Transistor gain and Data Sheets

[Brymus]

OK I was under the impression that the Hfe is the gain of a transistor.
But when trying to select different gains based on the data sheets, it shows the Hfe of the 3904 , 4401,  5088, to all be the same.
100-300 ?
I often see them referenced here as the >3904 low gain 4401 med gain and 5088 as high gain.

Would someone be so kind as to explain how to interpret the data sheets #s and graphs so I can do my own picking,based on what I want?
I know most of you know this already and I feel dumb just asking.
So your help would be very appreciated,thank you

[R.G.]

OK I was under the impression that the Hfe is the gain of a transistor.

It is. It's the *current* gain of the transistor measured under some kind of conditions, usually conditions that make it look best in the datasheet. This is in direct conflict with what we usually want - voltage gain. About all you can take away from hfe is that higher hfe offers the possibility to get higher voltage gain if everything is done properly. It's certainly not a guarantee.

But when trying to select different gains based on the data sheets, it shows the Hfe of the 3904 , 4401,  5088, to all be the same.
100-300 ?I often see them referenced here as the >3904 low gain 4401 med gain and 5088 as high gain.

A lot depends on the manufacturer. If you are looking at the data sheets provided by a second- or third-tier supplier, they may have pulled the "NTE" trick - well, we make these transistors and they kinda match the specs for all these transistors - let's just sell the same ones as whichever type number we want."

In general, a typical 2N3904 will have a higher current gain than a 4401, and a 5088 will typically be higher than a 4401 if all other things are equal. Of course, all other things are NEVER equal, so it varies. Not much help, I know.

Would someone be so kind as to explain how to interpret the data sheets #s and graphs so I can do my own picking,based on what I want?

The bottom-line suggestion here is something my circuit design professor drummed into my head: count on a minimum hfe, not the hfe you want. Then make the design work for any hfe bigger than that minimum. Hfe is too variable to rely on.

I know most of you know this already and I feel dumb just asking.

Not at all. I suspect that most of the people reading here would say that higher hfe devices give higher voltage gain, but not be able to say exactly how that relates.

[PRR]

> my circuit design professor drummed into my head: count on a minimum hfe, not the hfe you want.

All my prof's problems started "assume Hfe>50".

You don't need any Hfe to get voltage gain.

You typically want Hfe to raise input impedance.

In a cascade of identical stages, maximum voltage gain per stage is equal to Hfe. However we almost never have long cascades of identical stages. And the "max" is hard to reach.

In a single resistor-loaded stage, voltage gain has nothing to do with current gain. The maximum voltage gain is simply Vcc/0.030. Although if you also want voltage swing, you get less; and higher Vcc will run into Early Effect too.

> >3904 low gain 4401 med gain and 5088 as high gain.

No. They are low medium high CURRENT. 50mA, 200mA, 500mA.

Hfe falls at low current due to recombination, at high current because the junction chokes-up. The curve is a hill. The datasheet naturally takes a point at/near the top of this hill. '3904 Hfe is quoted for 10mA, '4401 Hfe is quoted for 150mA.

The 2N5088 datasheet cites Hfe for 100uA, 1mA, and 10mA; and none are "100-300". Are you looking in the right places?



Beware "typical" versus "minimum". The '4401 at 0.1mA is typical near 200, but the MINimum at 0.1mA (and Vce=1.0V) is just 20. If you design for 200, many of your products may work. If your product won't work correctly with a Hfe=20 part, too bad: the maker will not take your complaint, the sheet clearly shows that Hfe=20 at 0.1mA is possible. (Not that makers take complaints until you buy millions of parts; and in DIY if you -really- need Hfe it may be cheaper to test/sort than to try to buy high-Hfe parts.)

> most of you know this already

It took me many years to -understand- transistors. (And I still think they work by magic; just that I have a grasp of the magic-spells to predict performance.)

"Most" folks on most audio forums could not figure 9V across 4.7K to 1-digit accuracy.

[Brymus]

Thanks RG, PRR,
I guess I have the wrong data sheets as all of the ones I mentioned appear identicle except for the part #
The one pic you showed shows the 100-300 I mentioned.

I was reading up trying to understand,and got into how the electrons flow from/through the base out the collector,the base is about the thickness of a 1/10th of a lightwave
http://hyperphysics.phy-astr.gsu.edu/hbase/solids/trans2.html
I was reading at the link above maybe you can suggest someplace better?
Its quite daunting when I am just trying to figure out how to select different gain transistors for a pedal build,not really designing anything per say.
So I guess I need to figure Vbe and Vce using real numbers and then look at the data sheets?
Since I am doing this for my tonebender build I still want to look at the operational range when used  as an amplifier correct ?
Or are the Qs in the FF part driven into saturation?

Figuring out 9V across 4k7
P=V*I  = V to the second / I = I to the second *R
Ohms law to figure power dissipated in watts is that what you mean ?
Since you gave (V) and (R) ,I(myself) would then have to solve for (I) in order to solve for watts dissipated ,correct ?

[PRR]

> pic you showed shows the 100-300 I mentioned.

I hasty-assembled the sheet-clips so it may be unclear.

The 2N5088 snippet shows Hfe 300-900, 350-?, 300-? at different currents. Not the "100-300" you seem to be seeing... what sheets are you looking at?

> select different gain transistors for a pedal build,not really designing

But what "gain" do you select? Surely this is a "design" issue, in reverse.

Why do you think you should be selecting transistors?

What R.G. and I were taught is how most electronics MUST be done. Transistors come out of the oven with 10:1 variability: Hfe may range from 10 to 1000. They sort them into bins so they can charge more for the higher gain parts. But they don't want to wind up with hundreds of different order-codes. And the exact Hfe takes too long to measure. Hence the 2N508? batch is sorted 2N5088: 300-900, 2N5509 400-1200. You pay for 3:1 variability.

But your customer (the audio or pedal consumer) needs results within 20%, maybe 5%, of target. If one pedal has gain of 80 and another has gain of 120, two guys with the "same" pedal may notice the difference; two channels of stereo hi-fi ought to match better than 5% or the sonic image is lopsided.

The art of electronics design is very much about NOT needing any specific part-spec. This is a little easier when affordable gain is "high". The basic goal of any amplifier is gain higher than 1; if we get any gain, we can stack more stages to any result. I was told "assume Hfe>50". 50 is much-much higher than 1, We can design for substantial gain which hardly varies if the specific part is Hfe=50 or Hfe=500.

Hfe is current gain (not voltage gain). If we are told "assume Hfe>50" and 10% tolerance on final current gain, we design the whole stage for current gain of 5. If we get a Hfe=50 part, actual current gain will be 4.5; if we get a Hfe=500 part, actual current gain will be 4.95; if we get a Hfe=5000(!) part, actual current gain will be 4.995. This extreme 100:1 variability in transistors only gives 11% variation in final current gain. A gain of 5 per stage seems small, but 2 stages is 25 and 4 stages is 625. Or for parts Hfe=50-5000, 4 stages gives 410-625 current gain. If we need a whole Ampere to a speaker, a conservative 4-stage amplifier only needs 1.6mA-2.4mA input signal, which can be sucked from a strong preamp. (Actual loudspeaker amps tend to use a sharper pencil or more stages to keep input signal current down to microAmps, but that's an Advanced Topic.)

> suggest someplace better?

Many many people have learned from TAoE:
The Art of Electronics
Paul Horowitz , Winfield Hill
ISBN 0521370957
Yes, $75-$50 is a gaspy price. You won't find it a lot cheaper: it IS a good book and in high demand. It covers a lot of stuff you don't want to learn, and I think it skims a few things. But it seems to be the current go-to book for basic useful electronic theory.

I read this book for many years before I understood things:
Analysis & Design of Transistor Circuits
Cowles, Laurence G.
1966, so some of it is dated.... well, maybe not. Youse guys fool with Germanium. In 1966 we were just leaving Ge totally behind, but Cowles covers the Ico problem well. OTOH his treatment of Si devices assumes what we would call "modern" Si with negligible leakage, and he shows how this really simplifies design. abe.com shows copies at $10-$30, and well worth those prices.

[PRR]

> Figuring out 9V across 4k7
P=V*I  = V to the second / I = I to the second *R
Ohms law to figure power dissipated in watts is that what you mean ?
Since you gave (V) and (R) ,I(myself) would then have to solve for (I) in order to solve for watts dissipated ,correct ?

Someday I'll start a boot-camp, and teach you to spit-out "2 m A, almost!" before I finish the question.

Yes, I did not say what answer I wanted: current, power, color-code. That does depend on context.

But in small-audio context, we almost NEVER care about "wattage". The whole pedal draws maybe 9V 10mA or under 1/10th Watt, the smallest parts we can handle are 1/8th watt, nothing can melt.

OTOH, a lot of circuit operation is based on current ratios. Hfe is one ratio, albeit poorly specified. We'll often have a resistor in the base circuit and a resistor in the collector circuit. These define another current ratio. Oh, the exact ratio depends on voltages-- but these are clues too.

IMHO, being able to glance at a part and rough-guess voltage and current is essential to real understanding of circuits.

BTW: Power is Voltage squared divided by resistance. You can get it by solving for I and then for P, and this is the way when you also want to know I.

[Brymus]

Thanx again the data sheets I have were just downloaded from IDK where anymore,

"I think I need" different gains based on what others report giving the "best" results ,trying again to somewhat copy the results others have posted.

The fact that you and RG are the only ones to comment leaves me to believe that more people here "paint by numbers" than I originaly thought.
My goal is in fact to understand the circuits rather than just copy and measure the voltages ,ect
So I can bend them to my will and get the result I here in my head.
My usual method is to define the paramaters that dictate what controls the sound I am tweaking then to a-b it within a range and listen.
This is tedious but usually effective,however this is not the same as what you ,RG,Mark,Gus and all the others do,those who actually understand the fundamentals of the components and how they interact to give a desired result.

[R.G.]

IMHO, being able to glance at a part and rough-guess voltage and current is essential to real understanding of circuits.

It is, yes. But try explaining that to someone who has recently found an on-line calculator app at some obscure web site.