A good question. On the contrary, I would say you should use w=1 for the PMOS, because you have 2 in parallel what makes them act as doubled single one. hence, currently it is like you have a CMOS inverter of 4:1 ratio... He mentioned several times "worst case of PD/PU path". according to that, in case of the nmos PD, one should make it lesser resistive, unlike the pmos PU. still, it doesn't fit with intending to have a normalized delay as the CMOS inverter... in fact, there is a limitation with normalizing it. for example, try to normalize it in NAND3. (You can't because you don't have a pmos less then w=1...)
Thanks for such videos.
but now width of nmos and pmos becomes same, so how will gonna follow delay symmetricity?
A good question.
On the contrary, I would say you should use w=1 for the PMOS, because you have 2 in parallel what makes them act as doubled single one. hence, currently it is like you have a CMOS inverter of 4:1 ratio...
He mentioned several times "worst case of PD/PU path". according to that, in case of the nmos PD, one should make it lesser resistive, unlike the pmos PU.
still, it doesn't fit with intending to have a normalized delay as the CMOS inverter...
in fact, there is a limitation with normalizing it. for example, try to normalize it in NAND3. (You can't because you don't have a pmos less then w=1...)
The quality of teaching is non-pareil.