Yes, but this is a bit of reverse thinking.
Agreed, but the OP is kind of coming at this, not as a designer might, but as
someone just trying to understand, broadly. I decided that backing into this,
this way, might communicate the process at one level without needing to then
also know a level above it.
Decide what your spec is *first*. *Then* *chose* a transistor that will
meet the spec. For example, if you are after low noise, you might
consider that the optimum noise (approx) is setting
re=Rsource/sqrt(hfe), to set Ic. Looking at the data sheet for the
transistor to *chose* the current in the first place makes little sense.
Hehe. Agreed!But then, I'm just imagining that the OP has a particular data
sheet or transistor and is then seeing about what might be various
considerations, given that. It's a concrete thing to imagine a single
device/datasheet.
The whole discussion would have been just that more complex and raised to yet
another level, had I folded in what you are suggesting. As a designer already,
I'm sure that's easy for you -- a "been there, done that," kind of thing. But I
imagined starting short of that broader view.
You are designing for the *application*, not the device.
hehe. Yes.
You might want to be able to drive a large capacitive load at a given
frequency. You would then need to chose a highish current.
Yup. Good addition. Actually, I think all this is good, by way of expanding on
ideas. And it really helps me as a hobbyist, to see your point of view, too.
For high speed amplifier design you might want to select a transistor
with low Ccb.
There are many factors in BJTs that can become important, depending on the
application, as I'm sure you know far better than I do. Over time, anyone just
getting started on understanding will develop an 'eye' for more of these as
applications they try teach them.
Or where the simpler mental models basically fail you. One example of this is
where you don't have a little r(e) model in your mind and you simply ground the
emitter of the NPN, for example. What's the gain? You might have previously
figured it as R(C)/R(E), but what does this mean when R(E) is zero??
But it did help me some to start easier and roughly correct for a smaller range
of things and then, gradually to fold in additional concepts (I'm at the
fortunate state where I still have much more to learn, too.)
But if *want* to use a high current, chose a different transistor.
Yup.
Again, its the application that should be driving the choice of
operating current and device, not the other way round.
Agreed. But I chose this route because that's the way I'd want it explained to
me, had I no skills designing for applications but still wanted some idea about
how to calculate things given some particular part. (And more, as a hobbyist, I
don't always want to order a transistor for a project. I will grab my little
box of the few I have (and in my earlier days of being a hobbyist, this was
nothing more than sorted by NPN or PNP and otherwise all together in a place)
and select one that seems "big enough." If I were doing a light bulb switch,
I'd probably pick a "bigger one" and if it were a simple audio amplifier I might
pick a smaller one but where I still may have several (so that I can consider
doing several stages, for example.) That was about my level of thinking, then.
One can take it in either direction. For just understanding the calculations
though, I thought it was helpful to take it in the direction I did. But I like
the additions you've made!
Other curves to look at might be the turn-on and turn-off times
versus I(C) [higher I(C) generally means 'faster'], but that usually
isn't your problem for audio amplifiers, for example.
Often they are. Audio amplifies usually use feedback. The game plan here
is to have the amplifier as fast as possible so that one can apply
lashings of feedback in order to reduce distortion. To keep things
stable with lashings of feedback, one needs to minimise phase shift from
the transistors. This means very fast transisters.
I wanted to avoid the bigger picture, while at least giving negative feedback
some mention so that the OP would at least trigger on the phrase in later
reading. There is a world of beauty in understanding negative feedback from a
variety of dimensions and eventually it becomes a good friend. Your point here
is neatly and tersely put, yet entirely understandable from my point of view. I
love the clear, full, yet economical of use of words. But the OP isn't even at
my modest level of understanding, I fear, and the idea of phase shift (or even
group delay) is probably way past the point of meaning.
Even the idea of exactly what 'distortion' means is probably not quite there to
the OP, yet. I tried to imply that having a gain that fluctuates as your signal
voltage does causes it, but the OP may yet need to actually *visualize* this in
mind before it becomes clearer. Working an NPN amplifier example with an R(E)
that is very tiny, but taking r(e) into account for gain calculations and
figuring the V(C) as V(B) wavers, would probably help the OP see what happens to
the original sine shape. But this takes getting out a graph paper and plotting
it by hand. Getting a spice program to do it for you might let you see the
result, for example, but it's really in the calculations themselves and doing
them by hand that really makes this clear. At least, I think so.
Agreed, but the gain is gently sloping upwards on I(C) up to a point.
hehe. Sometimes, it's important to state it, though. Yes?
It does vary a bit, but is often not a major issue.
Not usually a dominate distortion effect, but can be.
Just trying to list what I could off the top of my head.
-- through the use of negative feedback.
I generally use 1V as a reasonable guideline. I try not to go lower,
but higher is fine. With this and knowing that I(C) is 1mA and that
I(E) is roughly equal to I(C) [when operating normally, anyway], you
can figure that R(E) should be 1V/1mA = 1k ohm.
Probably the "best" reason for an emitter resister is to reduce
distortion.
Yup. And if the OP works out the math on what happens to a sine going from base
voltage to collector voltage, when R(E) is in the ballpark range of, say 10
Ohms, and while taking into account r(e), then I think the point will become
very much clearer.
Once you see the hand-plotted results and have the calculations fresh in mind
that you used to generate it, it sticks with you.
A simple transistor amplifier has around vi(mv)% distortion.
That is 1mv will give 1% distortion. This is a *huge* amount for such a
small signal. An emitter resister will reduce this by around (re/Re)^2.
http://www.anasoft.co.uk/EE/index.html
I think the OP will need to start with just understanding what this distortion
is and viscerally why it arises before getting this newer picture. The OP needs
to hand plot this.
{snip tedious calculations}
Ok.... This can in fact be done automatically in SuperSpice
http://www.anasoft.co.uk/DeviceDesigner.html
Simply chose node voltages and device currents, and press the button, it
will calculate out all the resister values for you.
Of course! I actually have a basic degenerative amp design, with the series RC
leg (and some other topologies) parallel to R(E) and with bootstrapping, where I
can simply set a few design parameters and let it compute the results and plot
gain and phase over frequency, etc. I also have the basic DC amp and the
non-bootstrapped AC amp, for comparisons.
Natually, being a cheap-minded hobbyist, it's in LTSpice, though. Since I
managed to scarf up a bunch of ORCAD model libraries, I've filled in for some of
the really big lack of LTSpice, which is it's relative lack of complete,
non-"Linear Corp" model sets.
You seem to have done pretty well, all things considered.
Thanks. I tried. But I'm also just learning, too, and have much more yet to
gather.
Jon