What happens inside an amplifier?

[wyn palmer]

You do realize that if you charge or discharge a capacitor 50% of the energy involved in the transaction is lost each time? It doesn't generally appear as heat in the capacitor and is independent of any series resistance so the usual explanation is that the energy is dissipated in an emergent EM field- i.e. it disappears as EM radiation?
I know that this is only marginally relevant, and I only mention it to show that the power/energy issue is a complex one even in very simple circuits and that the concept of instantaneity has its own consequences...

 

[bowtie427ss]

What happens inside an amplifier?

Gain.

 

[wyn palmer]

Gain.

The input level into a preamp can be 0.25mv or less and the output from a power amp can be 100v or more.
The input resistance into the preamp can be say, 47k ohms and the speaker load can be, say, 4 ohms.
So, the voltage gain can be 400,000 or more and the power gain can be 10,000,000,000,000 or more.
That's pretty impressive don't you think.

 

[bowtie427ss]

The input level into a preamp can be 0.25mv or less and the output from a power amp can be 100v or more.
The input resistance into the preamp can be say, 47k ohms and the speaker load can be, say, 4 ohms.
So, the voltage gain can be 400,000 or more and the power gain can be 10,000,000,000,000 or more.
That's pretty impressive don't you think.

Zoom in further.

What's the input voltage at the detector stage of an FM tuner?

 

[wyn palmer]

Well,

Zoom in further.

What's the input voltage at the detector stage of an FM tuner?

I, for a long time, designed receivers for cell phones and an LTE receiver I worked on had c. -130dBm sensitivity.
0dBm is 1mW, so we're talking 10^-16W!
The transmit power was c. +20dBm, so the power gain (in some sense) is >10^15, or two orders of magnitude greater than the audio case, so they're in the same ball park.

 

[wyn palmer]

An interesting angle on transistors I was one told about.
The word 'transistor' comes from transfer resistor. As a small voltage varies at the base connection, the resistance between the collector and emitter will vary. If the power supply is connected to the collector, a small variation in voltage at the base, a large variation in voltage appears at the emitter at a constant ratio. Hence, the circuit amplifies.

I thought that was quite a tidy explanation.

Actually that better describes the operation of a MOS Field Effect Transistor (MOSFET), which changes the source-drain resistance as a result of changes in the gate voltage.
To a first order a bipolar transistor changes its collector current as a result of changes in base to emitter voltage (base and emitter more or less track each other with about a diode drop difference between them). The collector load resistor has this changing current flowing through it and thus the transistor converts a small base-emitter voltage change to a larger collector voltage and gives you voltage gain. Ideally the collector output looks like a voltage controlled current source, not a voltage controlled voltage source, but in reality it's not perfect by any means.

 

[birchoak]

I never really "got" bipolar transistors until I saw this. It's like a "valve" but with some closed-loop thing going on ...

That graphic really helps. Thank you.

 

[petehall347]

its magic same as combustion engines . work out times per second them things are going round at and not self destructing .

 

[wyn palmer]

That graphic really helps. Thank you.

The only problem is, that isn't how a bipolar transistor works.
The base current is a dependent variable, it exists because the collector current exists, and the transistor in its simplest form is governed by,
Ic=Kexp(Vbe/vt) where k is a process dependent constant, Vbe is the base emitter voltage and vt is the thermal voltage.
No mention of Ib or hfe in this. IC is just controlled by changes in Vbe and by temperature which changes vt. Basically you stick a voltage (vbe) in and you get a current (Ic) out.
Also Ie=Ic+Ib, and Ib=Ic/hfe.
Yes, there's more to it than this, but basically that's it. Hfe is process dependent and in certain kinds of bipolars it can be >1000 and for others <10. Hfe also varies as a function of the transistor operating conditions and over wafer sites and wafer to wafer and lot to lot.
Yet in all these cases ic=kexp etc. applies even if k changes a bit.
Yes, understanding the physics is tricky and there are all sorts of subtleties that can amaze (the translinear principle anyone?) but all you really need is those simple equations to get a basic grasp of how it all works.

 

[loopstick]

The only problem is, that isn't how a bipolar transistor works.
The base current is a dependent variable, it exists because the collector current exists

Have you ever tested a functioning base-emitter PN junction with a functioning analog ohmmeter?

 

[wyn palmer]

Have you ever tested a functioning base-emitter PN junction with a functioning analog ohmmeter?

Yes, but that's not the same thing as a functioning transistor. If you forward bias the b-e junction it becomes a diode so of course there will be measurable current flow.
If you short the collector to the base it becomes a better diode- so what's your point?
Just in case you hadn't noticed my signature, by the way, I'm an ex Analog Devices Inc. IC engineer- recently retired. You can look me up if you want- I believe that there's still some ADI web presence for me. I was an ADI Senior Design Fellow and had been an ADI Fellow for nearly 30 years.
I worked extensively on bipolar processes for a number of years including being on the team that developed ADI's first complementary bipolar process where I designed numerous circuits and assisted in the optimization of the transistors and the development of their SPICE models. This included the recognition that there was a problem in the process- known as "quasi saturation"- which was before it became a "Thing" in the industry and was included in the generic industry wide models. I worked closely with ADI's process and modelling engineers to improve the devices, including optimizing device implant profiles etc., and to include extensions to their models to allow for simulation. Some of my circuits included using the "translinear principle" to take advantage of the near ideal exponential characteristic of the transistor Ic vs. VBE relationship to do neat things- like building a low power RMS to DC converter using only four bipolars and a capacitor in the compute core. So, yes, I understand how bipolar transistors work from the semiconductor physics and fabrication details, to the models, to the circuit implementation.

 

[lini]

Ic=Kexp(Vbe/vt)

Wouldn't that equation imply that K would have to be a current?

Greetings from Munich!

Manfred / lini

 

[wyn palmer]

Wouldn't that equation imply that K would have to be a current?

Greetings from Munich!

Manfred / lini

Yes, you are right of course.
K is actually Is- the saturation current, and it's a function of device geometry and process.
http://www.ee.ic.ac.uk/pcheung/teaching/aero2_signals&systems/transistor circuit notes.pdf
The above link covers the basics quite well.
https://leachlegacy.ece.gatech.edu/ece3050/notes/bjt/thebjt.pdf
The second link goes into the device equations in more detail showing deviations from the simple (and incomplete) equation that I quoted originally and much more besides.
It's here for interest's sake and doesn't really change the validity of the original post.
Note that BJTs although generally drawn symmetrically in the descriptions of the transistors are usually highly asymmetric. If you reverse the emitter and collector they don't work at all well.
Transistors are usually vertical structures- that is for the BJT the emitter is on top, the base is around/below the emitter and the collector is around/below the base.
The emitter is small and heavily doped- there's a high concentration of the impurity that's needed to produce the N in NPN or the P in PNP.
The base is larger and is doped P in the NPN and the distance from the emitter to the collector through the base is usually quite small.
The collector has the same impurity as the emitter (N in NPN) but has a much smaller concentration throughout most of the region- the exception being the areas that are needed to
collect the current and provide a low resistance path to the collector terminal- these are heavily doped in N.

Greetings from North Carolina!

 

[wyn palmer]

Yes, you are right of course.
K is actually Is- the saturation current, and it's a function of device geometry and process.
http://www.ee.ic.ac.uk/pcheung/teaching/aero2_signals&systems/transistor circuit notes.pdf
The above link covers the basics quite well.
https://leachlegacy.ece.gatech.edu/ece3050/notes/bjt/thebjt.pdf
The second link goes into the device equations in more detail showing deviations from the simple (and incomplete) equation that I quoted originally and much more besides.
It's here for interest's sake and doesn't really change the validity of the original post.
Note that BJTs although generally drawn symmetrically in the descriptions of the transistors are usually highly asymmetric. If you reverse the emitter and collector they don't work at all well.
Transistors are usually vertical structures- that is for the BJT the emitter is on top, the base is around/below the emitter and the collector is around/below the base.
The emitter is small and heavily doped- there's a high concentration of the impurity that's needed to produce the N in NPN or the P in PNP.
The base is larger and is doped P in the NPN and the distance from the emitter to the collector through the base is usually quite small.
The collector has the same impurity as the emitter (N in NPN) but has a much smaller concentration throughout most of the region- the exception being the areas that are needed to
collect the current and provide a low resistance path to the collector terminal- these are heavily doped in N.

Greetings from North Carolina!

One thing I will add- yes, you can indeed drive a current into the base of an active BJT as was shown in the original explanatory diagram- but if you do then what happens is that the base emitter voltage rises and the collector current increases and the collector load R bottom voltage falls as the transistor attempts to balance out Ib and the driving current. If the driving current is too large the device will saturate (the collector approaches ground so that the B-C junction is forward biased) and the beta will fall and the currents will indeed balance.
BJTs are, more or less, never operated this way, and in many cases designers will go to great pains to avoid having this occur in an uncontrolled fashion as it can cause really nasty things to happen- like circuit latch up or oscillation.
So, as a rule, BJTs are operated as voltage in, current out, devices.

 

[loopstick]

Yes, but that's not the same thing as a functioning transistor. If you forward bias the b-e junction it becomes a diode so of course there will be measurable current flow.
If you short the collector to the base it becomes a better diode- so what's your point?

A simple low-side drive lamp circuit. Base-emitter current flowing, collector-emitter current flowing. Bulb is lit. Remove bulb. Collector-emitter current stops flowing. Base-emitter current continues to flow.

 

[wyn palmer]

A simple low-side drive lamp circuit. Base-emitter current flowing, collector-emitter current flowing. Bulb is lit. Remove bulb. Collector-emitter current stops flowing. Base-emitter current continues to flow.

The transistor is no longer operating in the active region when the bulb is removed, in fact it's not even in the saturation region. it's just acting as a rather poor diode and not a transistor at all so the equation doesn't apply. It seems likely that when the bulb is removed Ib will jump up by roughly a factor of beta.
I'm really not seeing your point- if it's somehow trying to prove that Ib is the independent parameter and Ic is the dependent parameter then it's really not applicable, and if the point is that there are regions of operation where the IB is largely or completely independent of IC then well, duh, yes there are, but it ain't no linear amplifier in them...
We were talking about amplifiers were we not, and if you want every possible operating condition for the various junctions to be described then I suggest you refer to all of the various equations describing operation in the various possible modes including non amplifier ones.
The point remains, the BJT, when operating as an amplifier, is a transconductance (voltage in, current out) element and not a current controlled resistance (current into the control port, modulated resistance at the output port). Ib does not define the current out (from the collector), rather Ib is defined BY the current out which is defined (to a first order) by the value of Vbe, thus Ib is the dependent variable in all of this and is two steps removed from the actual independent variable- the control voltage, Vbe.
Think of it this way, if Ic was set by fixing Vbe at a given value and you could miraculously change the value of beta dynamically then Ib not Ic would change.

 

[wyn palmer]

A simple low-side drive lamp circuit. Base-emitter current flowing, collector-emitter current flowing. Bulb is lit. Remove bulb. Collector-emitter current stops flowing. Base-emitter current continues to flow.

Ok, perhaps the problem is in my choice of words "the base current exists because the collector current exists". That is not true if read exactly as written. Better to have said something more along the lines of "the base current is what it is because the collector current and beta are what they are".
If there is no collector current and the base is biased by a voltage then current (electrons) will flow from the emitter to the base and will be defined by the same exponential relationship as was stated before- we have a basic PN junction.
However, once the collector current flows this picture changes. Electrons pass through the base region and onto the collector and when that happens a process known as recombination occurs which results in the bulk of the electrons that used to go to the base now going to the collector terminal while a small fraction of them combine with holes in the base region, giving rise to a small base current of magnitude of IC/beta as Kirchoffs law must be satisfied- that is all the currents into a node must equal all of the currents out of a node- or Ib=Ie-Ic.
There are other components in the base current as well, but we won't get into that.
Explaining the effect of recombination is well beyond anything I could write here and it's really hard to make it sound sensible using words.

 

[loopstick]

Ok, perhaps the problem is in my choice of words "the base current exists because the collector current exists". That is not true if read exactly as written. Better to have said something more along the lines of "the base current is what it is because the collector current and beta are what they are".
If there is no collector current and the base is biased by a voltage then current (electrons) will flow from the emitter to the base and will be defined by the same exponential relationship as was stated before- we have a basic PN junction.
However, once the collector current flows this picture changes. Electrons pass through the base region and onto the collector and when that happens a process known as recombination occurs which results in the bulk of the electrons that used to go to the base now going to the collector terminal while a small fraction of them combine with holes in the base region, giving rise to a small base current of magnitude of IC/beta as Kirchoffs law must be satisfied- that is all the currents into a node must equal all of the currents out of a node- or Ib=Ie-Ic.
There are other components in the base current as well, but we won't get into that.
Explaining the effect of recombination is well beyond anything I could write here and it's really hard to make it sound sensible using words.

Now I remember why I liked the cartoon.

 

[petehall347]

i totally get npn transistors .. pnp is something else .

 

[ConradH]

A lot of this seems way too technical. IMO, there's a really important concept the OP needs to understand. Water analogies fall short in many ways, but are good to start with. You have a garden hose shut off at the nozzle. It's pressurized from the house outlet, but no water is flowing. In electronic circuits, that pressure is voltage. It may or may not be accomplishing anything, but it's what provides the "push". It's also known as EMF for electromotive force. Think about that name; it's important.

Now we open the nozzle. Water flows through the hose. In electronic circuits that's current flow. We control the flow with restrictions like valves. In electronics in the UK they control current flow with valves. Smart. In the US we know them as tubes. You can also control current flow with other restrictions like transistors, resistors, capacitors (for AC), inductors (for AC) and photocells.

Most audio power amplifiers take an input that may be low pressure and low flow (low voltage and low current) and use it to control transistors or tubes that in turn control an output of high pressure and high flow (high voltage and high current). That's a highly flawed analogy, but a good place to start. BTW, there's no requirement that a power amp increase the voltage. It could easily output the exact same voltage that was put in, but with high current (flow) capability. It's just convenient if it also increases the voltage so everything upstream can run at lower voltages.

I haven't touched on the difference between DC and AC, but for the above it doesn't matter... yet.