What's All This hFE Stuff About?

We see lots of discussion here and on other forums regarding hFE, but I get the feeling that many of us don't really know what it means or why it's important.

First, a little background. Bipolar transistors are current-mode devices. We have to supply base current to get collector current flowing (turn the transistor on). We vary the base current to modulate the collector current and achieve amplification. The definition of hFE (DC current gain) is the DC collector current divided by the DC base current. Notice that the "FE" is upper case, denoting DC conditions. hFE is important for setting the bias on a transistor. The definition of hfe (small-signal AC current gain) is the change in collector current divided by the change in base current. As an example, if a 5uA change in base current causes a 300uA change in collector current, then the hfe is 300/5 = 60. The "fe" is lower case, denoting small-signal AC. If we draw a graph of collector current vs. base current, hfe would be the slope of the curve at any given collector current. hfe is important for determining how much amplification we can get from a transistor, but it's not the only factor. hFE (and hfe) vary considerable from one part number to another and from one manufacturing run to another. The datasheet for 2N3904 says the hfe at 1mA collector current will be anywhere between 100 and 400. This kind of tolerance is typical.

Even for the same transistor, hFE and hfe are not constant, they vary with collector current and how they vary depends on how the transistor was designed and manufactured. When someone says a transistor has an hFE of 100, that means at some collector current the base current is 1/100th of the collector current. With the types of transistors we use and the current levels we have in pedals, when collector current goes up, hFE goes up, but it's not proportional. All this means you can have a transistor that has an hFE of 100 when the collector current is 1mA, but when you run it at 200uA, the hFE will be lower. It's difficult to predict the hFE at various collector currents, so designers depend on testing. Depending on the circuit, cherry-picking transistors may be necessary.

When someone recommends a certain hFE or hfe, it's really just a guideline to get you in the ballpark. Some circuits require hFE to be within certain limits to bias correctly. Tone Benders and Fuzz Faces for example. Other circuits are deliberately designed to not be sensitive to hFE or hfe. The Big Muff Pi is an example of this. Local feedback around each transistor in the BMP stabilizes the bias point and sets the AC gain.

At the beginning of this article, I said we have to supply base current to get collector current flowing. That's only partially true because many germanium transistors, especially vintage ones, are leaky enough to supply their own base current. There's still base current, but it's coming from inside the transistor instead from the circuit outside the transistor. All three stages of the Maestro FZ-1 (and its variants) are examples of leakage biasing. This leakage current makes measuring hFE more difficult because we have to account for the effects of leakage current. DMMs that measure hFE usually do not account for leakage current and as a result, grossly over-estimate hFE. Real transistor testers, like the DCA55, make the measurement correctly.

Variability in leakage and hFE are why some modern germanium fuzz circuits have trimpots in them. When Steve Daniels at Small Bear sells a matched set of Tonebender or Fuzz Face transistors, he also includes the biasing resistors for that set. Those resistors are part of the matching and accommodate the part-to-part variability in the transistors.

So how do you know you have the right hFE (or hfe)? Follow the BOM in the Build Docs, buy quality transistors and cross your fingers. If you're substituting transistors or building with germanium, use sockets and plan on trying a few different ones.

Next time: The Other Way to Describe Transistor Gain.
 
Last edited:

SYLV9ST9R

Well-known member
Thanks Chuck! This is a great initiative for people like me who like to build, but only have a basic grasp of this stuff. Will make for nice reading material during the winter months here.
 

Chas Grant

Well-known member
Well as an electronics expert I know that DC = District of Columbia and AC = Air Conditioning, but what does hFE (or hfe) stand for? And speaking of BJT current ratings, how much can a bipolar bear?
If I come across a bipolar bear, I'm running!!
Bipolar.jpg

But to answer your question, it all depends on the transistor! Continuous collector current on a 2N5088 is 100mA, where as on a a big power transistor like a MJ15003 it is 20A. Both are BJT, one can be used in a pedal, one can be used in the power stage of a solid state amp. So it depends on its ratings and what its designed for.
 
Last edited:

jubal81

Well-known member
Dang, Chuck. You're pretty good at explaining this stuff so pedal builders can understand.

So, as far as using BJT as small-signal audio amplifier:
1. Can it be assumed that most typical silicon transistors we use will perform very similarly, regardless of part number, as long as the amplification factor is set up to be less than the hfe?
2. Once the gain factor exceeds the hfe, does it matter by how much?
3. Would it be bitchin' to bias a BJT fuzz with a servo and control the gain by adjusting the voltage supply/headroom?
 

Dan0h

Well-known member
Wow dude, its almost like you are reading my mind. I have been scouring the internets trying to decide if I should pay up for the OC71 to use in the Duocast build, or buy a hand full of "substitutes" and try them out with sockets. Its amazing to me that pedals are not all over the place when it comes to sounds coming from them. I wish there was a pdf that listed all the vintage parts and all substitutes that would work, but then you are dealing with personal taste and everyone's ears are different, not to mention taking the fun of the hunt out of the game.
 

Danbieranowski

Well-known member
My question is why are hfe/hFE tolerances so broad? Is this a "nature of the beast" type situation that's just inherent with transistor technology? Or is it that the way they are manufactured is cheaper by some companies (thus wider tolerances), but tighter for higher cost companies?
 

Chuck D. Bones

Circuit Wizard
My question is why are hfe/hFE tolerances so broad? Is this a "nature of the beast" type situation that's just inherent with transistor technology? Or is it that the way they are manufactured is cheaper by some companies (thus wider tolerances), but tighter for higher cost companies?

Parametric variation is inherent in the manufacturing process. Sometimes an entire production run of parts is rejected at the factory. How much tolerance is acceptable is a business decision on the part of the semiconductor manufacturer and the OEM (original equipment manufacturer). The parametric variations are loose enough to ensure a good yield and tight enough to work in the circuits. When an OEM requires tighter tolerance, they can pay extra for the manufacturer to cherry-pick parts for them, or buy a larger quantity than is needed and cherry-pick them themselves. Places I've worked have done both, depending on the parts. Again, it's a business decision. Some pedal builders actually perform the necessary engineering and design pedals that are not dependent on tight tolerances to work. Some cherry-pick parts. Others apparently just slap shit together, eat the rejects and hope the rest don't get returned. That behavior is more common than you might think and extends far beyond the guitar pedal industry.
 

Chuck D. Bones

Circuit Wizard
Wow dude, its almost like you are reading my mind. I have been scouring the internets trying to decide if I should pay up for the OC71 to use in the Duocast build, or buy a hand full of "substitutes" and try them out with sockets. Its amazing to me that pedals are not all over the place when it comes to sounds coming from them. I wish there was a pdf that listed all the vintage parts and all substitutes that would work, but then you are dealing with personal taste and everyone's ears are different, not to mention taking the fun of the hunt out of the game.
Some of them are all over the place. Legend has it that Hendrix would have his tech buy several Fuzz Faces just so he could audition them and hopefully find one or two that sounded good to him. Parts substitution is very circuit dependent, will probably involve cherry-picking, and as you said, there is the personal taste factor. In other words, don't hold your breath for a germanium transistor substitution guide.
 

Chuck D. Bones

Circuit Wizard
Dang, Chuck. You're pretty good at explaining this stuff so pedal builders can understand.

So, as far as using BJT as small-signal audio amplifier:
1. Can it be assumed that most typical silicon transistors we use will perform very similarly, regardless of part number, as long as the amplification factor is set up to be less than the hfe?
2. Once the gain factor exceeds the hfe, does it matter by how much?
3. Would it be bitchin' to bias a BJT fuzz with a servo and control the gain by adjusting the voltage supply/headroom?
1. & 2. Amplification factor (voltage gain) does not correlate with hfe. Stand by for an article on that topic. And yes, a great number of the Si trannies we use are interchangeable. There are a handful that I prefer and will work in most circuits: NPN: MPSA18, 2N5089, 2N5210, BC549C, BC550C. PNP: 2N5087, BC560C. They are all high hfe and low noise.

Regarding #3, the FZ-1 throttles Q2's collector current to vary the gain and clipping threshold. Moving the supply rails up and down may or may not affect the gain, but will definitely affect the headroom.
 

Chuck D. Bones

Circuit Wizard
I know of two good ways:

Analytical - you analyze the circuit using a simulator or some other engineering tool and look at what effect varying the hfe has. In many pedal circuits, it will become evident that hfe has little or no effect on performance unless the hfe is so low (<25) as to make the transistor useless.

Empirical - you build the circuit, either on a breadboard or on a PCB with sockets and you try different transistors. This method has two big advantages. 1. You don't need to know how to run a simulation and interpret the results. 2. You get to hear the results. The only downside, if you can call it that, is you need to have a good collection of transistors. Serious pedal builders already have that.

There is a third way: ask someone. The main problem with that is their idea of what sounds good might not be your idea of what sounds good. The other problem with this method is they might be completely full of :poop:. Either scenario, you'll have no way of knowing you received good advice until you build the circuit and hear it for yourself.

I do the first two, and I listen to other peoples' opinions with a healthy dose of skepticism.
 

jubal81

Well-known member
Spitballing here, but I'm wondering about finding a way to use low-gain/leaky germaniums and if pairing them up as darlington or Sziklai is an option worth pursuing.

I recently did up a layout for the "Skreddy trio" and built a hybrid fuzz driver because I've got a handful of NPN Ge's that have been laying around for years I wanted to use up. I'd thought all this time they were pretty good, but turns out they're turds.
 
  • Like
Reactions: fig

Chuck D. Bones

Circuit Wizard
My experience with the HFD is that the bias of the last stage is critical to getting the tone right, particularly the note decay. I think you'll have better luck using various Ge trannys there if you dial-in the bias. Once again, I believe it's more about leakage than HFE because the 2nd stage can hit the 3rd stage plenty hard to drive a fairly low HFE tranny into saturation. When I built my HFD, I socketed the 3rd stage transistor. First, I installed a 2N1308. Low leakage and HFE around 150 as I recall. I could not get it to sound the same as a real HFD. I could get close, but required different control settings. Then I replaced the 2N1308 with an AC127 (similar HFE, but more leakage) and the sound was dead-on with nearly identical control settings. I checked the bias with each transistors and, not surprisingly, the AC127 ran at a higher collector current and lower collector voltage due to the higher leakage. Anyone who has fiddled the bias trimmer on a FF knows how much difference it makes.

Ge trannys have been used in Sziklai pairs; take a look at the Small Bear Oh My Darlington or the Human Gear Animato (Aion Polaris) for examples. Here's a good article on Sziklai pairs.

A good use for low-gain transistors is as clipping diodes. Just be careful they're not too leaky.
 
Top