BJT Biasing

[Faber]

Okay, so I've been looking through the BJT biasing stuff and I've been wondering about the load lines from datasheets.
I think I understand what to do with them (draw the line), but then I designed and breadboarded a pedal with random Ic and it worked just fine.

I don't know if that was just a fluke or if the load lines are as useless as Hfe.

Help?

[alanlan]

The use of a load line superimposed on the output characteristic of a BJT can be useful to get a graphical view of how Vce will change for a change in Ic (in turn caused by a given change in Ib).  You've probably seen someone in a text book somewhere draw a sine wave on the Ib lines and draw across to where the peaks of the sine wave meet the load line and then draw down to intersect the Vce axis.  All very complicated.

Other than that, I've never found them to be an essential tool.  I think it is more helpful to consider the DC bias conditions with a bit of simple maths and then look at the AC conditions separately.  I don't think a load line is essential for any of that (in fact when I see a load line I have to think hard what it is trying to tell me - it just doesn't appear to be all that intuitive but it could just be me).

[R.G.]

Okay, so I've been looking through the BJT biasing stuff and I've been wondering about the load lines from datasheets.
I think I understand what to do with them (draw the line), but then I designed and breadboarded a pedal with random Ic and it worked just fine.
I don't know if that was just a fluke or if the load lines are as useless as Hfe.

BJT biasing and you breadboarded a pedal with a random IC? Huh?

Neither load lines nor hfe is useless. They're both extremely valuable - for what they do. But neither one is a magic key to how circuits work, either.

[frank_p]

Mr. R.G.
Talking of load lines and BJT curves, I saw that the link on GEO on curve tracers is not opening anything.  I was looking what was a OK solution to make a curve tracer (?)

This is the name of the link that don't work on GEO:

Basic Testing of Semiconductor Devices and other information including Introduction to Curve Tracers.

thanks

[Faber]

I mean collector current, sorry for the confusion.

Yeah, I just picked a random collector current and Vce and it worked.  (I think it was like 1mA and 4.3V).  It turned out to be a pretty good boost.

About Hfe being useless, I remember the thread about BJT biasing that was extremely informational that said if you make the divider current? 10x the emitter current, then Beta becomes useless.  Maybe Beta was the word that I was talking about.

Hfe = Beta?  Maybe that's another place I'm going wrong.

So load lines are not useless; is it better to pick a random Ic and Vce and get lucky or use the load lines?  (I like and use R.G.'s way of setting the drop in Rc, Vce, and Re first.)  Also, when I look at the load lines, there's only Vce for up to 1V, and I need them up to the 4.5V for 9V batteries.  That confuses me too... 

[brett]

Hi
I'm no expert, but here's a couple of thoughts of a beginner anyway.

Yeah, I just picked a random collector current and Vce and it worked.  (I think it was like 1mA and 4.3V).  It turned out to be a pretty good boost. 

There are also other combinations that wouldn't work well, or would work better. 
One reason why Q1 in a fuzzface gives pleasant distortion is that it it has Ic of about 0.02 mA.  Examining the load line for 0.02 mA tells you why.

About Hfe being useless, I remember the thread about BJT biasing that was extremely informational that said if you make the divider current? 10x the emitter current, then Beta becomes useless.  Maybe Beta was the word that I was talking about.

Hfe = Beta?  Maybe that's another place I'm going wrong.

hFE equals Beta sometimes.  Gain in many circuits (common emitter circuits at least) is the lesser of Rc/Re and hFE.  Again, taking Q1 in a fuzzface as an example, there is a low effective Re (there is no resistor on the emitter, so the resistance is only that of the base-emitter junction diode).  Rc is a whopping 33k.  In this case hFE is much less than Rc/Re, and consequently gain = hFE.

cheers

[R.G.]

The real skinny on this is that (a) competent (b) commercial (c ) industrial practice is to design circuits in such a way that the highly, highly variable hfe, HFE, beta, etc do not matter over a wide range of variation. The fallacy with thinking in terms of hfe and load lines on characteristic curve diagrams is that both hfe and the curves vary so much that it's impossible to use just one chart. You have to do at least min, max, and typical/average/nominal charts to get a sense of what is going on.

Industry realized a long time ago that it's much more cost effective to demand that your engineers design circuits that are as immune as possible to device variations. The common four-resistor BJT circuit with a base divider and both emitter and collector resistors is one of those immune-seeking circuits.

We used to have to take a transistor circuit and compute the sensitivity of its bias point, gain, etc to the component values as they changed and also to the variation of the transistor parameters. Sensitivities of 1:1 or worse (i.e., what you want varies exactly as much as what you can't control)  are bad indeed. Sensitivities of less than one indicate that what you want varies less than what you can't control. Big raises all around and no one works overnight tweaking things on the production line.

Effects are a mixture of such careful design and seat-of-the-pants (i.e. uninformed) design, plus design for side effects. The FF is a design for side effects.

[Faber]

And I realized that in my other post I said that Id needed to be bigger than Ic, then I read in an old college textbook that Id needs to be 10x bigger than Ib, which I read to be Ic/Beta.  But since Beta is so random for a set of transistors, how do I get around that like you were saying R.G.?

Thanks for all the help! 

[alanlan]

I think in these cases it is common practice to make certain assumptions.  i.e. think about what could be the worst case DC bias current into the base.  This will be the steady state/quiescent/DC collector current divided by the worst case (lowest) hfe.

So, look at the datasheet to determine the minimum hfe and use that figure to work out the maximum base current for your required collector current.

Your bias divider current can then be set to 10X this.  This should guarantee that all will be well.

e.g. if you want your Ic to be 10mA and the datasheet says that the minimum hfe = 100, then your worst case base current will be 100 uA.  Your divider current in this case would be set to 1mA to ensure that the bias voltage stays pretty rock solid.

It's worth pointing out that there are plenty of other ways to get a stable bias; here I'm only considering the potential divider method (i.e. as used in the four resistor circuit R.G. mentioned).

[Faber]

Okay, cool that makes things much easier.  Thanks again!!!   

[Gus]

Look at the input section of the Hot Si.  Controlled gain and stable bias and OK input resistance.

[demonstar]

The real skinny on this is that (a) competent (b) commercial (c ) industrial practice is to design circuits in such a way that the highly, highly variable hfe, HFE, beta, etc do not matter over a wide range of variation.

I have lost track of how many times I have seen this typed here by you even in the short time I have been here.

[frank_p]

Mr. R.G.
Talking of load lines and BJT curves, I saw that the link on GEO on curve tracers is not opening anything.  I was looking what was a OK solution to make a curve tracer (?)

This is the name of the link that don't work on GEO:

Basic Testing of Semiconductor Devices and other information including Introduction to Curve Tracers.

thanks

I think I found it on an other site:

http://repairfaq.ece.drexel.edu/sam/semitest.htm

[frank_p]

It's the same link as the link just over it (at GEO).  So I presume it was the same content for the too links (?)

[DougH]

Effects are a mixture of such careful design and seat-of-the-pants (i.e. uninformed) design, plus design for side effects.

This is the problem with trying to learn electronics from an effects board.

There's a lot of "wild west shoot from the hip" B.S. going on with fx design- some of it is intentional to achieve a certain response- some of it is not and is just poor design. If you can't discern that, a lot of bad habits, fallacies, and otherwise just wrong thinking gets promoted, adopted, and accepted as "truth". Which is exactly what is going on throughout a lot of the "bo-tweak" industry.