Are hybrid parameters useful for design?

[[email protected]]

Who cares who he is?

You're another foreigner. In American usage, that is a declarative and not an interrogative.

 

[[email protected]]

On Tuesday, October 16, 2012 10:35:02 AM UTC-4, Jim Thompson wrote:

All these flames from unknowns are just more of his aliases coming out of the woodwork...

 

[rajibbandopadhyay]

Yup, back-to-back diodes, with the lower one about a 6 volt zener. And
the mentioned controlled current sink (whose gain rolls off at high
frequencies and low temperatures.) Add some capacitances, and some
emitter series resistance, and that covers 98% of the cases. Add in a
little output slope (the Early voltage) and some leakage if it matters,
which it usually doesn't.

Okay, great! Thanks.

 

[John S]

Static beta, aka hFE (not hfe!) is on basically all bipolar transistor
data sheets. The other h-params rarely are.



here:========================================================

On Sat, 13 Oct 2012 17:24:52 -0700 (PDT),


Yeah, mostly older. Of the transistors that we use, most data sheets
have hFE, and only one out of the dozen that I checked had any other
h-params.

http://www.centralsemi.com/PDFs/products/2n3903.pdf

http://www.fairchildsemi.com/ds/2N/2N3904.pdf

==============================================================




"They" is plural. hFE is singular. We call it "beta." The point of
this thread isn't whether beta is on data sheets; the question asked
was whether anybody actually uses h-paramater analysis, as is taught
in university courses.

Just hFE isn't enough to do the academic matrix h-param stuff. So the
answer to the OP is "nobody here seems to use it."

I did a search and found that you are correct about hfe/Hfe/hFE. I had
no idea that the capitalization meant anything. For all these years I
thought hfe was just hfe. My mistake.

I hope you and the group will accept my apologies.

JohnS

 

[[email protected]]

He's been polite, and the question about h-params was reasonable.

He's no Indian- and his conversation has been nothing but broad innuendo and
generalities. His understanding of meritocracy is that of classic white
trash American, that's the main giveaway, but you're the same so you didn't
pick up on it.

....and Democrats think Republicans are xenophobic racists. You lefties are
really something.

Gee whiz, you sound like a round-the-bend angry s**t slinger. STFU.

He's just an American version of Phyllis. Ignore.

 

[John S]

It doesn't matter much. hfe and hFE are similar, because transistors
are fairly linear as current amplifiers. The range of beta on a data
sheet might be 5:1 min/max, or some min and no specified max, so most
people go with the min and try to design circuits that are tolerant of
anything better. So this is rough stuff, not rocket science, so a lot
of fancy math (like formal h-param matrix stuff) isn't justified.

I still remember that day (I must have been 11 or 12) when I finished
assembling my Knight-kit oscilloscope, the first DC-coupled scope I'd
ever had access to. I rigged it up to sweep the base current of a
transistor and plot Ic vs Ib. It was a straight line. Something dawned
on me that instant.

I remember younger days as well but I guess I have not faced the
required challenges.

Thanks.

JohnS

 

[[email protected]]

I did a search and found that you are correct about hfe/Hfe/hFE. I had

no idea that the capitalization meant anything. For all these years I

thought hfe was just hfe. My mistake.

There is no need to apologize and you didn't make a mistake because for practical purposes the ac-current gain and so-called static current gain are identical. This follows because IC in ratio to IB by factor hFE makes deviation in IC, ic, to deviation in IB, ib, that same ratio, of hfe=hFE. This is why the only distinction between hFE and hfe is one of analytical notation, and /also/ why the datasheet specification is so widely used for both. And this has nothing to do with hFE variation with VCE, Early effect, and level of current injection IB.

 

[[email protected]]

Phil Hobbs wrote:






























Re prototyping: Datasheets have deteriorated so very badly that prototyping is

more necessary now than it was 10 years ago, IME. JL mentioned the problem of RF

parts having very few DC specs, which is one I run into all the time. I also do

a lot of niche things, e.g. using an SA614 limiter/detector IC with its RSSI

output wrapped round to the gate of a MOSFET variable attenuator, to get wider

dynamic range in a tactical optical communications system for DARPA. Youcan't

do that based on data sheets, or even SPICE models--you have to prototype, and do

intelligent engineering characterization to make sure it'll work in the corner

cases, e.g. high temperature, strong signals, bright sunlight right in the field

of view.



So I do a fair amount of prototyping--overall, I probably spend almost asmuch

time on protos as on Spice, but less than I do on analysis or on actual

schematics.



But then I'm a physicist.



Cheers



Phil Hobbs

That's one reason the datasheets are becoming so scant on small signal infois because they provide a SPICE model. I don't see why you can't rig up partial SPICE sims to test out things like the front end attenuation control loop before you actually start prototyping, and this stuff about testing a candidate circuit against environmental/operating extremes can get well beyond what one would call prototyping.

 

[John S]

There is no need to apologize and you didn't make a mistake because for practical purposes the ac-current gain and so-called static current gain are identical. This follows because IC in ratio to IB by factor hFE makes deviation in IC, ic, to deviation in IB, ib, that same ratio, of hfe=hFE. This is why the only distinction between hFE and hfe is one of analytical notation, and /also/ why the datasheet specification is so widely used for both. And this has nothing to do with hFE variation with VCE, Early effect, and level of current injection IB.

I'm not apologizing as much for my mistake as I am for my attack on John
L. It was uncalled for considering my lack of technical knowledge in the
matter.

Thanks for your input.

JohnS

 

[[email protected]]

Hey, it's only a newsgroup. It's not life.

Maybe- but you tried to make it sound like the two current gains were distinct somehow.

 

[John S]

Hey, it's only a newsgroup. It's not life.

Thanks, John. That was very gracious of you.

Cheers,
JohnS

 

[[email protected]]

If you mean hFE and hfe, they are different. It's rare to see both on

a data sheet, but when you do, they are often different.

Oooh boy, here we go again...different only in definition, but identical numerically.

 

[Jamie]

Oooh boy, here we go again...different only in definition, but identical numerically.

of course
$68,$46,$45 with $68,$66,$65

It's obvious!

Jamie

 

[Jamie]

Yo' mama. (How's that for American?) ;0

Cheers

Phil Hobbs

Watch it now, they may become contagious.

Big brother "OBumma" is watching, Can you dig it!

Jamie

 

[rajibbandopadhyay]

.

Phil wrote a great book on electro-optical design, with a goodly bit of
low-noise electronics included.

....
I bow in deference
....

He also serves as our reference LPTM (Large Physicist Test Mass).

Truly, I believe the entrepreneurs and physicists like Phil H, John L,
Jim T, Tim W, et al, should consider organising Apprenticeship programs
(like in the olden days of Tycho Brahe - Nicholas Copernicus type) so
that learners could do hands-on training without filling their minds with
unnecessary academic hodge-podge.
I say this from an Indian, and a personal, perspective.

 

[[email protected]]

They are distinct. If you try building amplifiers that really need to

be linear over wide ranges of collector current, the distinction is

vital.



One of the major limitations of laser noise canceller performance at low

frequency is beta nonlinearity, i.e. 1/h_FE - 1/h_fe.



Cheers



Phil Hobbs

The key condition to your observation is range of collector currents, usually decades. Obviously hFE=hfe for infinitesmal excursions in operating current. But for large signal, you not only have hFE<>hfe but significant variation in each gain with current. That's why they have things like feedback, feedforward, push-pull and a bunch of other techniques.

 

[josephkk]

Can I have the current picture in the developed countries on h-
parameters? I believe, with so much computing power at their hands,
people should not use h-parameters nowadays. They have no reason to. In
case of complex circuitry, you could do simulations, and in practice, you
use ICs in place of transistors. what is your experience with hybrid
parameters? Could you discuss this with people in the univ. laboratories?

I would mention an important aspect which intrigued me
while I was studying in Univ. The theory books put so much emphasis on
hybrid parameters, but to me, they are good for later analysis, i.e.,
not to invent something new, but when invented, they try to find how
the system work, not for learning electronics principles. And also, in the
practical world I did not come across specific cases of using
h-parameters. I thought - such wastage of time and effort!

While i have used h-paramters for design, it is very uncommon. It can be
useful for composing sectional (or block) designs. Not all that useful
for designing less than four transistors worth. H-parameters are sort of
a co-delopment with the Eberly and Gummel-Poon physics based models
(SPICE) to some extent. See if you can find a GE transistor manual 7th
edition (best of breed) for a helpful discussion about various early
transistor amplifier models.

?-)

 

[[email protected]]

No, hFE is a DC measurement (Ic/Ib) and hfe is small-signal, the slope

of Ic/Ib at some operating point. They are usually similar, but on

data sheets that show both, they are often numerically different.

I don't need you reading me EE 101 on the subject. The hFE is usually a min-max range with guardbanding and, hopefully a 'typical' hFE variation with IC,Q. The hfe whenever presented are usually typical, at fixed IC,Q and VEC,Q and freq, and should be larger than hFE at Q-point because it doesn't have to contend with that small cut-in region where hFE trends to 0.

But since nobody here uses small-signal h-params, it doesn't matter

much.

Dunno how you even select a transistor for a non-trivial app without at least mentally computing a hybrid-pi. The BE diffusion capacitance can be a biggie with finite source impedance drive, and, God forbid, you might even use Miller's Rule here and there to do a quick estimate of bandwidth and risetimes.

 

[[email protected]]

While i have used h-paramters for design, it is very uncommon. It can be

useful for composing sectional (or block) designs. Not all that useful

for designing less than four transistors worth. H-parameters are sort of

a co-delopment with the Eberly and Gummel-Poon physics based models

(SPICE) to some extent. See if you can find a GE transistor manual 7th

edition (best of breed) for a helpful discussion about various early

transistor amplifier models.



?-)

The h-parameter models combine trivially for series-shunt feedback configurations.

 

[josephkk]

On Mon, 15 Oct 2012 12:06:15 -0400, Phil Hobbs

On 10/15/2012 11:57 AM, Jim Thompson wrote:
On Mon, 15 Oct 2012 11:45:27 -0400, Phil Hobbs

[snip]

And h parameters don't actually simplify anything, anyway. They were
originally touted as a way to get round the measurement problems of y
and z parameters, namely oscillation on the one hand and RC rolloff on
the other.

But all they are is a way of organizing a whole two (2) linearized,
frequency independent approximate equations for a transistor circuit
that is far better described by hybrid-pi. The labour saving is
trivial, far less than the effort required to calculate the stupid
things in the first place.

Honestly, folks, is h_re really more useful than VAF? VAF and C_CB are
actually a decent first cut at predicting low frequency
reverse-transfer behaviour over a wide range of conditions, which h_re
isn't.

H-parameters are textbook curiosities.

Cheers

Phil Hobbs

So are PhD's ;-)

Seriously, VAF is just a parameter in another STYLE of modeling. And
math is math is math. "Textbook curiosities" tweak the mind to think
of alternate solutions to untenable problems.

...Jim Thompson


Sure thing. But it's a reasonable approximation over a wide range,
because it's physics-based. h_re depends linearly on collector current,
for a start.

And at least I'm not in a _pathology_ texbook.

Cheers

Phil Hobbs

h-parameters can be complex math expressions, not just fixed numbers.

...Jim Thompson

Sure again. But then where's the benefit? A 2x2 matrix written out
with all sorts of algebra and stuff sort of defeats the point of the
exercise--it's far clearer just to write it out the usual way.

Cheers

Phil Hobbs

The 'benefit' is the ability to easily model things that don't yet
have models. Let the simulator cope with the Algebra.

...Jim Thompson

Hmmm. Perhaps a lot of "behavioral models" come from sloppily executed
h-parameter versions of ICs.

?-)