Confused about MOSFET symbol

[Mutad0r]

I'm making MOSFET logic gates and wikipedia uses some strange MOSGET. What I find confusing is those little balls at the gate of the top MOSFETS. I think they might mean that they are PNP instead of NPN, but I haven't found any of those symbols used anywhere else.

The picture is huge, so I didn't directly add it. It's in the link.

https://upload.wikimedia.org/wikipedia/commons/thumb/e/e2/CMOS_NAND.svg/2000px-CMOS_NAND.svg.pnghttp://

[Yansi]

Ball on gate symbolizes P channel, without it N-channel.

[Richard Crowley]

The little circle is the convention for showing "inversion".

The circle on the gate indicates that the transistor is "depletion mode" 
You can think of a "normally-on" switch where applying voltage to the gate turns OFF the channel conductance. 

And conversely the symbol without the circle is "enhancement mode".
Equivalent to a "normally-off" switch where applying voltage to the gate turns ON the channel conductance.

Before I read the URL, I was going to say that looks like a NAND gate, but it is self-identifying.   
If both "A" and "B" are high, then the output is pulled to ground.
Else, the output is pulled high.

[rfeecs]

The little circle is the convention for showing "inversion".

The circle on the gate indicates that the transistor is "depletion mode" 
You can think of a "normally-on" switch where applying voltage to the gate turns OFF the channel conductance. 

And conversely the symbol without the circle is "enhancement mode".
Equivalent to a "normally-off" switch where applying voltage to the gate turns ON the channel conductance.

...

This is incorrect.  All the devices in the schematic are enhancement mode.  The circle just means p-channel, no circle is n-channel.

For the p-channel, all the voltage polarities are reversed, so it needs a negative Vgs to turn on.  With no gate voltage, it is turned off.  So it is enhancement mode.

[T3sl4co1l]

Note that IC schematics traditionally use the shorthand of a simplified symbol (the insulated-gate-over-a-channel above, or an erroneous BJT-with-square-arrows one), because they use a metric shitload (that's a technical term) of them.

Of particular note, the substrate connection is almost always strapped to VSS/VDD, so it can be omitted.  There's also little to no distinction between kinds of MOSFETs, as they will most likely all be identical process (N/P and size (width/length) being the only variables).

They also typically leave off the bounding circle, which I have always interpreted as the boundary of the chip; so that in an IC, everything should be drawn within some other bounding shape (usually a rectangle), but in a discrete design, all individual devices should be encircled.

It's also common to see a MOSFET symbol with an antiparallel diode, but this is also almost always erroneous; there is no additional diode in the device, and the body diode action is already spelled out in the traditional insulated-gate-over-three-line-segments-with-arrow shape (which, apparently, all are quick to forget that it's actually telling you about that very diode to begin with).

Tim

[rfeecs]

It's also common to see a MOSFET symbol with an antiparallel diode, but this is also almost always erroneous; there is no additional diode in the device, and the body diode action is already spelled out in the traditional insulated-gate-over-three-line-segments-with-arrow shape (which, apparently, all are quick to forget that it's actually telling you about that very diode to begin with).

Tim

The diode is typically only shown with power MOSFETs, like HEXFETs, where the body diode is deliberately made large and an important part of the device function:

[T3sl4co1l]

I haven't seen a single power FET that isn't rated for full forward current in the body diode.

Possibly RF parts, but that might be as much because, if you inject charge carriers into the slow-ass substrate junction, it freezes up like a PIN diode.  RF transistors are rarely rated in usual terms, anyway.

Even the GaN SOI (semi-on-insulator) parts, that well and truly do not have a channel-substrate junction -- for familiarity's sake, they are still rated as if they have them.  The similar behavior arises from Vgs(off) being fixed, since sufficient reverse bias simply turns it on in inverted mode, as any regular FET ought to (but usually doesn't because body diode Vf << Vgs(th)).

(And yes, the GaN SOI parts should use a different symbol to reflect this -- I believe the manufacturer's recommended symbol is the traditional one, omitting the arrow and substrate connection since they are indeed physically absent.  You could also use the easily drawn insulated-gate-over-a-channel symbol, from IC-land, though an un-dashed "channel" is supposed to mean depletion mode, which they aren't.  But that would be clear enough from the circuit, so it's not a big deal in my opinion.)

Tim

[rfeecs]

Not sure where that came from.   I didn't say anything about forward current in the body diode. 

So you don't like the phrase the body diode is "an important part of the device function"?  OK.  Just pointing out that if you look at most datasheets for power MOSFETs they include the diode symbol.  Probably because if you forget about the body diode you can be in big trouble.

As for RF power MOSFETs, like LDMOS, I agree no diode symbol and no one cares.  These are always used as common-source power amps and the body diode can be ignored.

Have no idea where  you are going with the GaN parts.  I don't think you will find many GaN MOSFETs in production.  They are all MESFET-like (Schottky barrier gate depletion mode HEMTs); or JFET-like (the eGaN enhancement mode parts that EPC makes among others).  There is only one diode junction in the device and that's the gate junction. 

And I agree the schematic symbol is just a short-hand place holder and is "not a big deal".

[T3sl4co1l]

I probably misread that.. on second thought, you were agreeing with me.

That was coming from, 'what if the diode part was small and can't handle the current' -- a concern I've heard before, but never seen reason to (RTFDS, am I right?).

Tim

[leblanc]

There are a number of symbols for MOSFETs. I highly recommend that you get into the habit of using Q5 and Q10 (or Q6 and Q11). The body diode is important to consider. You may forget it exists my omitting it and run into major problems.

As for the symbols in your image, these are used when you just want a simple functional diagram, not a full schematic. Functional diagrams omit a number of things (such as protections). As others have mentioned the circle at the gate means that the switch is active low.

[T3sl4co1l]

Q4 and Q9 should be preferred.  Q5 and Q10 are redundant (as I've been saying), and those particular ones indicate a zener, which isn't true -- although MOSFETs are frequently rated for avalanche (check the datasheet), MOSFET breakdown is more like BJT avalanche than diode breakdown, and has important consequences for switching capability when subject to it.  Put another way: a MOSFET of given capability has considerably less avalanche capacity than a zener diode of equivalent area (which, to be fair, won't be the same die area as the MOSFET, but even one of a modest fraction in area (like 20%) has better ratings).  So when that type of service is required, you are better off using an outboard diode -- in which case, the dioded symbol is triply redundant and misleading!

Tim

[leblanc]

Hmm. Can't really argue with that. I thought the little arrow had a different meaning. Thanks Tim!