Understanding Uno's FDN304 MOSFET

Perhaps this is more of a general mosfet question. I'm trying to understand how this and mosfets work in general so I can select one for another project.

If I look at Uno's schematic (http://arduino.cc/en/uploads/Main/arduino-uno-schematic.pdf) it appears this mosfet is ON with 0 V at the gate since the op amp is used as a comparator and it makes sense to use USB power when the external power is < 3.3V. However, when I look at the data sheet for the FDN304 (Intelligent Power and Sensing Technologies | onsemi) the graphs seem to show drain current as 0 and a high drain resistance for 0V at the gate (ie off). I'm a little confused why everything is marked as negative on those graphs so maybe someone who understands those graphs can explain them. My understanding is that most mosfets are enhanced mode and those are off with 0 gate voltage. Depletion mode mosfets are on by default and what I want. The datasheet for FDN304 doesn't say what kind of mosfet it is, but it has the standard enhanced mode mosfet symbol in the datasheet. So my observation is that the Uno circuit seems to say it's on with 0V at gate, but the data sheet graphs/symbol says it's off with 0V at the gate. I'm not 100% sure about either one so I'm not sure where my mistake is.

What I want to do is turn off battery power when USB is connected to my device. To me I either need an enhanced mode mosfet with an inverter to make the switch to the batter off when there is 5V on USB, or I need a depletion mode transistor with no inverter. The problem is depletion mode transistors seem to be fairly low current and I need something around 750mA at a 2V.

Summary: Can anyone explain why the UNO's seemly enhanced mode MOSFET seems to be working as a depletion mode mosfet in that circuit? And do you have any suggestions on how to best turn off battery power when USB is connected to a device?

Thanks!

[Edited to fix typos]

My understanding is the reverse of yours... but I'm not a big mosfet user:

My understanding is that most mosfets are enhanced mode and those are on with 0 gate voltage.

To me... an enhancement device is normally "OFF" when the gate bias voltage is equal to zero and turns on otherwise.

pwillard, that was a typo and I've fixed it in my post.

The FDN304 is a P channel FET. This means to turn it on you have to put the gate to 0v, or make it more negative than the source by the turn on voltage.
It is rather like the FET equivalent of a PNP transistor. This is sometimes best understood with the term "upside down transistor". You use P channel FETs for top switching, that is sourcing current rather than sinking current. That is why the data sheet has all the negative signs.
A data sheet that says what sort of FET it is can be found here:-
http://www.fairchildsemi.com/ds/FD/FDN304P.pdf

Thanks that was helpful!

I remember reading the that the gate on a MOSFET has a large internal resistance so there is no need to put something like a 10K resistor between the voltage source and the gate to limit current. Is that correct? Bonus points if you can tell me what on the datasheet gives me this info.

GlacialWanderer:
Thanks that was helpful!

I remember reading the that the gate on a MOSFET has a large internal resistance so there is no need to put something like a 10K resistor between the voltage source and the gate to limit current. Is that correct? Bonus points if you can tell me what on the datasheet gives me this info.

Not resistance but rather high gate capacitance causing higher current draw when switching gate voltage on or off. Once fully charged or discharged a mosfet gate draws zero current. Any mosfet datasheet will show device capacitance value. One trade-off is that any series gate resistor will slow down the switching transition time from off to on or on to off, due to the RC time constant, causing the device to be in it's 'linear' mode longer and causing more device power dissipation (heat);

Lefty

retrolefty:
One trade-off is that any series gate resistor will slow down the switching transition time from off to on or on to off, due to the RC time constant, causing the device to be in it's 'linear' mode longer and causing more device power dissipation (heat);

Lefty

Most of what you said made sense. I was confused by "trade-off". You didn't explain the positive side of use a series resistor to the gate of a fet. Is is just that it's required to keep peak current down? And is there generally a good sized resistor people tend to use in general cases (100, 1k, 10K)?

Thanks!

Hi

I'm also trying to understand the power system on the Uno, and as it's very closely related to what's being discussed here, I thought I'd tag it on here. Could somebody tell me whether my following ramblings are correct or not please?

USB provides a regulated 5v source, so if this is the only power connected, the gate of the FDN304 is 0v and USBVCC passes through and becomes the 5v rail. The op amp is using a 50:50 voltage divider to compare Vin from the external source to the regulated 3.3v source. Once Vin hits 6.6v (and the op amp sees 3.3v on this pin) the output is switched on, producing 5v at the gate of the FDN304, meaning the USBVCC is "switched off". This allows Vin, having passed through regulation, to become the 5v power source. But why are there two of the MC33239 chips? I just can't get my head round that bit of the circuit :~

Thanks!

Olly

Most of what you said made sense. I was confused by "trade-off". You didn't explain the positive side of use a series resistor to the gate of a fet. Is is just that it's required to keep peak current down? And is there generally a good sized resistor people tend to use in general cases (100, 1k, 10K)?

Yes, a series resistor is to limit the peak current the voltage driving source has to provide to charge or discharge the gate capacitance. As an Arduino output pin has a absolute maximum current rating of 40 ma then a 125 ohm or higher series resistor will prevent the gate current draw from exceeding that safety rating.

The tradeoff is again any series resistor will cause a RC time constant delay the will 'slow down' the transition time between fully on and fully off or fully off to fully on for the mosfet source/drain current path. This may not be a big deal if the mosfet is switching a current load well below it's maximum rated capacity and/or is heatsinked well.

For very high voltage/current mosfet devices many advice using special gate driver ICs that can provide several amps of gate current drive, thus keeping the mosfets heat dissipation to a minimum possible by having the fastest possible switching time.

Lefty

But why are there two of the MC33239 chips?

You need two in the schematic to give two footprints on the PCB. These two footprints a different for different types of regulator, if you look at the schematic you will see one is MC33269ST-5.0T3 and the other is MC33269D-5.0. In practice only one of theses regulators is fitted, it gives the factory a choice of components to use if one gets to be in short supply.

Thanks for answering all my questions. You folks are awesome!

Grumpy_Mike:
You need two in the schematic to give two footprints on the PCB. These two footprints a different for different types of regulator, if you look at the schematic you will see one is MC33269ST-5.0T3 and the other is MC33269D-5.0. In practice only one of theses regulators is fitted, it gives the factory a choice of components to use if one gets to be in short supply.

Thanks Mike, that makes me a lot happier!

I ordered a few of these (http://www.mouser.com/ProductDetail/Fairchild-Semiconductor/FDN304PZ/?qs=sGAEpiMZZMv4eh0jmGe023MliyXefJywkNxtsL75n1Q%3D) to test out and they aren't working as I expected from the discussion above.

My test circuit it very simple. I have 5V attached to the drain. I have a 470K ohm resistor, an LED, and ground attached to the source. I have a 1k resistor attached to the gate and then connect that to either 5V or Gnd expecting it to switch on/off the LED, but it doesn't. The LED is always on.

Does anyone know why this part isn't acting as a switch like I expected it to?

Try this:

The problem with my original circuit was that the source and drain were switched. Thanks!

Grumpy_Mike,
I am really confused by schematic that all the Arduino have for their power supply. You might have not noticed but think all the power supplies have been rubber stamped from one design to the next (UNIO or the MEGA 2560). So a little mistake has been transposed on all the drawings. I have the MEGA 2560 so I can personally speak from that model when I try to describe what I am seeing. (From here on out, I will refer to the Arduino MEGA 2560 as just MEGA- just for clarification.)

First, I think IC1 and IC2 of all the power ICs are swapped with their symbol on the Arduino schematics. I am surprised that no one has picked up on this or caught the swap. I am talking about the MC33269x-x.x which you can see from the data sheet here >> http://www.onsemi.com/pub_link/Collateral/MC33269-D.PDF

Grumpy_Mike, you said there is only one regulator on the board, I think there are both regulators on the board. I think who ever did the drafting for the power supplies were not paying attention to how they were drafting.
Second, if you go looking for the MC33269ST-5.0T3, you will not find it at any part house. Both Mouser and DigiKey do not have that part in 5 Volts. YES, the ‘ST’ part is suppose to be in a SOT-233 package, but if you look at that data sheet the ‘ST’ part only comes in the 3 Volt flavor. That does not change the fact that if this is the correct Mfr. Part # there is a SOT-233 package on your Arduino board (at least it is for the MEGA). Now here comes the part ‘B’ of my confusion- on any of the Arduino schematic they show this part as a part with more than 4 pins. That would make this part a SO-8 part not a SOT-223 part. And the MC33269ST-5.0T3 is clearly a SOT-223 part, which would mean that this is IC2 on all our shematics.

NOW where in the heck you ask is this SOT-233 part on your Arduino board you ask? Well for the MEGA board; if you hold the board like this >> http://www.hobbyengineering.com/pics/i4312-ProductFront-400x223.jpg (thanks hobbyengineering.com for your picture- Sparkfun at this time does not have their pictures updated- they still have the old FTDI chip picture showing) it is north of the power jack soldered to the right side of the silver square (which is being used as a heat sink- just north of the power jack for the regulator). Also, you might have not notice that’s missing from all our schematics is the really cool poly fuse that’s between the SOT-233 package and the USB connector. I really wish Arduino would update their schematics- because the use of the poly fuse is so cool to use for current protection. Yes, I have read about it, just not seen any datasheets on which one they are using. And just like a good bone, the Lab loves to collect datasheets on cool things.

Now you’re asking where in the heck does the Blacklab think he’s sniffing the second MC33269 chip? Well if you look at that JPG picture that I gave you, I will walk you were it is at. Now look at the USB connector, look right of it and you can see the NEW ATmega8U2 that replaces the old FTDI chip for USB communications. South of that ATmega8U2 chip you will see a silver oval which is the crystal for the ATmega8U2 chip. And south of that is an 8 pin SOT-233 chip, which if you get your magnifying jeweler glasses out you will see it connect as you see IC1 in the Arduino schematic. Therefore IC1 is really the MC33269D-5.0 (NOT IC2). If you look at the two silver cans right of the Power jack those are the 74u capacitors. And the block just a little up and to the left of the two capacitors is the D1 M7 diode. Now, what I just noticed C2 is on the 5 Volt side in the MEGA schematics and it’s on the supply side in the Uno schematic { http://arduino.cc/en/uploads/Main/arduino-uno-schematic.pdf }.

I hope I have proven that both power regulators are on board, and it might give inspiration for you to copy their power design on the next board you make. Don’t forget to add that poly fuse to your next design too.

please don't cross-post (post the same content in multiple threads.)