Very low current in amplifiers

Well beta drops at low current, so that would be an issue. But the beta roll off is a function of current density rather than strictly current. That is, a very small BJT could have decent beta at low current levels. Are we talking discrete or IC design? WIth ICs, there is parasitic diode leakage to consider. Anyway, Ie under a microamp isn't exactly groundbreaking in IC design, but discretes are targeted for higher current applications.

The only micropower bipolar design I've done has been in bandgaps. Probably if you researched hearing aid papers, you'd get your answer.

Yes and no. In general, drive capability increases with higher bias currents, but with BJTs noise is lower at low bias currents. This is why the input stage of low-level preamps are often operated at sub-mA currents, but there's a catch. Increasing the resistors also introduces increased thermal and other noises originating in the resistors. A good design will set the desired bias level without excessively increasing the resistor values.

From the datasheets I've looked at hfe doesn't drop that much for low currents at around 100uA. It seems hfe is affected more by high current and/or low Vce than low Ic. This is for discrete application. In my case I'm using a jfet which I'm not sure if it matters. Simulations in multisim show my circuit works fine but I'm not sure if the models are effective for sub-mA drain currents or not.

Yes, of course, this is why I mentioned the issue with noise. I don't believe i'll have any drive issues since I can buffer the signal. The noise may or may not be an issue but I don't believe it will. The main issue I need to get over is if there is a point where the device starts to behave very differently when there is not enough current(excluding noise). What if I were using fA's and assuming everything else was ideal, could I expect to have a reasonable audio amplifier? Or is there some point where the transistor simply shuts off completely or goes bonkers and requires a completely different model? Again, excluding thermal noise and random fluctuation. fA's maybe to small to conceive but I'm trying to understand the limiting case.

The dynamic range is likely to drop precipitously.

Also stage gain is often a strong function of collector current.

These games were played with vacuum tubes where they were called "starved current amplifiers"

--
Many thanks,

Don Lancaster                          voice phone: (928)428-4073
Synergetics   3860 West First Street   Box 809 Thatcher, AZ 85552
rss: http://www.tinaja.com/whtnu.xml   email: don@tinaja.com

Please visit my GURU's LAIR web site at http://www.tinaja.com

Bandwidth will almost certainly be ~300 lower. The stray capacitances will stay the same and you will have ~300 times less current to charge and discharge them.

The loss of bandwidth is the primary reason. Leakage currents become more of a nuisance and you have to be more careful about what happens when the circuit gets warm, but the fact that the circuit becomes glacially slow is usually the biggest problem.

-- Bill Sloman, Nijmegen

On a sunny day (Sun, 24 Jul 2011 14:18:13 -0700 (PDT)) it happened Jenkins wrote in :

You get high frequency roll of due to the RC constants being bigger. So gain decreaes at high frequencies. Beta depends on Ic too.

There is a limiting case related to the number charges in a coulomb, = about

6e19. You need enough charges per second to adequately produce the desired waveform. This does seem to be well past reasonable expectations of device behavior repeatability. Does this help?