov 23, 2012
8,009
The big benefit of high speed switching is to allow smaller (cheaper) inductors made from cheaper materials (have you notices how light a AXT power supply has become?).

If you are switching a motor or doing almost anything besides an AXT power supply, you are better off at lower PWM frequencies. Also, check the DATASHEET of your Microcontroller. The higher Pwm frequency you use, the fewer bits of PWM resolution you get. At the lowest frequencies, a midrange PIC will allow 10 bit (1024) different duty cycles. At the highest PWM frequency, you get only 5 bits (that is from memory - I didn't bother looking it up just now). In any case, you are down to only 32 different duty cycles.

Also, if you are controlling a motor, the inductance and DC resistance in the motor will filter out all high frequencies anyhow. Just stay above most people's audible range (10 k usually works because a motor's ability to reproduce anything that fast is poor and barely audible).

Cheers
 

Ron H

Joined Apr 14, 2005
7,063
This is more than I've been able to drag out of the OP in 6 hours. It represents the idea that a 1 KHz PWM load of 800 ua would cause dips in the 800 volt supply of less than 1%.
Where did you get that schematic?
 

Thread Starter

whatisgoingon

Joined Jan 30, 2013
9
This is more than I've been able to drag out of the OP in 6 hours. It represents the idea that a 1 KHz PWM load of 800 ua would cause dips in the 800 volt supply of less than 1%.
That is what I am doing. THe load side is just a load bank that I can vary. But the problem like I said I am having is turning the mosfet on and off. I was orignally using an optocoupler connected to the microcontroller.

I use a 2500 resistor between microcotnroller and optocoupler (4N26) to limit the 5 V to less than 2ma (through simulations I get about 1.5 mA). THe pull up resistor of 8 kOhm is what I used to get this working. BUT since the 8 kohm is so high I have been told the RC time constant will be large. Therefore I cannot switch the mosfet on and off fast enough. This is my problem.

Through simulations I found If I decrease the 8 kohm to something low, when I try turning the mosfet on (by giving 5 V on the microcontroller side) i get 11.8 V at the gate (it doesn't go down close to 0).
 

Attachments

Last edited:

#12

Joined Nov 30, 2010
18,224
You can use an optoisolator to drive this chip, then use this chip to drive your MOSFET.
and I don't want to hear any more about, "I have been told". We have the math. We can do the math. It works a lot better than, "somebody told me".
 

Attachments

Last edited:

Thread Starter

whatisgoingon

Joined Jan 30, 2013
9
From the datasheet I coudlnt find what the minimum current required is for the input of the drive. After my optocoupler I can only get around 1.5 mA output. Do you think that will be enough for the drive to work?

Thanks for your help by the way.
 

#12

Joined Nov 30, 2010
18,224
I couldn't find the input current either, but the driver chip datasheet lists an input maximum of 2 ma. That tells me it is a voltage driven switch and it requires a lot less than 2 ma. The switching voltage is about 1.8 to 2 volts.

Fig. 7 shows the driver chip as if it is a MOSFET so I might have steered you wrong and only handed you the same problem in a different package. This chip also does not list input capacitance so I can't tell. I wish I could find a CMOS, Schmidt trigger input, driver chip.

Can anybody else confirm?
 
Last edited:

#12

Joined Nov 30, 2010
18,224
This chip has 400K pull up resistors on its inputs, schmidt triggers, and internal amplifier circuits. It might be what I'm (we're) looking for.
 

Attachments

Ron H

Joined Apr 14, 2005
7,063
This looks like the same thing in an 8 pin dip package.
http://www.mouser.com/Search/Refine.aspx?N=6785017

Repeating myself...HELP. I'm foundering here.
Well, I'm not totally up to speed on this, but I have some comments:
1. 4N26 is blazingly slow.
2. 4N26 guaranteed current transfer ratio is only 20%. 2mA of input drive will yield, worst case, only 400uA of collector current, and that 20% spec is at Iin=10mA, which is the sweet spot (see Fig. 2 in the Vishay datasheet).
3. The driver you linked to looks good, but I couldn't find any input capacitance or input current requirements, which concerns me when I have such a wimpy predriver (4N26).

Has the need for the optocoupler been discussed?

EDIT: I looked over the thread again. I guess if your PWM is a 1kHz, and your duty cycles are not too extreme, the 4N26 might work, but I think you will need more than 2mA of input drive. Why is this number so low?
 
Top