BJT with hight Hfe at very low Ic

As you are aware of, Hfe drops at low Ic current. Unfortunately this is not specified at very low Ic.
I have in mind bjt drivers in Darlington or Slizard configuration for low power.

Do you know BJTs that do not drop much ? Is there a process that gives not much Hfe droop at very low current.
 
Can you be more specific? How low is low Ic? A pretty generic KSC1845 maintains gain down to 10uA which is the left edge of the graph I have for it. The graph does not tail off rapidly, so I don't know how low it will go and maintain beta. At Ic=10uA, beta is shown as 500. The graph did not indicate which gain class the graph was referring to.
 
  • Like
Reactions: mchambin
In the sixties I fiddled with slizard pairs for phono input. Testing with a test vinyl I had noticed best results in the bass with very large input impedances.
I remember, this was highly dependend of the driver bjt. I had my favorite for this. I do not remember what part, germanium I presume.
Writing this, I remember this equipement has been saved from the trash bin. I will look at what i used for stereo phono input.
It could very well be, this bjt had little beta droop at very low Ic current. I’ll see, since I know where to find two of them.
 
Good process can reduce some of the drop in beta at low current, but the base-emitter recombination effect is fundamental (plotting base and collector currents against Vbe shows that at low current levels the base current rises w.r.t. what you'd expect from the diode equation. Since these effects depend on current density, its best to pick a very small transistor (die size isn't usually given in datasheets, but max collector current is, and is a plausible proxy for die size).
 
  • Like
Reactions: mchambin
Transistors made using modern processes are very good regarding low current beta. Even S9012,13, 2N3904, 6 from Chinese origin (dirt cheap) are excellent in this respect.
Earlier, the transistors were plagued by contamination, surface currents, etc. but now you can pick any recently-made transistor and be confident about its performances at sub-nA Ic levels.
A good indicator is the behaviour of Vceo and Veco: earlier types showed a clear snap-off characteristic (ie. negative resistance) at ~80V and 8V respectively, but recently made types simply behave like zener (or avalanche) diodes , with no hint of negative resistance.
That is a clear sign of low leakages, and thus good behaviour at low current
 
  • Like
Reactions: mchambin
Can you be more specific? How low is low Ic? A pretty generic KSC1845 maintains gain down to 10uA which is the left edge of the graph I have for it. The graph does not tail off rapidly, so I don't know how low it will go and maintain beta. At Ic=10uA, beta is shown as 500. The graph did not indicate which gain class the graph was referring to.
Thanks, 10uA Ic is the low I am looking for.
And wondering about, in the nA realm.
 
At that time ( pre 1966 ) I hardly knew what I was doing but dared anything. My stereo system made from A to Z did work.
Then had decades of good service. May be not the best RAAI stuff. But a decent 2 x 10W rms class A amplifier.

I disgrested, my interest in very low Ic is about experimenting in topologies where I look for hight collecter impedance.
 
I probably grabbed the KSC datasheet from digikey. If you can't find one online, I can email the pdf for it. It is a low noise device with high beta like you want. It is as Mark mentioned, limited to a max current of 50ma, fairly low for even a TO-92 case device. It does have a pnp complement as well, KSA992.
 
  • Like
Reactions: mchambin
Thanks, I downloaded the two datasheets KSC1845 and KSA992, very interesting transistors indeed. Very affordable at Mouser.

I had not seen: Noise figure 7 dB. What does it mean ? Could this be translated in Rbb or better known noise figure.
This 7dB is specified for a 100K source. I found, the meaning is: The transistor adds a noise 7dB times the thermal noise of à 100k Ohm source.
Not doing any calculation that is obviously terrible noise.
I do not see the relevance of that 7dB at 100k.
I deal with way lower impédances ( like 150R ribbon microphone ) where serious ultra low noise transistors have rbb less than 10R. Low noise op amps have less than 5nV / sqrt Hz noise density
Do I missed something ?

BTW. I am grateful for all your inputs. I was not aware of transistors specialized at such low Ics. I was aware of progress in performances but not aware of that much.
 
Last edited:
Transistor datasheets used to have "noise contour" graphs like the one below for the 2N5087.

The combination of 100K source impedance with 1mA collector current results in poor noise performance. I am not sure why the KSC1845 was spec'ed that way. It can do much better at lower collector current.
Ed

2n5087-noise.png
 
Noise figure and contours plots really serve to obscure understanding I think - they can all be derived from noise spectral density specs of the transistors (nV/√Hz and pA/√Hz), typically the voltage noise is fixed, the current noise depends on the base current using the shot-noise formula, with the caveat that 1/f noise will appear at lower frequencies (hopefully the datasheet has a graph of this).

Noise figure in particular is unique to a bandwidth, current and source impedance, and no two datasheets are likely to quote it for the same conditions - the same transistor can have a noise figure of 1dB, 10dB or whatever in different situations, yet its noise spectral density graphs are fairly easy to compare (especially for voltage noise).
 
  • Like
Reactions: mchambin