IC diagrams

[SavannahLion]

I'm not entirely sure where I need to ask this. I figure I'll check here and see if anyone knows. If not, I'll check elsewhere. Mostly it's because I'm too tired to pull up my login information for the other fourms.

I see symbols like this from time to time in IC schematics but never PCB schematics. I know this specific diagram is for a NAND gate but other that that, I'm mystified as to what the symbols represent. Based on the function of the NAND are these intended to be an IC version of PNP and NPN transistors?

[Ed_McCarron]

FET?  Like a transistor, but the trigger is voltage rather than current?

 P type
 N type

Monmotha will be along shortly to write a 3 page dissertation on the topic.

[lilshawn]

a MOSFET or just FET is very similar to a transistor in that it acts like a switch, but that is where the similarities end.

a FET switches on very fast and turns off very fast. therefore can be switched fast. there is almost no period of partial conductance so it is very efficient. it's often used in switching power supplies to rapidly turn the voltage off and on.

a transistor (depending on model) has a range of voltage where the transistor can be just partially on.. where it's not fully on, but it isn't quite off. a FET is either 100% (or as nearly possible) on or off.

[WhereEaglesDare]

a MOSFET or just FET is very similar to a transistor in that it acts like a switch, but that is where the similarities end.

a FET switches on very fast and turns off very fast. therefore can be switched fast. there is almost no period of partial conductance so it is very efficient. it's often used in switching power supplies to rapidly turn the voltage off and on.

a transistor (depending on model) has a range of voltage where the transistor can be just partially on.. where it's not fully on, but it isn't quite off. a FET is either 100% (or as nearly possible) on or off.

...this is why they are called SemiConductors.  This is an old book from the 50s called the ABCs or Transistors...  It is a really dumbed down barney version of what transistors are and how to use them.  Anything from the 50s early 60s is really good for beginners cuz everyone was beginners then.

[lilshawn]

when i took electronics class in high school, the books for the curriculum where from the 50's and 60's, all the way from brush cut military haircuts to cheezy porno looking mustache dudes and all...... it was told to us that NOTHING had changed in the last 30 years (90's at the time) and it really hasn't 60 years later.

I had to laugh about the cautionary tale at the beginning of one of the beginner books... "DON'T BE A CLOWN - sometimes there will be a class clown who will charge up a capacitor and toss it to you only for you to receive an electric shock. DON'T BE THAT CLOWN!" **spot color added JUST like in the original book**

guess what the first thing we did was had the book not even mentioned it in the first place, there would have been a chance that no one would have even thought to do it.

[MonMotha]

Hi!

Yes, this is a MOSFET.  In this CMOS digital logic application, they are indeed being used as switches.  For an N channel FET, if the gate voltage is substantially higher than the source voltage, the FET will be "on" and allow current to flow from drain to source, and if the gate voltage is close to or below the source voltage, the FET will be off and no current can flow.  P channel FETs work the opposite way (current flows freely from source to drain when the gate voltage is substantially lower than the source voltage).

You can see from your circuit then that if either A or B is near ground (low), one of the top two P channel FETs is on, which will allow current to flow from VCC into the output, setting the output to a logic high.  If A and B are both near VCC (high), then both N channel FETs are on, and current flows from the output back into ground, pulling the output down to a logic low.  This is indeed a NAND gate.

CMOS is so-name "complementary MOS logic" because it uses both N channel and P channel FETs.  You can also make logic using only one channel by substituting a resistor for the other side.  This resistor often lives outside the IC, and you end up with an "open drain" or "open source" output.  Other logic topologies exist that use the more commonly known bipolar junction transistors (BJTs), instead.  The most common is "TTL" such as the 74xx (no inner letters) series logic.

MOSFETs are actually bidirectional.  For an N-channel MOSFET, as long as the gate voltage is sufficiently more positive than the source voltage, current can flow either way.  However, conventional MOSFET construction techniques end up with a so-called "body drain diode" with the anode connected to the source (since the body is connected to the source) and the cathode connected to drain.  This diode allows current to flow from the source to the drain whether or not the MOSFET is "on", albeit with a full diode voltage drop (typically ~0.7V).  Sometimes this can get in the way, but it can also serve as a built-in freewheeling diode in inductive load switching applications, and you can buy MOSFETs with this built in diode highly characterized for this and similar purposes.


However a MOSFET does have a very usable "partial conduction" range: it's called the saturation region, unlike a "saturated"  bipolar transistor which is "saturated" when VCE is as low as it can get, and the bipolar transistor is acting as a switch.  Confusing, I know.  The BJT equivalent of the MOSFET's saturation region of operation is the "active" or "forward active" region.

Many car stereos (used to) have MOSFET linear output stages, though "class D" amplifiers, which are similar in many respects to a switch mode power supply, are very common now due to their higher efficiency and minimal cost in terms of sound quality.

The "switch" region on a MOSFET is commonly called the "ohmic" region because it effectively acts like a low value resistor (milliohms to dozens of ohms, typically).  A BJT being used as a switch acts like a constant voltage drop (typically about 0.2V).

You can often interchange a MOSFET and a BJT with just a few circuit tweaks.  MOSFETs are typically used in switch applications due to their effectively zero static gate current requirements and lower on state losses at low-medium currents.

A hybrid of the two also exists: the IGBT.  These have a voltage controlled gate like a MOSFET, but the on-state characteristics more closely match a bipolar transistor.  These are often used in high power applications.  There also exist other kinds of FETs, though they see limited action these days.  JFETs are still around, mostly as inputs on op-amps.


And nope, basic electronics haven't changed a ton in 40 or so years (the FET is about that old), but how they are used definitely has.  Miniaturization has a lot to do with that.  Your quad core processors have some 800 million plus transistors on them.  That was unthinkable in the 70s.

[Ed_McCarron]

Told you so.   

[Hoopz]

Told you so.   

You know, I read all of his stuff because it's so detailed, in depth and obviously factual.  I just wish I had some idea of what language it is because it makes no sense to me at all.  It's not even over my head as much as in another world. 

The dude knows his stuff.   

[RoyalScam]

I cut my teeth, so to speak, in electronics in the late seventies early eighties.  As such my strength was in 74xx TTL and 4000 series CMOS logic. The two books other than semiconductor reference guides I used mostly were Don Lancaster's TTL and CMOS cookbooks.  If you happened to have a design that needed an extra gate of some sort, or even a gate with an unusual input, he used a method he called IIRC Mickey Mouse Logic.  This usually involved creating your own gates using discreet transistors. Although I've known this and even built discreet component gates just on breadboard, I've never really used the method. But this guy wins the prize;

http://waitingforfriday.com/index.php/4-Bit_Computer
 
Although I know how it works, and how it's built, I've never even contemplated doing this. Looks like it was a fun project.  BTW It's a good read if you want to learn the fundamentals of logic and adders.

Regards,
Scam

[SavannahLion]

Holy crap in a hand basket monmotha. I'm still trying to digest what you said. However something escapes me. Why aren't discrete FETs used more often as opposed to your typical PNP/NPN (These are the BJTs right?) transistors?

Royalscam that site rocks! I love how he built the 4-bit computer. The schematics alone are worth the reaD. I was racking my head on how to test my latest arcade add-on without ordering more ICs. Sadly my local supplier is out of business for good. Now I have to scavange for parts or order throug Mouser or Digikey. Kind of sucks.

[MonMotha]

Holy crap in a hand basket monmotha. I'm still trying to digest what you said. However something escapes me. Why aren't discrete FETs used more often as opposed to your typical PNP/NPN (These are the BJTs right?) transistors?

Yes, a BJT = "Bipolar Junction Transistor" aka your bog standard NPN/PNP stuff.

Discrete MOSFETs are actually used fairly often as load switches, these days, and it's very rare to see a SMPS design using a discrete switch that uses anything other than a MOSFET.  Small FETs are much more available in surface mount packages than traditional through-hole type packages, so you might not be noticing them in hobbyist designs as they can be inconvenient from that POV.

There also seem to be fewer "standard" MOSFET part numbers (compare to your 2N3904 type BJT parts that are relatively exactly crossed by everyone in the industry), which can make cross refs difficult.  This also may lead to less usage in hobbyist type projects.  BJTs also tend to be more available at retail (e.g. Radio Shack), which could again lead to hobbyist projects skewing towards bipolar devices.

BJTs also have some robustness advantages over MOSFETs at the expense of efficiency.  You can have a fairly small SOT-23 MOSFET switch a load that would require like a TO-220 bipolar transistor to switch due to the lower on-state losses and total lack of base current on the MOSFET, but the MOSFET can be ESD zapped fairly easily, and relatively minor failures such as floating the gate may (literally) blow the device up, depending on the load.  This is much less likely to happen with the giant bipolar transistor.  Hence, big bipolar transistors are still sometimes used in applications where size is not a factor and robustness is (such as lab bench supplies) whereas the MOSFET may be used in a similar power application where size and efficiency are a major concern, and the environment can be more carefully controlled to prevent failure (such as a mobile phone).

I think there may also still be a number of (mostly older) engineers doing designs that are very comfortable with bipolar transistors but less comfortable designing with FETs.  This may contribute to some "odd" decisions on commercial product designs.

[SavannahLion]

Googling for "floating gate" gets me FGMOS. I take it that's not what your talking about. So when a gate floats, that means its "disconnected"? As in an input that's not tied to high or low?

[MonMotha]

Ah yes, "floating gate" is also a term used in IC fabrication for something different.  Yes, when I said "floating the gate", I meant not driving it with anything i.e. leaving it disconnected.  This can happen inadvertently.  For example, many microcontrollers leave their IO pins in a floating (high impedance) state during reset since the software has not yet run to configure them as inputs, outputs, or whatever.

If the gate of a MOSFET is allowed to float, the gate may rise to a voltage sufficient to partially turn the MOSFET on but not enough to take it into its low on-resistance state.  If this happens on a MOSFET used as a switch for a large load, the MOSFET will conduct, but it will have much higher than intended drain to source voltage.  This may result in a rather large (several dozen or even hundreds of watts, in some cases) amount of power dissipated in the MOSFET, which can literally blow them up.  And people wonder why I wear safety glasses while troubleshooting SMPSes

(There is a solution to this, btw: put a pull up or down resistor on the gate of the MOSFET)

[lilshawn]

I think it's just a cost or availability issue. Although superior in many ways, the FET is often more expensive than a transistor. In circuits where speed or current carrying ability are low, or just simpLe switching is needed, its more cost effective just to use a transistor.  A few pennies here and few there over 10's of components, over thousands of units adds up really quick!

Besides, the transistor has always been a transistor, the fet has gone through several incarnations from the old tube style to what you see now (grid, source, drain? Yes inside the vacuum tube was an actual GRID of metal that ions of electricity would emanate from)

From its crude beginnings...

[MonMotha]

I think it's just a cost or availability issue. Although superior in many ways, the FET is often more expensive than a transistor. In circuits where speed or current carrying ability are low, or just simpLe switching is needed, its more cost effective just to use a transistor.  A few pennies here and few there over 10's of components, over thousands of units adds up really quick!

Besides, the transistor has always been a transistor, the fet has gone through several incarnations from the old tube style to what you see now (grid, source, drain? Yes inside the vacuum tube was an actual GRID of metal that ions of electricity would emanate from)

From its crude beginnings...

I'm not sure about cost.  Discrete MOSFETs are CHEAP, now, possibly cheaper than many conventional bipolar devices, and they're waaaaay better in many applications than a BJT.  I use both: it depends on what I need to do.  Likely availability is more an issue than cost due to the aforementioned difficulty in cross referencing them should one supplier run out (which is happening a LOT these days).

I'm not aware of anyone ever calling the "gate" of a FET the "grid", though it may have happened and it is true that FETs are possibly the closest semiconductor equivalent to an amplifying tube (with the conventional 3-terminal FET being close to a triode; you can also get FETs with several gates just like you could get tubes with several grids aka tetrodes, pentodes, etc., but they're not overly common).  The internal physics aren't the same, but they're analogous.

It is true that there are several kinds of FETs, but MOSFETs are far and away the most commonly used and have been for years.  Just like BJTs and tubes, you can get them with varying spec tradeoffs.

[Ed_McCarron]

If I'm switching a solenoid or a motor via the output on a PIC, I like the FET as a low side switch.  Simple, low loss, works like a charm.

For power work, I still like the venerable 2N3055.

Guess that means I'm older.

[lilshawn]

For power work, I still like the venerable 2N3055.

what not a fan of the 2N2222?? i suppose you loath the 555 too! 

[Ed_McCarron]

For power work, I still like the venerable 2N3055.

what not a fan of the 2N2222?? i suppose you loath the 555 too! 

It's all about power.  2N2222's pop when you pass 10A thru them.  I still miss the LM3909 LED flasher chip that got many a kid following Forrest Mims' guides into electronics.

[MonMotha]

I did an off-line flyback SMPS design that used a Fairchild FCPF16N60 as the switch.  This is a true monster of a MOSFET for its size (TO-220 package): 16A w/ a rated blocking voltage of 600V.  Honestly, it was more than the design needed on those fronts, but the very low Rds(on) of less than a quarter ohm combined with the low gate capacitance (55nC) made it a good choice.  The supply operated from either 22-180VDC or 22-120VAC and delivered a regulated 24V DC at about half an amp.  The humongous input voltage range was a challenge, but the sucker worked.

I had a small heatsink on the FET, but it probably wasn't strictly necessary.  I put it on there as a precaution since Rds(on) increases with junction temperature, and without an easy way to measure the FET's temperature and perform a thermal shutdown, it would have been possible for the device to run away thermally.  I was able to use the isolated TO-220F rather than the metal tab version, at least, so there was no electrically hot heatsink.

Interestingly, MOSFETs do NOT suffer from thermal runaway in a typical class AB push-pull linear amplifier design, unlike bipolar devices.  This may be one reason why were popular in automotive applications.


I've been cooking up a three-phase inverter design for a while, now.  My goal is 60A out at 240V (yes, you read that right: 60 AMPS at 240V).  Probably going to be using some International Rectifier IGBTs (I forget the exact part number), not MOSFETs, though.  The MOSFETs just don't seem to be quite there at that power level, though they'd have lower on-state losses.  I bet that'll change over the next 5-10 years as small, low cost power electronics are driven by demand for the electric vehicle market.  And yes, the power electronics are going to be tested inside an explosion proof housing!