Please comment on the definition of V_OH and V_OL for the inverter, Ch. 11?

In your Fig. 11.2 you define V_OH, for example, as the maximum inverter output

voltage and V_OL as the minimum inverter output voltage. However, I've seen a few

others, including wikipedia.org, that define V_OH (V_OL) as the output voltage

when the input is at V_IL (V_IH).

Good question. Defining V_OH (V_OL) when the inverter's input is V_IL (V_IH) is more

exact for specifying the noise margins, Eqs. (11.1) and (11.2). However, if you

look at the difference in a practical inverter the numerical difference between

the two methods is minor. It's much easier (and more practical) to determine

V_OH and V_OL using the definitions presented in Ch. 11.

For example, looking at the short channel inverter VTC in Fig. 11.4 the inverter's

V_IL = 350 mV (at this voltage V_OH = 944 mV) and V_IH = 514 mV (at this voltage V_OL

= 65 mV). So

NM_H = V_OH - V_IH = 430 mV

NM_L = V_IL - V_OL = 285 mV

Now using the book's definitions, V_OH = VDD = 1 V and V_OL = 0 V we get

NM_H = 486 mV (difference of 56 mV)

NM_L = 350 mV (difference of 65 mV).

However, let's look at how much V_IL and V_IH vary for VTC slopes changing from -0.5

to -1.5 instead of just looking at the values when the slope is -1. In this range

V_IL varies from 300 to 370 mV (70 mV) and V_IH varies from 500 to 555 mV (55 mV).

This variation is comparable to the difference seen using the two different methods.

Perhaps this minor difference is why most books use the engineering approximation of

V_OH /V_OL corresponding to the max/min inverter output voltages. Again, good question.

One last comment; CMOS is a unique logic family where the logic outputs can swing

from VDD to ground. In many logic families the outputs don't swing all the way to

the power supply rails. V_OH isn't equal to VDD and V_OL doesn't equal ground as

discussed in the book.

Return