CMOS Schmitt Trigger: contradictionary information

Question

Weste and Harris in "CMOS VLSI Design" describe the CMOS trigger as follows:

Let's call the transistors in the first columns from bottom to top M1,M2; M4,M6 and the transistors in the second column M3 and M5.

Note that M3 and M5 are indicated as weak, implying that their sizes are smaller than the other transistors.

On the other hand, Jacob Baker in CMOS Circuit Design, Layout, and Simulation it is implied that the same transistors must be much larger for having hysteresis. For example, in example 18.1. he derives W1=10u, W3=22.5u, W6=20u, W5=26u; W2 and W4 larger than the other transistors.

Which version is now correct?

Answer 1

Both the versions are correct. consider the explanation below.

Suppose I give a square wave signal as an input. When the voltage at A is rising from 0 to rail voltage (Vdd), the bottom nmos M1 turns on slowly as A reaches Vt. This is the time you can see the actual effect of the M3 nmos in feedback. During this transition, you can think of the M1,M3 pair of transistors to be like a potential divider circuit where the impedances of M1,M3 depend on their sizes (W1,W3). Ron for a transistor is inversely proportional to it's width .If W1 is less than W3, you'll see that the voltage at the source terminal of M2 is larger than what it would've been in the case where W1 is greater than W3. Hence, when W1 is less than W3, the Schmitt trigger would've switched it's output low at an input voltage higher than when W1 was sized larger than W3. The same holds true with the upper pmos during the other transition of input signal.

So, I feel the choice of the sizes of feedback mosfets are to alter the Vih and Vil which are the voltages at which the circuit will switch from High to Low and Low to High. Both the scenarios will produce a hysteresis band needless to say.