MOSFET vs. transistor

hi,
What are the major differences between MOSFETs and transistors? Can one be used in place of the other?
jared

What are the major differences between MOSFETs and transistors? Can one be used in place of the other?

First of all, MOSFETs are transistors, that's what the 'T' stands for. So your question really should be "What are the major differences between field effect transistors and bipolar junction transistors".

At the risk of oversimplifying things one major difference is that field effect transistors are responsive to voltage changes at their input and bipolar junction transistors are responsive to current changes. I'm sure others will pop in soon with more, or look at some of the other nearby threads.

Can one be used in place of the other?

Sometimes.

Don

First of all, MOSFETs are transistors, that's what the 'T' stands for. So your question really should be "What are the major differences between field effect transistors

what separates MOSFETs from other FETs?

There are many different types of "Field Effect" Transistors from 'gateless' measurement devices (A PH probe is one application) and there are J-Fets, junction fets very similar to the device that Baird, Bardeen and Schockley? (sp) invented in the '50's' I think at Bell Labs. They act similar to a PNP transistor except the gate is Never forward biased. The device name was coined from Transfer Resistor...

Doc

thanks for the information.

More information if you care to read it:

http://arduino.cc/forum/index.php/topic,112294.0.html

The practical offshoot of the theory is that MOSFET transistors are generally more efficient and can carry a heavier load.

I was spooked of them until recently, I had been a fan of Darlington NPN pairs like TIP120, or ULN2003 arrays of Darlingtons.

Now I use N-Channel Logic Level MOSFET for most switching applications, particularly my power LED applications, where I am switching amperes at times.

Here's the ones that I have been using:
http://www.ebay.com/itm/160768659447

They are rated for 60v 11A. For twenty cents a pop. Love them. Just make sure to look for "Logic Level" in the name of the component, these are ones that will be (mostly) switched on by 5v, rather than the usual 10-12v at the gate terminal.

focalist:
Just make sure to look for "Logic Level" in the name of the component, these are ones that will be (mostly) switched on by 5v, rather than the usual 10-12v at the gate terminal.

Thank you!

focalist:
Just make sure to look for "Logic Level" in the name of the component, these are ones that will be (mostly) switched on by 5v, rather than the usual 10-12v at the gate terminal.

Are you referring to VGS(th) Gate Threshold Voltage? One transistor I am looking at is IFR9540 which gives that value at -2.0 typ and -4.0 max. That would be logic level, right?

Yes, that's the gate switching threshhold. As I understand it, it's not actually fully 'ON' until somewhere around 10v under a heavy load, so it provides a bit more resistance at 5v and therefore runs somewhat hotter... Though MANY times cooler than a BJT or Darlington. I'm using these straight from the pin with no gate resistance, and also no gate to source resistor, as the microcontroller provides a discharge path for the gate capacitance. I know that there is some merit to a gate resistor to limit the current charging the gate, but I just don't see that small a capacitance needing it. As for the gate to source resistor, I've read it is unnecessary in most microcontroller use because there is enough of a discharge path internally to handle the gate charge dissipation. Even using as a stop-motion photo strobe, I have not seen any turnoff delays caused by gate charge being held too long.. So I'll likely continue to do it the wrong way per usual :slight_smile:

For example, I have three of these, with their gate terminal directly connected to a PWM pin, so I can PWM fade via the FETs. The FET is connected so as to be "switching ground" to string of LEDS, one each Red Green and Blue.. Each draws 2A at 12v (24 watts per color) and the transistors, even without heat sinks, never even get warm.

focalist:
Yes, that's the gate switching threshhold. As I understand it, it's not actually fully 'ON' until somewhere around 10v under a heavy load, so it provides a bit more resistance at 5v and therefore runs somewhat hotter... Though MANY times cooler than a BJT or Darlington. I'm using these straight from the pin with no gate resistance, and also no gate to source resistor, as the microcontroller provides a discharge path for the gate capacitance. I know that there is some merit to a gate resistor to limit the current charging the gate, but I just don't see that small a capacitance needing it. As for the gate to source resistor, I've read it is unnecessary in most microcontroller use because there is enough of a discharge path internally to handle the gate charge dissipation. Even using as a stop-motion photo strobe, I have not seen any turnoff delays caused by gate charge being held too long.. So I'll likely continue to do it the wrong way per usual :slight_smile:

For example, I have three of these, with their gate terminal directly connected to a PWM pin, so I can PWM fade via the FETs. The FET is connected so as to be "switching ground" to string of LEDS, one each Red Green and Blue.. Each draws 2A at 12v (24 watts per color) and the transistors, even without heat sinks, never even get warm.

I recommend a gate resistor of atleast 100 ohms since the gate capacitance could potentially damage the pins.

Hehe I suppose you are right, in theory it's close to a short circuit in terms of current draw, even if it is for only a nanosecond or so.

I wish I had a scope, for looking at something like this. I'd be interested to see what the actual situation is, to see what the actual draw is in reality... But the resistance on the gate is probably a warranted safety measure...lol...

essentially the port current is drawn through the on resistance of a cmos gate and is quite forgiving, the real issue is the total dissipation of the processor and the noise created on the Vcc bus by the charging transient. Adding the resistor serves more to keep Vcc noise down than protecting the output gates. At 40 Ma the pin is guaranteed to be at Vcc or ground or close enough that it isn't sensed by the connected load as not quite 1 or 0. The other consideration is when a large number of outputs are active the total dissipation should be as low as possible because the heating effects will change reference voltages, internal oscillators and some timing as heating causes high/low thresholds to change slightly.

Doc

Docedison:
essentially the port current is drawn through the on resistance of a cmos gate and is quite forgiving, the real issue is the total dissipation of the processor and the noise created on the Vcc bus by the charging transient. Adding the resistor serves more to keep Vcc noise down than protecting the output gates. At 40 Ma the pin is guaranteed to be at Vcc or ground or close enough that it isn't sensed by the connected load as not quite 1 or 0. The other consideration is when a large number of outputs are active the total dissipation should be as low as possible because the heating effects will change reference voltages, internal oscillators and some timing as heating causes high/low thresholds to change slightly.

Doc

Even if the CMOS is forgiving, the small metal wires that connect the die to the pin may burn like fuses on current pulses.

From the datasheet the output transistors (they are FETs, not gates) on-resistance is about 40 ohms with a 5V supply about 60 ohms with a 3V supply. That's about 65mW and 90mW dissipated in the output transistor at the abs max current of 40mA (at 5V and 3V respectively. That's a lot of power for one CMOS FET (it'll be quite a large FET compared to most on the die of course, but it'll still be microscopic), and exceeding the current limit will cause thermal stress, or thermal damage, or in extreme localised melting of the chip and interconnect.

If the manufacturer says the abs max limit is 40mA, stick to that limit if you want reliable operation (unless you are prepared to run your own series of long-term reliability tests to further characterise the chip's response to current pulses!)

If you short an output pin at 5V it will likely dissipate 600mW in the output transistor (10 times abs max), probably raising its temperature at something like a few thousand deg C per second - that rate of heating will cause mechanical stress even if only for a few milliseconds - mechanical stress can lead to failure too.

I think you misunderstood me I stated several reasons why it wasn't a good idea to max out a Proc pin and they are gates. 2 totem pole transistors one, the top one a P-Ch fet with a modified gate drive for biasing on the weak input pull-ups and the second the N-Ch fet to pull down the output as the top is switched off to allow the the lower one to pull down the output just an inverted drive from - to the top P-Ch fet gate.

Doc

You mean the kind of gate made from two complementary FETs that carry the current?

A gate is an abstraction that processes logic values, a transistor carries current. If talking about the current/voltage/resistance/etc, you hop down an abstraction layer to the actual transistors to talk about voltages and currents surely? Actually I wasn't sure if you were talking about the gate terminal on a FET or a logic gate.

Semantics aside you call it what you will I was referring to it as a gate because thats what that classic structure is was called when I worked for a living and very similar to a CD40106 hex inverter W/O the 'biasing' network for the input function. Gates are made of transistors of all kinds, I would point out that this also meets the classic definition in that the gate can be an input, an output and either high, low or open collector...

Doc

Actually I wasn't sure if you were talking about the gate terminal on a FET or a logic gate.

This really shouldn't have been any problem, just look at the 'Subject' of this thread.

Did anyone else notice that the original poster dropped out (with Reply #4) after he got the answer to his question?

Don