Tim Williams
Normally we operate BJTs with the collector high, and base voltage variable (and current-limited), relative to the emitter (for expedience, I'll use NPN polarity here; for PNP, everything is flipped, plus for minus!). This gets us the common set of circuits, for power/logic switching, current source/mirror, amplifier, clamper, and so on, and even analog switching with some effort (if somewhat approximate).
Inverted Mode
What happens if collector voltage goes below emitter? Recall that an N-P-N transistor reads the same backwards (the difference lies in the details); indeed, it still works, albeit not usually as well. This is inverted mode. Besides having lower "VCE", hFE is usually quite small -- a typical 2N3904 might hit 5 or 10. Both are a consequence of the doping levels being backwards; the emitter "emits" better when it's highly doped, and high doping level causes a narrow depletion region, hence the low breakdown voltage (typ. 6-8V).
Transistors have been designed for inverted mode operation. 2SC2878 (obsolete/vintage) and 2SD2114K (a rather tasty little transistor in its own right) come to mind. Typically, VEB is higher, and some effort has been made to keep reversed hFE comparable (typically still lower than forward, but manageably so). They aren't usually specified in terms of inverted characteristics, but a resistance spec is added, which one can assume is an acceptable ohmic approximation (i.e. it has low distortion while saturated).
Another consequence of inverted operation: the built-in potential due to the doping gradient (heavier in emitter, lighter in collector), while it doesn't manifest as free energy (of course), it does affect Vce(sat). 2N3904 might (forward) saturate at some 10s of mV, a respectable level, but doesn't go any lower, even at very small IC, or large IB (indeed, IB eventually works against you due to emitter resistance). Inverted or "VEC(sat)" can be single mV. In past centuries, engineers used this to make high accuracy D-A converters, before integrated versions (or MOSFETs) were developed. (JFETs were a competitive option, which might be chosen for or against based on other factors like switching speed or charge injection.)
Read more »
Caleb W.
BleakyTex
jbb
Thanks for that comprehensive article. I used to use reverse biased BE junctions as cheap zener diodes, particularly if the gain was low or the transistor was otherwise substandard, because they were much cheaper than real zeners. A drawback was that you had to do manual characterisation and selection. There wasn't much call for these ersatz zeners after integrated regulators became cheap enough.