Early voltage

The Early voltage $V_A$ is a single parameter that measures how flat a transistor’s output current curves are in its amplifying region — and hence how good a current source the device is. It is shared between the MOSFET and the BJT, with the same role in both.

For the MOSFET it is the reciprocal of the Channel-length modulation parameter:

$V_A=\frac{1}{\lambda }$

where $\lambda$ (units ${\text{V}}^{-1}$) sets how much the saturation current rises with $V_{DS}$ in the modified square-law $i_D=\frac{1}{2}{k}_n{V}_{OV}^2(1+\lambda V_{DS})$. For the BJT, $V_A$ plays the identical role via the Early effect: $i_C=I_S{e}^{v_{BE}/V_T}(1+v_{CE}/V_A)$. The name “Early voltage” comes from the BJT and was carried over to the MOSFET by analogy.

What it means geometrically

Extrapolate the (slightly upward-sloping) saturation/active output characteristics back to the left. [Background from general knowledge, not the source PDF: the geometric extrapolation interpretation; the PDF defines $V_A$ only as $1/\lambda$.] All the curves for different $V_{GS}$ (or $V_{BE}$) converge on a single point on the negative voltage axis, at $-V_A$. A large $V_A$ means the curves are nearly horizontal — almost an ideal current source. A small $V_A$ means they fan out steeply — a poor current source.

Output resistance and gain

The slope of a saturation/active curve is a finite output resistance, and for both device types it works out to the same compact form:

$r_o=\frac{V_A}{I}$

where $I$ is the DC bias current ($I_D$ for the MOSFET, $I_C$ for the BJT). See MOSFET output resistance. Because the maximum intrinsic gain of a single stage scales with $r_o$, a larger $V_A$ directly buys higher gain. Typical MOSFET values are $V_A\approx 20$–$100\text{V}$ ($\lambda \approx 0.01$–$0.05{\text{V}}^{-1}$); BJTs are often higher still. The practical takeaway: $V_A$ is the “quality” of the device as a current source, and it is the same concept whether you arrive at it from channel-length modulation or from the Early effect.