FETs Vesus Bipolars, Why More Efficient?

[Winfield Hill]

On 26 Nov 2004 12:24:59 -0800, Winfield Hill wrote:

Consider an npn differential amplifier with a 1mA current sink at
the two joined emitters, and a pair of 10k collector resistors.
Analyze the amplifier's gain using transistor beta theory, i.e.
the base current.

and he wrote,

We're looking for Dr. Slick's answer, I'm curious to see what
he'll come up with.

Dr. Slick eventually answered...

Tell you what, i'll give my answer as soon as Kevin can convince
Motorola, Philips, and all the other corporations to stop including
hFE or Beta in their data sheets for bipolars.

Aha, I think we finally flushed him out. Too bad he can't find
a credible answer. Well, perhaps it was too much to expect.


--
Thanks,
- Win

 

[Active8]

On 27 Nov 2004 02:59:05 -0800, Winfield Hill wrote:

Active8 wrote...

If'n yer drivin' another diff stage. One side is showing a gain
of 174 with 3904s in PSpice. That's close to the 200 I perdicted.

Besides load impedance, one has to include Early effect for full
accuracy (spice models usually include an Early voltage estimate).

Is "Vaf" the parameter in der model? I found Vaf = 74.03 in the
PSpice model. That changed my gain calc to 186.

Considering the innacuracies of the PSpice plot and cursor snap, I
can't say for sure that 174 is what I was *really seeing* in PSpice.

--
Best Regards,
Mike

 

[Winfield Hill]

Dr. Slick wrote...

Fuck you.

Our conversation is finished.


--
Thanks,
- Win

 

[keith]

On Sun, 28 Nov 2004 11:54:39 -0800, Dr. Slick wrote:

Winfield Hill <hill_a@t_rowland-dotties-harvard-dot.s-edu> wrote in message news:<cocgk3030iq@drn.newsguy.com>...

Dr. Slick eventually answered...

Tell you what, i'll give my answer as soon as Kevin can convince
Motorola, Philips, and all the other corporations to stop including
hFE or Beta in their data sheets for bipolars.

Aha, I think we finally flushed him out. Too bad he can't find
a credible answer. Well, perhaps it was too much to expect.


Who the fuck are you to get me to jump
through a hoop? Fuck you.

Hey slick! You quite obviously don't know who you're talking to.

go figgr.

--
Keith

 

[Winfield Hill]

lemonjuice wrote...

And it can easily be proved that you can include beta in many
other ways. (grin)

Yes of course, as a barely-measurable 2nd-order effect.


--
Thanks,
- Win

 

[Active8]

On Mon, 29 Nov 2004 14:46:07 -0500, Active8 wrote:

On 29 Nov 2004 03:31:42 -0800, lemonjuice wrote:

V= Integral E.dx ie a scalar quantity

Hey Poindexter. Do you thing you can find Integral f(x) dx with only
one point? Even if one limit if integration is at infinity, it's
still a point, i.e. dx has two ends.

Oh, sorry about the math. My experience in the volts per meter field
is that the meter has two distinct endpoints.
--
Best Regards,
Mike

 

[Active8]

On 29 Nov 2004 12:35:20 -0800, lemonjuice wrote:

I don't see it written that current flow is not english . LOL
^^^^^ ^^

charge flow flow

Stop trolling. It is redundant.

<snip the rest of the troll>
--
Best Regards,
Mike

 

[Active8]

On 29 Nov 2004 12:49:46 -0800, lemonjuice wrote:

On Mon, 29 Nov 2004 08:20:12 GMT, "Kevin Aylward"

I don't think you can make an "insulated base"!


Arguable, there is. However, they are called Insulated Gate Bipolar
Transistors, or IGBT.

Error.

No way. The current goes through A MOSFET's gate that is insulated and
is connected to a BJT.
Thats how a IGBT is made. Get your facts straight.

Yeah but you've been saying there is *no* current through a MOSFET
gate. Now you say it does, duh!
--
Best Regards,
Mike

 

[Winfield Hill]

Active8 wrote...

Are you sure they're constant Vbe curves and not Ib curves like
you find in data sheets?

You should get nearly (30%) the same answer either way.

The two I'm looking at don't get linear except possibly at the
far right extreme and that puts Va on the positive Vce axis.

What parts are these, and what conditions?


--
Thanks,
- Win

 

[Jonathan Kirwan]

On Mon, 29 Nov 2004 14:54:33 -0500, Active8 <reply2group@ndbbm.net> wrote:

On Mon, 29 Nov 2004 10:08:10 GMT, Jonathan Kirwan wrote:

On Sun, 28 Nov 2004 23:30:42 -0500, Active8 <reply2group@ndbbm.net> wrote:

On Sun, 28 Nov 2004 22:42:38 GMT, Jonathan Kirwan wrote:

On Sun, 28 Nov 2004 15:31:06 -0500, Active8 <reply2group@ndbbm.net> wrote:



I don't have Var in my model. But this Vaf I have... is there a
parameter which shows at which point on the I-V curve that value of
Va refers to? Looks like something I might adjust for different
operating points.

The forward Early voltage shows up on the negative side of the V-axis. It's an
'extrapolation' backwards,

Yes, but the tangent is drawn from the point on the I-V curve (the
slope of which isn't constant for a given Vbe) that you're operating
at, i.e., Vce,Ic ... It looks like it'll intersect the Vce axis
farther to the right for higher Vce.

I'm looking at one of several charts right now with several Vbe curves drawn.
When I place a ruler on the slopes (once Vce reaches the point where the curves
are roughly linear at higher Vce), they all converge at about the same point on
the Vce line. I think this is often -50V to -70V, or so.

Are you sure they're constant Vbe curves and not Ib curves like you
find in data sheets? The two I'm looking at don't get linear except
possibly at the far right extreme and that puts Va on the positive
Vce axis. At any rate, the line starts at the Vce,Ic operating
point.

They are kind of idealized, but that's how the model works. The basewidth
modulation is a kind of (1+Vbc/Va) thing. I'm pretty sure that spice simulates
it as exactly linear and that this seems good enough for real situations. Are
you looking at measured curve traces?

Who knows where they came from. One from Win's AoE and one from a
web page. Try:

http://www.mtmi.vu.lt/pfk/funkc_dariniai/transistor/bipolar_transistor.htm

start reading below fig 4., heh, the *second* fig 4, that is.
There's a bunch of base width modulation stuff above all that, too.

Maybe I don't understand the point/question, though.

They call it the "point of interest" in this article.

Gotcha. I see what you are talking about. I've seen the same thing in real
curves, as well. The modeled term, Va, (which was first developed, I think, in
the EM3 model and carried forward into later models) does NOT address this
reality. And if you use a spice program (I have done exactly this test) to plot
these curves for a transistor model, you'll see __perfect__ straight lines. I
think this demonstrates that the models do not incorporate the actual facts.
But the residual error appears to be "good enough" for most design purposes.
The difference *will* add some distortion to a signal, though I'm a hobbyist and
never needed to be that concerned about what difference it might make.

The modeling argument goes like this: (1) it is assumed that the BJT is
operating in the linear region; (2) results taken from simple, constant doping
analysis are assumed to hold in the general case. Given these, the
simplification is justified by noting that basewidth modulation is itself just a
2nd order effect, so a first-order analysis of it is acceptable. Thus, the
Taylor's expansion of the modulation effects is reduced to the 1+Vbc/Va I
referred to, earlier.

Does that make sense?

Jon

 

[Active8]

On 29 Nov 2004 13:11:46 -0800, Winfield Hill wrote:

Active8 wrote...

Are you sure they're constant Vbe curves and not Ib curves like
you find in data sheets?

You should get nearly (30%) the same answer either way.

Ok.

The two I'm looking at don't get linear except possibly at the
far right extreme and that puts Va on the positive Vce axis.

What parts are these, and what conditions?

You tell me - one's in AoE The other's at the link I posted and
it's just an example.

--
Best Regards,
Mike

 

[Active8]

On 29 Nov 2004 12:45:10 -0800, lemonjuice wrote:

Thank You LemonJuice!

You're welcome Dr. Slick

That's gay.
--
Best Regards,
Mike

 

[Kevin Aylward]

Active8 wrote:

On 29 Nov 2004 12:49:46 -0800, lemonjuice wrote:

On Mon, 29 Nov 2004 08:20:12 GMT, "Kevin Aylward"

I don't think you can make an "insulated base"!


Arguable, there is. However, they are called Insulated Gate Bipolar
Transistors, or IGBT.

Error.

Not at all. "Arguable", was specifically included to immunize that
statement from attack

You have to get up a bit earlier to catch me out on this sort of thing.

Kevin Aylward
salesEXTRACT@anasoft.co.uk
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

 

[lemonjuice]

On Mon, 29 Nov 2004 17:04:02 -0500, Active8 <reply2group@ndbbm.net>
wrote:

On 29 Nov 2004 12:45:10 -0800, lemonjuice wrote:

Thank You LemonJuice!

You're welcome Dr. Slick

That's gay.

I thought you had a soft spot for Kevin

 

[Jonathan Kirwan]

On Wed, 01 Dec 2004 17:41:25 GMT, Jonathan Kirwan <jkirwan@easystreet.com>
wrote:

Is Ro a common emitter resistor?

By this use of 'common' I mean "shared in common," in case that's not clear from
the context. Not "common emitter configuration" which has nothing to do with
Win's posed problem.

Jon

 

[Kevin Aylward]

Dr. Slick wrote:

"Kevin Aylward" <salesEXTRACT@anasoft.co.uk> wrote in message
news:<0Uerd.56191$F7.15351@fe1.news.blueyonder.co.uk>...


Your level of confusion? You're right!

Oh dear...


Well, yeah, you should be worried!

I am shaking my head in pity, oh dear...

THE STUFF YOU POSTED IS IN EVERY BASIC CALCULUS BOOK!

Oh?

Care to name some "basic" calculus books that includes a derivation
of the Reimann curvature tensor?

Just the one will do.


You have the fundamental theorem of calculus
on your site, if you haven't noticed!

No, I havent noticed. Where would that be?

I could put it there if you need a refresher on it.

Why don't you do a "paper" on 2+2=4?

Because its way, way to difficult for me. Seriously. I have had a deko
at a pretty decent abstract algebra book, and it takes about 100 pages
to actually prove a result like this from fundamental group theory.

You don't know shit, Holmes.

Great argument you present.


Don't need to. It's clear everyone
here knows you are full of shit.

I actually have quite a decent reputation technically in this NG.
However, there always are few too iliterate that I am unable to help.

Also, your so-called "papers",
and your idiotic posts are more than
enough evidence of your amoeba IQ.

So you should have been able to identify some specific errors in them
then. Are you really Guy Macon in disguise?

Considering all the physics text books agree with me, I think it is
apparent who lacks the education.


You only think in your own mind that they agree
with you.

Actually, the basic electronics texts agree that
you can call a common emitter BJT a CCCS.

I agree many basic books do this. They are all wrong. None of the
academic degree level texts do this. So, sure, if you want to go by the
Bantam paperbacks, by all means do so. Maybe the National Enquirer is
even a better bet for you.

Really,

the fundamentals are more important that the more
complex texts. Your earlier education sets the
foundation for higher learning.

That's correct. I couldn't agree more. That's why it is important to
teach people that the transistor is a voltage controlled device right of
the bat. Once some get the idea that base current controls the emitter
current, it is essentially impossible to get all but the brightest to
realise their misconception. Clearly you are one of the duller knives in
the cutlery cabinet.

you obviously fell asleep during Circuits 101,
so i suggest you go back and re-take the course.

Ho humm...

I heard the song you posted on your
website... it sucks donkey balls like
your "papers"

Its not a song, its an instrumental. I suppose thee is no accounting
for bad taste though.


It's more like noise to me.

No doubt coming directly from one region of your brain to another.


Kevin Aylward
salesEXTRACT@anasoft.co.uk
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

 

[Active8]

On Wed, 01 Dec 2004 21:11:52 GMT, Jonathan Kirwan wrote:

But that doesn't explain to me why you think it is then okay to drop Vce. I
know that it seems to you that it "works better," but it seems to me that a
theoretical approach would insist that it must be included and that there is no
proper basis for removing it, unless you can formulate one.

How about Vce << Va ?

--
Best Regards,
Mike

 

[Jonathan Kirwan]

On Sun, 28 Nov 2004 15:31:06 -0500, Active8 <reply2group@ndbbm.net> wrote:

On Sun, 28 Nov 2004 17:29:04 GMT, Kevin Aylward wrote:

Active8 wrote:
On 27 Nov 2004 02:59:05 -0800, Winfield Hill wrote:

Active8 wrote...

If'n yer drivin' another diff stage. One side is showing a gain
of 174 with 3904s in PSpice. That's close to the 200 I perdicted.

Besides load impedance, one has to include Early effect for full
accuracy (spice models usually include an Early voltage estimate).

Is "Vaf" the parameter in der model? I found Vaf = 74.03 in the
PSpice model. That changed my gain calc to 186.

Yes. Vaf is the foward early voltage. Var is the reverse early voltage.

And the usefull result is max gain in a simple amp is Va/Vt.

I don't have Var in my model. But this Vaf I have... is there a
parameter which shows at which point on the I-V curve that value of
Va refers to? Looks like something I might adjust for different
operating points.

The forward Early voltage shows up on the negative side of the V-axis. It's an
'extrapolation' backwards, just as if you placed a ruler along the normal mode
traced curve in quadrant I and drew it backwards until it intersected the
V-axis. If you visualize the curves, you'll see that larger Early voltages
imply closer-to-zero slopes on the normal operation side. Like the saturation
current, Is, it's an extrapolated value. Not directly measured.

Jon

 

[lemonjuice]

Hey Poindexter. Do you thing you can find Integral f(x) dx with only
one point? Even if one limit if integration is at infinity, it's
still a point, i.e. dx has two ends.

OK... man then ... In fact 1 of the many possible definitions of
Potential says its related to the work done in bringing a unit
charge
from infinity to an electrical field.

So man you already got your second point defined. When I say whats
the
potential you already know its related to infinity.

Look up equipotential surfaces. Like I said, my Rx antenna has a
voltage at the feed that's proportional to it's length and the
strength of the electric field in V/m.

Electrical field strength = Potential ... Thats another new theory
(grin)

BTW equipotential surfaces have nothing to do with this.

Look up the definition of Electrostatic potential.

For a point (x,y,z) its mathematically its

V(x,y,z) = q/4 * pi * epsilonnought * r
As I said before the reference point is at infinity.

 

[Jonathan Kirwan]

On 4 Dec 2004 20:07:20 -0800, Winfield Hill
<hill_a@t_rowland-dotties-harvard-dot.s-edu> wrote:

You were going to post a link to this, "Modeling the Bipolar
Transistor," by Ian Getreu (when he was working at Tektronix and
probably with their STS (semiconductor test systems) group when
it still existed.)" Right?

I've no link, Win. I'll need to type it up in text form.

Jon