BJT - a current-controlled device?

[MrAl]

Hello,


My suggestion is that in order to reach any agreements that the subject material should be divided into small parts and discuss each small part so that the replies are not too huge and time consuming to read through.

My other suggestion is that if you want to show stability then you need to show the sensitivity functions for each parameter you want to show is somehow more stabilized because of one given way of doing it over some other way of doing it. This will result is a set of numbers that are easily compared and impossible to argue with.
For example, if one says that gain with Vbe is more stable using technique A or method A than is using technique B or method B, then they must be able to show that the sensitivity function is less for A than for B for the entire set of parameters that are of interest. The sensitivity functions can be expressed in terms of the first partial derivatives.

 

[ericgibbs]

Hi Winterstone,

As you are the OP who asked the question. BJT - a current-controlled device?

I would suggest it would be a good idea if you posted a brief 'closing' summary, stating how the replies you have received have helped you decide.

IMHO a concise, impartial summary, highlighting the points made by both sides of the argument would be helpful to the less experienced members who are interested in deciding how a BJT is controlled.

Eric

 

[Winterstone]

Your paper makes no sense. Is "g" supposed to be "gm"?

OK - it makes no sense. Clear statement, but without justification. However - you cannot deny that the voltage mode (A) resembles a closed loop that has a loop gain that is confirmed by measurements.
You are right - I didn`t explain the meaning of "g". I have assumed that you are able to guess that a block between a voltage (input) and a current (output) must be a transconductance (in Germany we normally use the symbol "g" instead of "gm").

. The small signal collector current ic can be expressed as hfe*ib, or gm*vbe. But what is vbe? To compute, we must consider Re', as well as r_pi. How do we get r_pi? We compute it as r_pi = hfe/gm. So it is impossible to omit hfe from any stage analysis/calculations, nor alpha.

It seems you didn`t understand the block diagram. The block g=hfe/(r_pi) contains everything you like to see. So - what is wrong? Surprisingly, the result (loop gain) is in accordance with measurements.

Also, your equations for "loop gain" make me ask the question "are you referring to parallel feedback, where output is fed back to input through 1 or more resistors forming a loop"? "Loop gain" usually refers to an amp with global feedback, where output is fed back to input via resistive divider. If "A" is open loop gain, and "B" is feedback factor (fraction of output returned to input), then loop gain = T = A/(1+AB). I don't think that model covers this case which is series feedback. What is "open loop gain" here, and what is "feedback factor". Please explain. Thanks.

OK - I have assumed that you are familiar with control loops. I agree - some explanations may be in order.

Loop gain is simply the product of all transfer functions of the various blocks connected in a closed loop (including the "-" for negative feedback). Of course, this is NOT restricted to resistive dividers.
More than that, of course this applies to ALL kinds of feedback (series or parallel).
The formula as given by you is NOT the loop gain but the gain for the closed loop (in case of negative loop gain).
(Now I understand your comment "paper makes no sense" - you are mixing loop gain with the gain of the closed loop).

Applying simply the rules of block diagram calculation the feedback model for case (A) in my paper result in the gain expression (with Re feedback) as given in the paper:

G=-g*Rc/(1+g*Re).

As another rule of block diagram interpretation: When the closed-loop expression is written in the so called "normal" form (numerator: gain without feedback, denominator: (1+F(s)), the function F(s) is identical to the loop gain LG of the system. This can be confirmed by simple optical inspection of the loop components. Thus, the loop gain for (A) is LG(A)=-g*Re.

For my opinion, all other parts of your reply are either repetitions of statements made earlier or basics that needs no further comment (collector current ic can be expressed as hfe*ib, or gm*vbe, ...A bjt has an input current ib and voltage vbe...So beta, or "hfe", is a measure of the current gain of the raw device...)

W.

 

[ericgibbs]

hi All,

As with ALL the other Threads raised on this Topic, it will never reach a conclusion.

I did request in my post #62 for the OP to summarise, but it appears the tit for tat posts will never end.

The Thread has been Closed

Moderation.