Transistors

[Nowhere-Man]

To be fair we're many of us amateur engineers not physics students. If we understand amplification that is fine. An engineer needs to know electrons flow from emitter to + and back through the circuit to the collector. The holes appear to flow opposite.
Last night I got a bit stuck on tuning indicators. These were designed to follow AGC (or is it AVG) through a milliameter. As we know when AGC triggers, electrons flow to a valve grid and current is reduced. The milliameter is placed in the anode circuit but the shunt resistance has to be set. It would have helped me if the whole AVG circuit was shown with the indicator which was a Triode. However that is off thread.

[ppppenguin]

AGC (automatic gain control) or AVC (automatic volume control). Never AVG.

An engineer needs to know electrons flow from emitter to + and back through the circuit to the collector. The holes appear to flow opposite.

We don't even need to know that to design and repair transistorised equipment. For NPN devices we need to know that conventional current flows into the collector and base, flows out though the emitter. For PNP devices it's the other way round.

Even for valve equipment where it can sometimes be useful to think about electron flow, most design and repair work is done by think of conventional current which flows into the anode and out of the cathode.

You only need to think about electrons and holes when understanding what goes on inside a transistor. For the most part you don't need to know that, just treat it as a black box with certain properties.

[Nowhere-Man]

I am wired a bit unorthodox. I need to actually visualise electron streams. I can't not fully understand something and proceed. It extends to everything and makes me pretty much unemployable. This extends to phase as well. I will be up late trying to work out the phases in oscillators or it will bug me a lot. I agree though it's removed from the skills you need to repair circuits. Friends figured part of it is OCD but I guess it's educational regardless. Plus in Winter I plan to get the soldering iron dusted off.

[Kalee20]

One of the most important things to understand about transistors is that they are voltage amplifiers, not current amplifiers. Yes a change in base current will give a bigger change in collector current but base current is actually a nuisance, something that we have to put up with. Designers have to make their circuits work despite base current which will vary a lot with teperature and between different devices.

Many people have written that transistors are current amplifiers. They are wrong. Any article that claims this should be taken with a large pinch of salt as the author clearly doesn't understand how transistors work.

I'm going to have to disagree with Jeffrey!


Fundamentally, I think of transistors as current-operated devices. You bias the collector-base junction in reverse, and no current flows.

You then put a bit of current into the emitter, and lo and behold almost all of it gets attracted to the collector. The ratio is alpha and is virtually 1 - highly predictable!

I have designed amplifiers and level-shifters using the predictable and stable property of alpha.

Once we move away from the fundamental connection (common base) to common emitter - admittedly the most frequent manner of use - then yes, beta is not to be relied on, and base-emitter voltage is the quantity to work with (and base current is a nuisance). But Vbe varies with temperature too, as does gm and re.

[Kalee20]

What a brilliant article written in the early days of transistors....

Some of his remarks about valves are not correct.

Some of his remarks about semiconductors are not correct either!

The first page, second column, "In a semi-conductor, electrons travel much more slowly than they do in a conductor...," actually they travel faster! Because there are so few conduction electrons (one per several tens of thousands of atoms according to doping level, as against one or more per atom in a metal), to force 1A to flow along a cross-section of 1mm², these electrons really have to whizz.

And that is why, Hall effect is much more pronounced in a semiconductor.

[ppppenguin]

I agree that alpha is predictable, it's near enough unity in all modern transistors. That's fine for common base amplifiers but doesn't help you for common emitter amplifiers.

Incidentally you might as well say that a valve is current operated when it's in grounded grid mode. Its "alpha" is exactly unity for a triode

[Kalee20]

I am wired a bit unorthodox. I need to actually visualise electron streams. I can't not fully understand something and proceed.

Me too! In fact I always think in terms of electron flow. Current entering the anode and screen-grid of a valve, uniting and flowing out of the cathode? Not for me!
NPN transistors, and diodes - the electrons just flow against the arrows. Never been a problem.
PNP transistors... on the few occasions I design with a negative supply rail at the top of the page, 0V being supply positive, I just ignore the PNP-ness and pretend everything is reversed. Fault-finding with a 'scope on old positive-earth circuitry is fun though, hoorah for the 'invert' knob!

[jimmc101]

Maybe I should modify my assertion to say that a bipolar transistor without any external feedback is fundamentally a current amplifier.

Even a common base stage will be horribly non-linear if fed from a voltage source (ie a low impedance).
Adding a series resistance to the input will reduce the non-linearity but then the resistor is effectively a voltage to current converter.

To generalise for a common emitter transistor, adding external negative feedback can stabilise either the current or voltage gain as desired.

To stabilise the voltage gain add an emitter resistor (series derived series feedback), the limiting case for this is the emitter follower with 100% voltage feedback. (not 100% voltage gain).
To stabilise the current gain add a resistor from collector to base (shunt derived shunt feedback), the limiting case for this is common base with 100% current feedback. (not 100% current gain).

Certainly I would regard a common emitter transistor with an emitter resistor as Jeffrey described i.e. a voltage amplifier with a parasitic input current but I would not apply this to the transistor on its own.

Jim

[ppppenguin]

Sorry, Jim. I must disagree. Let's also not compare apples and oranges. Common base inherently has a current input. As does a common grid or common gate stage. In theory an ideal common base amplifier has a zero input impedance and so you wouldn't attempt to drive it from a voltage. Unless you use a resistor as a voltage to current converter.

So referring to common emitter configuration only. The small signal collector current is defined by the base voltage according to the well known formula for mutual conductance. For better or worse it's highly dependent on the DC collector current so with large signals a common emitter amplifier is inherently non-linear unless you apply -ve feedback. https://en.wikipedia.org/wiki/Transcondu...ransistors

A common emitter stage with current feedback (resistor from collector to base) is rather like a crude opamp. You're creating something approaching a virtual earth at the base. But the current gain is set primarily by the resistor, not the device. As has been said before, the beta of a device is utterly unpredictable and in an ideal world would tend to infinite. Apart from the fact it would upset the bias you can do exactly the same with a valve or FET, both of which have infinite current gain because the grid/gate current is near enough zero.

Out of curiosity I dug out my copy of Horowitz and Hill (1st edition). It's unfortunate that they introduce the transistor as a current amplifier (hfe) and only later bring in the Ebers-Moll transconductance model. To give them credit, as soon as they've introduced hfe (aka beta) they say that it's not a "good" transistor parameter becuase it's so variable. To quote them: "A circuit that depends on a particular value for hfe is a bad circuit". They also say in their intro to transistors that the current gan model is only introductory and they will move on to Ebers-Moll.

[Kalee20]

Maybe I should modify my assertion to say that a bipolar transistor without any external feedback is fundamentally a current amplifier.

Even a common base stage will be horribly non-linear if fed from a voltage source (ie a low impedance).
Adding a series resistance to the input will reduce the non-linearity but then the resistor is effectively a voltage to current converter.

To generalise for a common emitter transistor, adding external negative feedback can stabilise either the current or voltage gain as desired.

To stabilise the voltage gain add an emitter resistor (series derived series feedback), the limiting case for this is the emitter follower with 100% voltage feedback. (not 100% voltage gain).
To stabilise the current gain add a resistor from collector to base (shunt derived shunt feedback), the limiting case for this is common base with 100% current feedback. (not 100% current gain).

A transistor's fundamental operating mode has to be common-base, in terms of internal workings. Current in, current out. (A valve's fundamental is common-cathode, voltage in, current out).

If you connect it in the 'most useful' configuration - common emitter - then it is best to think of it as a black box and not be too bothered about what goes on internally.

And in common-emitter, sometimes I think of it as current-operated (when used as a switch in logic circuits) or as voltage-operated (when used as an amplifier). As the switch, as long as given zero base current, it's 'off'' and as long as given sufficient, it's 'on'. As the amplifier, the input resistance is as ill-defined as beta is - so you have to design such that the effects are negligible.

As for a common-emitter amplifier with emitter degeneration ending up in a limiting case as an emitter follower, I have to disagree! Because raising the emitter resistor more and more drops the gain to the collector to less and less, ending up as zero. To get to the emitter follower, you have to make a step change to the output connection. The same goes for common-emitter with shunt feedback, yes the transresistance is stabilised (ratio of output voltage to input current), but again, to get to a common-base configuration you have to reconfigure the circuit - you can't get there by smoothly altering resistors to zero or infinity.