As with most postings, the topic has veered away from the original question. But that's OK.
I would like to address the question of whether a BJT should be considered as current driven or voltage driven. To a certain extent, the answer is 'both" or "either".
This is almost like arguing Norton's/Thevenin's Theorems in which, if you have an ideal current source of 1A in parallel with a 1ohm resistor inside a box that has the 2 terminals coming out and then you have another identical box with an ideal 1V voltage source in series with a 1 ohm resistor and those terminals coming out, how do you tell the difference without opening the boxes? (Ignore the fact that you can't create ideal current and voltage sources). Mathematically, they are identical so, in theory, you can't tell the difference. [Aside: In practice, you can tell the difference. This is a well-known interview question. Answer at the bottom of this posting - to allow you time to try to figure it out yourself. It's a tough one.]
The problem with the current-driven and the voltage-driven advocates (including me) is that they have not stated their position accurately. In many circumstances, a simple current model in which you have Ib going in to the base and beta*Ib in the collector is more than adequate. This is a simple first-order model and I'm a big advocate of always using the simplest model that will do the job.
The real statement that the current-driven advocates should use is "For my applications (or for many applications) looking at the BJT as current-driven works and works well." Extrapolating from that to advocating that this is the best way to look at it for all applications is wrong.
And the same applies to the voltage-driven advocates. Stating that "since this works for me in my applications it is therefore the best way for everyone else to look at in their applications" is also wrong.
Here is my more well-thought out position: If a current-driven model works for you, use it. If a voltage-driven model works for you, use it. But don't insist that yours is the only right one.
The voltage-driven model (and by this I mean the voltage-driven way of looking at the BJT) is a more natural way of looking at the BJT from a solid-state physics and understanding of the basic operation perspective. For example, the normal, active region of operation is defined as the BE junction is forward biased and the CB junction is reverse-biased. I can picture how that is done by providing the appropriate voltages across these junctions. I cannot picture how that is done by looking at the base current.
The genius of the analysis by Herman Gummel and Sam Poon in their description of how the BJT works is the realization that the transistor saturation current (Is) - which is not the saturation current of the BE or CB junctions- is a fundamental parameter of the BJT. It inherently contains within it the explanation for beta versus Ic, why the output resistance variation with Ic can be modeled by one parameter (the Early voltage - named after Jim Early) and all the other second-order effects. It also inhently models all other regions of operation (inverse, saturation and off).
It also explains the reciprocity property that people talk about when they derive the Ebers-Moll model. Everyone says "It can be shown that alphaE*IsBE = alphaC*IsBC". [These equal Is]. But no-one ever shows that. Primarily because no-one knows how to. But the Gummel-Poon model shows that reciprocity is just saying the integral of the doping level in the base from E to C is the the same as the integral of that doping level in the base from C to E - which is intuitively obvious. If you don't understand this paragraph, don't worry - it's not important.
Again, if your application is not affected by these second-order effects or they can be adequately covered by simple modifications then the current-driven model is perfectly fine.
The bottom line (and I'm sorry it took so long to get there) is the current-driven model is just as good as the voltage-driven model for applications where the first-order effects are the only important ones (and that includes many, many applications). However, if second-order effects (beta v Ic, ro v Ic etc) are important in your application (usually when you are stretching the limits of the device or trying to stretch certain specifications), then the voltage-driven model is a more natural one to use.
Answer to the boxes question: One is warm.