Designer’s Guide Community

niloun

Community Member

Offline

Posts: 53

Comparators input beyond supply rails
Nov 10^th, 2017, 12:33pm

Hi everyone

I have simulated a StrongArm comparator which its inputs are beyond its supply (for instance: Vdd=0.9, Vi+=1.35+0.45*sinwt, Vi-=1.35) and surprisingly it works properly (even in the simulation corners). If it is so, why don't we decrease the supply to save power, what is wrong with decreasing the supply as much as possible for any input range?

Thanks in advance

Back to top

comp.jpg

IP Logged

DanielLam Community Member Offline Posts: 76	Re: Comparators input beyond supply rails Reply #1 - Nov 10^th, 2017, 2:22pm Oxide stress. The Vgs of the input devices can also be > Vdd and runs the risk of overstressing the oxide. Btw for this nmos input comparator, the higher the common mode, the higher the input thermal noise. Also, lower Vdd will give lower speed.
Back to top	IP Logged

niloun

Community Member

Offline

Posts: 53

Re: Comparators input beyond supply rails
Reply #2 - Nov 10^th, 2017, 11:48pm

DanielLam wrote on Nov 10^th, 2017, 2:22pm:

Oxide stress. The Vgs of the input devices can also be > Vdd and runs the risk of overstressing the oxide.

Btw for this nmos input comparator, the higher the common mode, the higher the input thermal noise.

Also, lower Vdd will give lower speed.

THanks DanielLam. I have attached two image from Razavi's survey on StrongArm. So you mean that in phase "b" transistors M1 and M2 receive an unusual amount of voltage which may cause damage to them?

Back to top

IP Logged

neoflash

Community Fellow

Offline

Mixed-Signal
Designer

Posts: 397

Re: Comparators input beyond supply rails
Reply #3 - Nov 14^th, 2017, 4:44pm

If you don't need speed from the regeneration latch, you can reduce the supply to save power. Or alternatively, reduce the size of transistor at expense of larger offset.

niloun wrote on Nov 10^th, 2017, 12:33pm:

Hi everyone

I have simulated a StrongArm comparator which its inputs are beyond its supply (for instance: Vdd=0.9, Vi+=1.35+0.45*sinwt, Vi-=1.35) and surprisingly it works properly (even in the simulation corners). If it is so, why don't we decrease the supply to save power, what is wrong with decreasing the supply as much as possible for any input range?

Thanks in advance

Back to top

IP Logged

niloun

Community Member

Offline

Posts: 53

Re: Comparators input beyond supply rails
Reply #4 - Nov 14^th, 2017, 11:40pm

neoflash wrote on Nov 14^th, 2017, 4:44pm:

If you don't need speed from the regeneration latch, you can reduce the supply to save power. Or alternatively, reduce the size of transistor at expense of larger offset.

niloun wrote on Nov 10^th, 2017, 12:33pm:

Hi everyone

I have simulated a StrongArm comparator which its inputs are beyond its supply (for instance: Vdd=0.9, Vi+=1.35+0.45*sinwt, Vi-=1.35) and surprisingly it works properly (even in the simulation corners). If it is so, why don't we decrease the supply to save power, what is wrong with decreasing the supply as much as possible for any input range?

Thanks in advance

Thanks, so there will be no problem if one wants to fabricate this circuit? Since I have heard about issues including : "bulk current" (which I have checked in the simulations and it was OK) or "latch up" or " oxide stress" which Daniel mentioned?

Back to top

IP Logged

Frank_Heart

Junior Member

Offline

Posts: 29

Re: Comparators input beyond supply rails
Reply #5 - Nov 17^th, 2017, 12:08am

If your input device Vgs can handle (Vi+=1.35+0.45*sinwt), you don't have stress problem.

With lower supply for latch, as DanielLam and neoflash mentioned, your dynamic gain is small and speed for latch would be low.

And, how about the supply for the logics after comparator? You probably need a level-shifter to bring the voltage swing to normal supply range.

-Frank

Back to top

IP Logged

niloun

Community Member

Offline

Posts: 53

Re: Comparators input beyond supply rails
Reply #6 - Nov 17^th, 2017, 1:33am

Frank_Heart wrote on Nov 17^th, 2017, 12:08am:

If your input device Vgs can handle (Vi+=1.35+0.45*sinwt), you don't have stress problem.

With lower supply for latch, as DanielLam and neoflash mentioned, your dynamic gain is small and speed for latch would be low.

And, how about the supply for the logics after comparator? You probably need a level-shifter to bring the voltage swing to normal supply range.

-Frank

Thanks for your answer Frank.
You are right but we can lower the supply of the subsequent blocks and use a level-shifter in the final stage if it is necessary to have the original logic levels. Am I wrong?

Back to top

IP Logged

Frank_Heart Junior Member Offline Posts: 29	Re: Comparators input beyond supply rails Reply #7 - Nov 17^th, 2017, 8:19am Niloun, I agree with you. As soon as the overall speed/noise performances meet system requirements. It's a good idea to lower down supply for saving power. -Frank
Back to top	IP Logged