Do i need Isocell for a signal going from AON domain to SWD(off) domain?

[nitint08]

If yes, please explain. Isocell only needed to prevent x or z going to other domain but if destination is already OFF than it doesnot make any sense.

 

[narfnarf]

I think you have the right idea. My understanding is that isolation cells are only from possibly-off domain to always-on domains.

narfnarf

 

[nitint08]

Thanks. but do you think the possibility of latchup here.
say signal in1 from AON(0.9) going to SWD(1.1, 0) via LS(0.9 -> 1.1) so in1 as "1" being supplied to some cell in SWD domain. So may it develop a latchup situation here assuming designers has not deployed latchup protection circuity and at layout level it also get missed.
If such situation may arise latchup than using isocell with clamp value "0" will prevent such situation?

otherwise i don't understand any situation where isocell is needed from AON to SWD.

 

[devop]

hi， check your library , think about a AON cell driving a transfer gate , you need the iso cell

 

[nitint08]

Does it mean, isocell should be placed between AON and SWD domain(in power-down) and isocell is must from signal going from SWD domain(in OFF) to AON.
So 2 isocell needs, one for AON -> SWD and another for SWD -> AON?

Now if two domains PD1 and PD2 both getting ON and OFF at same time than when both are OFF and a signal from PD1 going to PD2 via another AON domain(PD3). Do i need two isocell here, one bet PD1 to PD3 and another between PD3 and PD2?
or no isocell is needed between PD1 and PD2 as both are OFF, so knowing pst-state table it should be calculated that if multiple domains have such states that those go ON and OFF simultaneously than all the signals between these domains should be marked ignored for isocell requirement?