Man I hate MOSFETS...

[sonnytiger]

But their usefulness is great. I only hate them cause I just don't get how to use them. OK. Here we go. I am trying to use this MOSFET:
http://www.vishay.com/docs/91015/sihf510.pdf
to drive a small brushless motor (basically a small number of poles stepper motor) in the same way it was being driven with the original circuitry. The original circuitry simply has the common of the star wound motor connected to +12V and three MOSFETS simply connected each winding to ground. This is what I am trying to do. I ran a simple simulation in LTSpice, using pulsed voltage sources to drive the gate of each MOSFET (there happened to be a 510 model in LTSpice). In real life I planned on replacing these pulsed voltage sources with 2n3904 transistors connecting the gate of the MOSFET to +12V when +5 was applied via a 1k resistor to the base by an arduino. I connected +12V power to the MOSFETS without the 2n3904s in the circuit and a large current flew, increasing quickly. I then figured there might be some capacitive coupling to the gates which would charge them up and cause the MOSFETS to conduct. Hey ho that's what was happening and now no current flows. With that solved it seems to me i need to have two transistors for each MOSFET, one to connect to +12V and one to connect to ground. Is this true? I will post the LTSpice schematic and you guys tell me what else I should do.

[chrisbrown]

The gate of a MOSFET is basically a capacitor, with the other end connected to the source. If you charge it up and remove the charging source, it's going to stay charged until you discharge it. Sometimes a bleeder resistor is good enough but if you need fast switching then yes, you need a low-side transistor. There are plenty of gate driver ICs that can do exactly what you want.

[free_electron]

why do you want to drive the gate with 12 volt ?
your arduino outputs switch between 0 and 5 volt. that is more than enough to drive a modern mosfet directly.
you can go directly from the arduino ouput ( which is push pull ) to the mos gate. no need for intermediate crap.

[Jonny]

why do you want to drive the gate with 12 volt ?
your arduino outputs switch between 0 and 5 volt. that is more than enough to drive a modern mosfet directly.
you can go directly from the arduino ouput ( which is push pull ) to the mos gate. no need for intermediate crap.

Would I be right looking at this datasheet that this model only starts to turn on at 2-4V and is fully on around 15-20? Logic 5V would not be enough. Would need a logic level FET for direct drive. It would be worth the hassle to buy some.

[sonnytiger]

Yeah I determined 5V would not be enough based on the datasheet.

[Paul Price]

You could use a 74LS06 or ULN2004 to drive the MOSFETs and interface to your MCU.

Otherwise you could use two 2N3904's per MOSFET and you will need  a 2k base resistor to the MCU output. The first 2N3904 is configured as a inverter and connects directly to the base of the second 2N3904. The first 2N3904 would have a 10K pullup resistor on its collector and the collector of the second 2N3904 connects directly to the gate of the MOSFET and has a 1K resistor connecting to the stepper motor power supply. Both 2N2904 emitters are grounded.

[free_electron]

why do you want to drive the gate with 12 volt ?
your arduino outputs switch between 0 and 5 volt. that is more than enough to drive a modern mosfet directly.
you can go directly from the arduino ouput ( which is push pull ) to the mos gate. no need for intermediate crap.

Would I be right looking at this datasheet that this model only starts to turn on at 2-4V and is fully on around 15-20? Logic 5V would not be enough. Would need a logic level FET for direct drive. It would be worth the hassle to buy some.

I said MODERN mosfet. Not that old relic you got there. 

[sleemanj]

Absent a mosfet with a low RDSOn at logic level (there are some through hole ones available but more expensive, there are more in SMD packages), then the simplest way to drive it is a classic totem pole driver.


(Not my diagram, just a random one from google image search)

If you really want to get into some gory details:
http://www.ti.com/lit/ml/slup169/slup169.pdf

[Zero999]

That circuit won't work with a 5V input because it has no voltage gain.

Here's an idea but it's inverting so you need to take that into account.

[Bertho]

Absent a mosfet with a low RDSOn at logic level (there are some through hole ones available but more expensive, there are more in SMD packages), then the simplest way to drive it is a classic totem pole driver.

The problem with a totempole driver is that it shorts in the transitions. The NPN/PNP have different ton/toff parameters where the NPN is already on and the PNP is not off yet, creating a huge surge-current in the totempole.

If you really need to switch fast, then you should use a standard low-side mosfet driver, which is optimized for the purpose.

[bingo600]

That circuit won't work with a 5V input because it has no voltage gain.

Here's an idea but it's inverting so you need to take that into account.

Thnx

/Bingo

[Stonent]

Your hatred of MOSFTs reminds me of the 1950s/1960s era film where the guys says he hates springs and a little sprite makes them all go away and he has to live a life without springs.

[Sigmoid]

Absent a mosfet with a low RDSOn at logic level (there are some through hole ones available but more expensive, there are more in SMD packages), then the simplest way to drive it is a classic totem pole driver.

The problem with a totempole driver is that it shorts in the transitions. The NPN/PNP have different ton/toff parameters where the NPN is already on and the PNP is not off yet, creating a huge surge-current in the totempole.

If you really need to switch fast, then you should use a standard low-side mosfet driver, which is optimized for the purpose.

Please hear me out, I'm quite new to design and would like to validate an idea.

So what if I place resistors on both sides?
It will slow switching down, but my guess is, not that much.
If I add a total of 500 Ohms to the totem pole (250 on both ends), at 12 Volts the wattage of the "short circuit" state is down to less than 1/3 Watts.
The input capacitance of this MOSFET in question is 180 pF... That gives us a cut-off frequency of 3.5 MHz...

Seems cheap and simple enough, and I'm sure 3.5 MHz is more than enough. Or is there a fault in my thinking?

[Andy Watson]

The input capacitance of this MOSFET in question is 180 pF... That gives us a cut-off frequency of 3.5 MHz...

Seems cheap and simple enough, and I'm sure 3.5 MHz is more than enough. Or is there a fault in my thinking?

Miller feedback/capacitance - don't underestimate it!

[sleemanj]

That circuit won't work with a 5V input because it has no voltage gain.

Oops my bad for nicking a google image search result for totem pole driver and not looking at it properly

[Simon]

just put another transistor between the "driver" and the arduino with a pullup resistor. that 2 transistor solution works quite well. you can also take out the base resistors

[sonnytiger]

Unfortunately I can't find my arduino so i cant try any of these circuits. But I am not switching very fast at most maybe 70Hz. Love MST3K by the way

[Stonent]

Unfortunately I can't find my arduino so i cant try any of these circuits. But I am not switching very fast at most maybe 70Hz. Love MST3K by the way

Well as you complimented a show from my country, I'll compliment a show from yours.  I'm a fan of the The Red Green show. 

[sonnytiger]

As it happens, I am as well! Thanks for all the info guys, I think I understand the driving characteristics of MOSFETS a little better now

[Stevve]

Hating something because you don't understand it is not exactly the best way to go through life. Embrace the learning experience.

Sooo.... take a MOSFET to dinner?

[WarSim]

Unfortunately I can't find my arduino so i cant try any of these circuits. But I am not switching very fast at most maybe 70Hz. Love MST3K by the way

The current point is about assuring dead time.  Dead time is the time neither FET is driven to prevent shoot through (massive current).  It take time for the FETs to saturate and cut-off.  These times are different, also the times on the data sheet are the time it is capable of if the gates are driven hard as in their test setup.  Many ccts do not use dedicated driver chips which insure dead time.  In a conventional cct other means need to be used to push the charge into the gate and sink it out again fast enough to meet you timing.  This is why many uC also have PWM drivers to insure the dead time.  No matter what the speed if you are switching with a square wave the time you have to switch Is short no matter how often you switch. 
If a FET is turned on and off slowly it has to dissipate allot of heat and will likely hit its Pd limit before it's current rating. 
Yes there TTL drivable FETs but they in general have slower switching times therefor have a lower channel area to current ratio. 

[hpux735]

That circuit won't work with a 5V input because it has no voltage gain.

Here's an idea but it's inverting so you need to take that into account.

This is a topology that I've been experimenting with.  It works, but I worry about the abuse that I'm doling out on the logic pin.
(the image is an attachment)

My thinking is that the gate will float to high unless the uC pin is high.  In that case, the transistor is a current amplifier that will draw the gate voltage down near zero.

[coreybus]

I have never and probably never will understand them

[azi]

But their usefulness is great. I only hate them cause I just don't get how to use them.

I think you are at a basic beginner levels with FETs, and I suggest you invest some time in basic education.

A few years ago, I found the MIT OpenCourseware introductory Electronic Engineering course, and it contains a decent explanation of MOSFETs. Try it. Wikipedia articles, too and anything else you can find online.

For me, the best thing was to understand the construction of FETs; that is, how they are made, and what the parts do. It's just source, gate and drain, and the source and drain are similar, so it's really not hard. After that, read the datasheets carefully and try to understand the basic curves for transfer characteristics.

Along with that, get some small signal devices (really, 2N7000), and play around with "safe" currents that won't melt your breadboard. After a short time, you will probably have a good intuitive idea of what's going on in the device and how to use it. Then you can apply the understanding directly to bigger MOSFETs.  Have fun with it.

[Icchan]

Hi. My first post since finally I figured I might help just a little at least...

It's not very short, but bear with me ok? 

MOSFET's are really a can of worms especially when trying to understand them in the beginning. I'm going to be really crude in my explanation since I wish to convey the general idea of why they act the way they act. Please, correct me if I'm using totally wrong explanation to convey the idea and I'll edit my post.

The reason why MOSFET's have the characteristics of capacitors can be understood quite simply: there's no galvanic connection between the gate and rest of the semiconductor layer and they create a capacitor as a result of that. More voltage the MOSFET can handle, the wider the gap between the gate and source/drain parts and bigger the distance between the junctions of the Source <-> Drain. And more current the MOSFET can handle, wider the gate/source interface is. Thus as a rule of thumb: bigger the MOSFET -> bigger the "power" needed to drive it fast.

The why these dimensions matter is because they create stray capacitances and inductances that have to be overcome before the system reacts to a change at the Gate. But what's more, the gate itself has to be proper size and charged to a proper voltage to get the resistance between Source and Drain low enough when switching.

The important thing in MOSFET's datasheet is "Total Gate Charge". That is the sum of all the known factors that affect the amount of energy needed to fully switch on the MOSFET.

As an example:
http://www.pololu.com/file/0J56/irfr3707zpbf.pdf

See the page 2. Total Gate Charge has been given in nano coulombs. a Coulomb is defined as: 1C = 1A * 1s or 1 coulomb = 1 ampere of current in 1 second.

This means in the case of a MOSFET, that if one has a "Total Gate Charge" of 1 coulomb, the current required to charge it in 1 second equals to a 1 amps. Since coulombs are a very large amount of charge, mostly we're using nano or even pico coulombs.

If you wish to charge the Gate faster, let's say in half a second, we can do the arithmetic and we'll arrive to the conclusion that it takes double the current in half the time to charge. The faster you want to switch the MOSFET on, more current you need in one second. After the gate is charged however, the current flow to the gate stops (there's a tiny bit of leakage current, but it's mostly negligible.). When you wish to turn the MOSFET off again, you'll need to discharge the gate and the same rule applies again, faster you discharge it, faster the MOSFET switches to off state.

Thus there's a real need to use external driver circuits when trying to handle large amounts of current and voltage over the MOSFET. Puny micro-controller I/O pins can't handle the current spikes or even give enough current even if they could handle it to drive big MOSFET's fast enough to handle large currents.

Since when the MOSFET is changing it's state, there's a time when the resistance is not infinite or near zero and this is the time the MOSFET is turning all that current going trough it in to a heat. The faster you go over this period (faster you drive the device) less heat you're dissipating in the MOSFET itself. I'm not going into the nitty gritty details of calculating this right now but it's not very hard if need to be.

Whoah that was a long post... 

I hope this helps a little and you gain more valuable insight on the why's and how's of the MOSFET's... it wasn't' very intuitive for me either when trying to piece it together but it's valuable information when trying to drive larger currents. 

I might edit this post or create a totally different tutorial type post with all the facts and figures and calculations later date, but right now I must sleep and please, ask if you don't' understand something or tell me if I got something totally wrong and i'll fix it.