for *I*_{B} = -10 µA, -20 µA, -30 µA, ... , -80 µA

big data file, smaller data file, postscript plot, pdf plot

Note that you can make the PNP characteristic curves look like the common NPN curves just by rotating the PNP plot by 180°. In swapping N for P so NPN PNP, we've reversed the direction of current flows (so currents are negative -- flowing out of the collector and base in a PNP) and the required supply voltage becomes negative for a PNP.

That is to say that the PNP is designed for negative power supplies and out-flowing (negative) base and collector currents -- the opposite of NPNs.

The behavior of a PNP bipolar transistor is largely controlled
by the current flowing *out of* the base.
For the usual collector-emitter voltage drops (i.e., the __active region__:
negative voltages from a fraction of a volt down to some breakdown voltage) the collector
current (*I*_{C}) is nearly independent of the collector-emitter voltage
(*V*_{CE}), and instead
depends on the base current (*I*_{B}).
(This is unusual behavior: usually more voltage
produces to more current, but here the current only increases
slightly with increasingly negative *V*_{CE}.) The __current gain__,
i.e., the ratio of the collector current to the base current, is
often denoted by or *h*_{FE}:

= *h*_{FE} =
*I*_{C}/*I*_{B}

Thus in the simplest approximation the characteristic curves of a PNP are a set of flat, evenly spaced, lines:

Each (flat) curve shows that *I*_{C} doesn't change with changing
*V*_{CE}. The different levels show that *I*_{C}
does depend on *I*_{B}.

A slightly more complicated approximation takes into account
the sloping characteristic curves through a constant Early
Voltage (*V*_{A}). Here we assume that the
characteristic curves all have a common *x*-axis intercept
at the large positive voltage *V*_{A}. (The dashed curves
are far from the active region and in no way represent the actual behavior of
the transistor for positive *V*_{CE}. In fact, the PNP transistor
is not designed to be operated with positive *V*_{CE}.)

(For the above measured 2N3906, the Early voltage ranges
from 40 to 50 V. That is the extrapolated characteristic curves do not
intersect at *a* point.)

The actual relationship between the collector current (*I*_{C})
and the controlling base current (*I*_{B}) and
collector-emitter voltage drop (*V*_{CE}) is some complicated
function which we can denote:

*I*_{C}(*I*_{B},*V*_{CE})

Like any function we can approximate it near a particular point using just the first terms of a Taylors expansion:

Clearly these *hybrid (h) parameters* are not constants.
For example *h*_{oe} is the slope of a characteristic
curve, which is nearly zero for small |*I*_{B}| and increases
for more negative *I*_{B}. (Note that slope on an *I-V*
is basically the inverse of the resistance. Thus 1/*h*_{oe}
can be described as the output impedance. A typical value for
1/*h*_{oe} would be 100,000 .)

Even the defining parameter -- the current gain
or *h*_{FE} -- is not exactly
constant and depends on *I*_{B}. Here is the measured
relationship for the above 2N3906:

The characteristic curves focus on the output of the transistor, but we
can also consider the behavior of the input. In the active region the base
is a forward biased diode, and so *V*_{B} would be about -.7 V,
typical for a conducting Si diode. Of course in greater detail the relationship
between *V*_{B} and *I*_{B} would be given by the
Shockley diode equation:

where *I*_{s} is a constant and the thermal voltage
*V*_{T} is given by:

Because of the exponential relationship between base current and base voltage,
the slope of this relationship (which could be called the
input conductance, or 1/*r*_{B}, or 1/*h*_{ie}) can be
approximated by:

where in the last equation (25 /*I*_{B})
the absolute value of the base current must be entered in mA.

A desirable characteristic of a transistor is that the outputs have little
effect on the inputs, but if we look in detail we find that *V*_{CE}
affects *V*_{B}. The actual functional relationship giving the
base voltage from
the base current (*I*_{B}) and
collector-emitter voltage drop (*V*_{CE}) is some complicated
function which we can denote:

*V*_{B}(*I*_{B},*V*_{CE})

Like any function we can again approximate it near a particular point using just the first terms of Taylors expansion:

The small value of *h*_{re} shows that the input is
largely unaffected by the output.

The spec sheet reports the following values for the 2N3906:

Characteristics | Symbol | Min | Max | Unit |
---|---|---|---|---|

Input Impedance | h_{ie} |
2 | 12 | k Ohms |

Voltage Feedback Ratio | h_{re} |
0.1 | 10 | ×10^{-4} |

Small-Signal Current Gain | h_{fe} |
100 | 400 | |

Output Admittance | h_{oe} |
3.0 | 60 | µ mhos |