Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

1/f corner deep submicron

Status
Not open for further replies.

drDOC

Member level 3
Joined
Jul 17, 2008
Messages
56
Helped
8
Reputation
16
Reaction score
1
Trophy points
1,288
Activity points
1,628
Hi,

I am experienced in cmos design with 0.7 micron. The 1/f corner off course depends on a transistor and it's bias condition. But generally in this process the 1/f corner is 5 to 10kHz.

So my question: what would generally be the 1/f corner in 65nm processes or below.

My feeling is that 1/f will not improve because the smaller mos devices have thinner oxides and suffer from more surface effects. So I would expect that the 1/f corner should go to 100kHz even MHz ranges.

Can anybody comment on 1/f corners in different processes of similarly biased transistors?
 

Haven't you tried running some simulations for that process?? It won't give you an exact value, but you could get an estimate of how much the 1/f noise deteriorates. I don't think it would go up to the MHz range, but it is possible that it gets into to the hundreds of kHz spectrum.

diemilio
 

Yes I could run simulations. And I am actually asking the question because I doubt if characterization of the noise parameters in those simulations.

However I am wondering if it is generally true that 1/f corners increase when feature sizes decrease or that it is a wrong assumption.
 

Input referred corner frequency in the 5 to 10 kHz range for min size (min area) device in 0.7um CMOS ?
It seems very low to me. At 200mV vgs-vth , for NMOS ?

1/f noise is not extensively/accurately measured in nanometer technologies, imho. No one cares much because the BW of the circuits are in the GHz range.

Don't you have the curve from the foundry showing the noise corner ?
 

1/f gets worse because the relative charge of the traps gets bigger. I would expect a linear scaling. Remember the 1/f is set by the intersection of the trap noise density with the white noise. The high-k gates improve by a factor so that at 65nm and below you will get little more analog lifetime. But be caution! I have seen minimum size 32nm device where the trap event reduce drain current as a step funtion by 30%. Is that an analog device?
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top