FET vs. MOSFET vs. Transistor vs. Opamp vs. Triode

[TryingToDo613]

Hi all. Wondering if there is an overview online of the comparisons of these gain devises, like what corresponds to what and how they work similarly.  I have an old army publication that has some comparison between tubes and transistors, but it's set up mostly as a sales presentation to get army designers to design with transistors. Seems to me like all these things work pretty much the same way and all you have to do is wrap your brain around them and these circuits will make a whole lot more sense. -ph

[R.G.]

I don't know of such a treatise on line, but it's not that hard.

Let's do the discretes first; JFET, MOSFET, bipolar transistor and tube.

All of these have three (effective) terminals. Two of them form a current carrying path that is turned more or less on and off by what happens on the third terminal.  EE textbook, those much-maligned instruments of torture that are always wrong, have a pretty good send up of this.

What you do is to change either the voltage on the control terminal with respect to one of the current-path terminals, and this changes the amount of current flow. One of the current-path terminals is a high impedance terminal, the other is a low impedance terminal.

The matching terminals are
High impedance current path terminal: drain for all the FETs, collector for bipolar, and plate for tubes.
Low impedance current path terminal: source for all the FETs, emitter for bipolar, and cathode for tubes.
Control terminal: gate for all FETs, base for bipolar, and grid for tubes.

In all of these devices, increasing the voltage on the control terminal with respect to the low impedance current path terminal increases current flow between the two current path nodes. This has the effect of making the low impedance node rise in voltage (for NPN/N-channel polarity and tubes) in the same direction as the control terminal moved, and making the high impedance node go lower in voltage. So they all act like inverters if you take the output off the high impedance terminal, and they all act like followers if you take the output off the low impedance terminal.

Note that Bipolars are unusual in that their action is best described as a current multiplication, the active item being the current into the base, but the base-emitter voltage does follow the description that I just gave.

Some JFETs, some MOSFETs and all tubes are depletion mode devices. This means that if you leave the control terminal alone, they conduct heavily. So you have to bias them off to get them to behave linearly.

Some JFETs, some MOSFETs and all bipolars are enhancement mode devices. If you do nothing to the control terminal, they don't conduct at all, so you have to set up a DC bias to turn them on slightly to behave linearly.

For clarity, I've chosen not to talk about the common base, common gate,and common grid circuits. They're confusing to this line of thinking. Just note that once you get the idea right, there are embellishments.

The reason I've not talked about opamps is that they don't act like any of the others. You can make the discrete devices act op-ampish, but it's not a clear leap of understanding. Best to initially get down what the discrete devices do, then go more complex with opamps.

Was that the kind of thing you were looking for?

[petemoore]

Yes.
 Amazing how you typed that up. For me it is a very concise explanation of the basic functioning of transistors and tubes.
 I'll try to remeber  FET vs. MOSFET vs. Transistor vs. Opamp vs. Triode, and dig this up when questions such as this get touched on, it'd be nice if there were a link to valued posts like this one, is there any way you could add this to your site?

[Peter Snowberg]

Wonderful 101 post R.G.

Pete, I just added a link to this thread in the "ISI" section. If you have posts you feel should be always on hand like the one above, please let me know.

[petemoore]

Thanks Peter   . !
 I'll take a look in there !!!

[aron]

archive

[Coriolis]

AND you could just bookmark good threads - I do it all the time!  

[robotboy]

Don't forget the wiki!

http://diystompboxes.com/cgi-bin/diystompboxeswiki.pl?DIY_Stompboxes.Com_Wiki

I've been adding interesting posts as I find them. You're welcome to do the same.

[TryingToDo613]

Thanks RG. I'm going to print it and read it a few times.  It is pretty much what I'm looking for.  Haven't found an EE text that has that stuff. Could you suggest one?  

The one thing I'd say is missing is why FETs and Mosfets were brought into the picture. As I said,  I have a military sales pitch on transistors so I understand why the world was so hot on them over tubes, but Marshall based a whole line of amps calling them specifically Mosfets, and I still see people on ebay saying that they sound just like tube amps because they do something similar with odd order harmonics that standard transistors don't. What are the useful differences? Are there still available texts with this stuff in them?

And on Opamps, seems like there are two sets of three in there, so I figured it was yet another plate/grid/cathode copy kind of thing. If it's too complicated to get into here could you at least tell us what makes them more complicated?  I'm a really lousy guitar player and I don't know if I'd know good tone if it bit me. From the replies to this post I guess I'm not the only one who just wants to figure out what's going on inside these things and be able to design my own.
-ph

[petemoore]

Here's my take, transistors will do a good clean tubey sound.
 They'll also do a heavy metal type crunch, pretty good on the 'inbetweens'.
 One sound at a time should be tweeked for best results, using Q's. Driving a 'certain' tone 'made by transistors and tweeked'] with a FUZZ box or OD, might not work the same as driving a tube, which, if tweeked may allow a wide variety of well tweeked tones to be 'Stomped In' the chain of the signal path.

[R.G.]

It is pretty much what I'm looking for. Haven't found an EE text that has that stuff. Could you suggest one?

Yeah - go to a used books store and find the section on engineering. Texts that were current for about five years around 1960 were hot on it. Today, I don't think EEs are taught much about it these days. There's not enough actual circuits design work to keep many of them employed at it. It's all logic and RF, which has made a comeback.

The one thing I'd say is missing is why FETs and Mosfets were brought into the picture.

Mostly because they were available, and were distinctly different from either bipolars or tubes in some ways. I guess it's "because we can."

There was a whole generation of EEs that faded out because they couldn't make the jump to transistors. The big deal about bipolars is that they were current fed devices, not voltage fed devices, and came in both NPN (similar to N-channel and tubes) and PNP which is impossible with tubes because of the laws of nature.

The tried-and-true ways of dealing with inputs were all thrown out because bipolar bases were low impedance, not high impedance, and they worked at insanely low (for the times) voltages.

JFETs were actual speculated accurately before bipolars, but the semiconductor fabrication methods could not make them work well until after bipolars were discovered. And we *still* can't make them consistently.  The longing for JFETs won't go away because they bias in a similar way to tubes, but work at low voltages and high currents. MOSFETs were an extension of JFETs originally but really came into their own when it was found that they made low power logic, lower power and as fast as bipolar transistor logic, and with the ability to make tiny MOSFETs for logic, the obvious extension to making all the zillions of baby MOSFETs on one chip link up in parallel for a power device gave us power MOSFETs. I'm obviously glossing over huge amounts of stuff, but this is moderately accurate. It depends on what you think is important in the history, I guess.

As I said, I have a military sales pitch on transistors so I understand why the world was so hot on them over tubes, but Marshall based a whole line of amps calling them specifically Mosfets, and I still see people on ebay saying that they sound just like tube amps because they do something similar with odd order harmonics that standard transistors don't. What are the useful differences?

Yeah, that's the standard misconception. Depletion mode JFETs do bias in the same manner as tubes, in that they conduct with zero bias, and you have to reverse bias the grid to get them to back down into the linear region. However, the internal physics and governing equations are quite different. Which is engineer-speak for "but they sound different."

Bipolar transistors are an exponential-law device. That is, the output current is an exponential function of the input voltage. MOSFETs and JFETs are square-law devices, in that the current through the channel is a function of the square of the voltage between gate and source. Tubes are a 3/2 law device - the output is proportional to the 3/2 power of the input voltage. All of the really ugly variations get linearized out in the circuit applications, which is why we get fairly linear operation out of them, but the little oddities remain, and that's what we hear as the side effects on tone.

There is a never-ending quest for replacing tubes directly with JFETs to get tube tone in a transistor circuit. While JFETs may sound good on their own, they are NOT direct replacements for tubes in tone. A person saying that they are is either in earnest but misinformed, or doesn't care as long as they get your money. JFETs and MOSFETs do clip more softly than bipolars and have a different sound, but they are NOT tubes, and don't sound the same way. But we'll likely have at least one more generation of techies "discover" the JFET/tube substitution. Use JFETs and MOSFETs because you like the sound, not because they "sound just like tubes". They don't.

Bipolars have a high power gain, and quite high usable voltage gains. JFETs are clever devices that do nice tricks (like voltage variable resistance) but have never fulfilled their potential because of some apparently ineradicable oddities. MOSFETs are becoming true workhorses, because they have nice features like softer clipping, no thermal runaway (the ugly secret of bipolars) and quite high power capabilities, along with predictability.

That wasn't what you hoped I'd say. I can't tell you "use bipolars this way and JFETs that way" because there is so much variation possible.

Are there still available texts with this stuff in them?

My opinions were formed over about forty years of reading electronics stuff in textbooks and out. Used book stores are a gold mine if you know what you're looking for. The technology you want to see was current in the early to mid sixties for everything except the MOSFET, and mid 70's for the MOSFETs.

Better than just texts is to do both textbook study and to voraciously eat up all the circuits you can find and tinker with. I always get slapped in the face by a few people for thinking that theory is worth anything; the self made men in the world love to think that "the textbook stuff is all wrong" when in fact, it's a useful guide. But it's not complete without you doing the tinkering and listening, and the textbooks certainly won't give you a rundown on what kinds of distortions you can make sound good. Study -and- tinker.

And on Opamps, seems like there are two sets of three in there, so I figured it was yet another plate/grid/cathode copy kind of thing. If it's too complicated to get into here could you at least tell us what makes them more complicated?

Opamps were developed out of the world of pure theory reduced to practice. Feedback's effects, good and bad, were systematized by a few people in the mid to late 1930's. The ideas were applied to existing amplifiers with reasonable success until someone came up with the differential amplifier (this was a theoretical development in academia) and reduced it to practice with triodes in the 1940s. Philbrick was a big name then - they made recognizable operational amplfiers out of 12AX7's that worked just like todays within the limits of the devices. Analog computers that did gun aiming on battleships were a military secret in WW2. These were good enough to let a 16" gun throw a one-ton projectile over 20 miles and hit within *yards* of the desired aiming point. These were tube opamp based setups.

So yes, opamps are a different bird. The essence of an opamp is a high gain amplifier with an inverting gain and a means for summing feedback into the input node. The differential amplifier made this easy and repeatable and was a feature on all opamps as soon as it made its way out of the schools. What makes opamps different is that you can make *everything* inside the feedback loop disappear and only the external component values control what happens on the output. This is true as long as certain conditions hold. The high gain and feedback connection give you complete freedom from worrying about what's happening inside the amp, even if it's not a particularly linear amplifier. You just don't care that the second stage transistor is only half the gain you expected - half of a million for the overall amp is still pretty high.

All of the negative comments you hear about opamps comes from operating them outside the certain conditions. If you make the output bang against the power supply rails, the effective gain drops to zero, and the opamp doesn't act linear any more. So output clipping on an opamp sounds distinctly different, and was positively ugly with some of the early semiconductor opamps. I don't know about those Philbricks, though   . Exceed the available slew rate, and they can't be linear any more. Stuff like that. Today's opamps have gotten incredibly better than they started.

I'm a really lousy guitar player and I don't know if I'd know good tone if it bit me. From the replies to this post I guess I'm not the only one who just wants to figure out what's going on inside these things and be able to design my own.

I'd say that most of the people here want to figure it out, guitar player or not. I've come to decide that I'm lucky that I'm not a better guitar player. If I had the talent in any great abundance, I'd have followed the smokey-bar-gig trail and burned out like so many musicians I've seen in Austin. As it was, other folks could play rings around me, so I had to give up and learn a trade, disreputable as it is.  

[TryingToDo613]

Forty years. sheeesh. guess I better get going. thanks as usual. I've got a couple more topics for the archives. Have to figure out how to ask them, and as you said, tinker. -ph