EKV Model in spectre cadence

[antaryami.mt.er09]

Hi...
Can anyone tell me if spectre supports EKV model
if so how can i get it?
I'm now using the BSIM3v3 model.

 

[erikl]

Can anyone tell me if spectre supports EKV model

Yes, Spectre supports the EKV models, s. p. 4 (top left).

if so how can i get it?

From the foundries, which support EKV models for Spectre.

 

[palmeiras]

Hi erikl.

Do you know some foundry that provide EKV MOSFET models for Spectre?
I´m curious because I have never seen. It is always BSIM3 or BSIM4.

 

[erikl]

Do you know some foundry that provide EKV MOSFET models for Spectre?

Sorry, no. You might search at the foundries' web sites, or ask the EKV inventors which foundry supports their models.

**broken link removed** you can find EKV v2.6 parameters for a 0.5µm CMOS process.

 

[palmeiras]

Hi erikl.

Thank you for your replay. I think that BSIM model is still the default model for foundries.
I thought that foundries should provide EKV or ACM models when these models are proper for low-voltage and low-power design. However, I think that due to the economical factors, these models are not available.

 

[oermens]

see EPFL | Model for a list of foundries pdks supporting ekv model

 

[antaryami.mt.er09]

Yes, Spectre supports the EKV models, s. p. 4 (top left).


From the foundries, which support EKV models for Spectre.

The reason why i want to switch is weak inversion model,
when i'm trying to simlate my MOS device in weak inversion with spectre using BSIM3V3,the parametr η should be betn 1.2 to 1.7 as theory says
And it should vary with L of mos but it is always remains at 1,which creates a big prob for my current equation in weak inv,
Can you please suggest me anything for this........
Thanks for ur reply....

 

[erikl]

... it should vary with L of mos but it is always remains at 1,which creates a big prob for my current equation in weak inv,
Can you please suggest me anything for this.

You could perhaps try with **broken link removed**, which are said to model the moderate/weak inv. region better - if all the necessary parameters are provided by the foundries. May be you want to try with PTM models first.

 

[oermens]

The "n" you are describing is not subthreshold slope parameter of EKV model, see **broken link removed** pg 194.

n=1 Junction emission coefficient.

bsim model does not calculate slope parameter, specific current, etc. that you are looking for, they must be approximated/derived from given bsim parameters and model parameters. slope factor can be approximated by simulating vg-id and observing inverse of slope of current on a log scale in subthreshold region

 

[antaryami.mt.er09]

The "n" you are describing is not subthreshold slope parameter of EKV model, see **broken link removed** pg 194.



bsim model does not calculate slope parameter, specific current, etc. that you are looking for, they must be approximated/derived from given bsim parameters and model parameters. slope factor can be approximated by simulating vg-id and observing inverse of slope of current on a log scale in subthreshold region

When i saw the BSIM3V3 manual the model equation for n is written,
it can be calculated that way ,but i when i plot a curve for n it shows constant @ 1which
should not be the case in weak inv............
I've tried log(Id) ~Vgs then it's slope at a point using the caculator it show 12.36,
but slope will give me 1/ηVt,Vt is thermal voltage ,
In that way the result is wrong..........

 

[oermens]

by log(Id) you mean you are taking log in calculator then plotting the result against Vgs? This is wrong. Plot Id vs Vgs and change the scale of Id from linear to log.

 

[erikl]

Do you really think this makes a difference, plotting linear values on a log scale, or their log values on a linear scale? Always equidistant orders of magnitude. I think just the scale legend will change.
Changing a scale from linear to log just calculates the log values and displays them on a linear scale. And shows a nice subdivision between orders of magnitude ;-)

 

[oermens]

the subthreshold slope parameter is the inverse of the slope in the subthreshold region which is measured by the amount of change in vgs to cause a 1 decade increase in current. that is why it is more useful to put it on log scale. if it is linear, subthreshold current is pA to nA and is difficult to place two markers 1 decade apart.

 

[erikl]

The OP did not talk about plotting linear data, s. above.
Displaying linear data on a log scale or logarithmic data on a linear scale results in identical plots - apart from the scale legends. Equidistant orders of magnitude, and in both cases you can choose lower and upper display limits. What's the difference, then? Or do I misunderstand you?

 

[antaryami.mt.er09]

by log(Id) you mean you are taking log in calculator then plotting the result against Vgs? This is wrong. Plot Id vs Vgs and change the scale of Id from linear to log.

Can you please tell me how to calculate the slope rather than doing it hand calculation?
i mean is there anyway i can use calculator for that ?

 

[oermens]

CMOS: circuit design, layout, and ... - Google Books

edit: just read over this, it is as erikl described (take log of Id, plot against vgs). first time i've seen subthreshold slope measurement described that way, since Id is already exponential in subthreshold region it seems easier to just plot it on a log scale.

 

[antaryami.mt.er09]

CMOS: circuit design, layout, and ... - Google Books

edit: just read over this, it is as erikl described (take log of Id, plot against vgs). first time i've seen subthreshold slope measurement described that way, since Id is already exponential in subthreshold region it seems easier to just plot it on a log scale.

Thanks to both of you......