The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Modeling >> Passive Devices >> PHY_RES SPECTRE
https://designers-guide.org/forum/YaBB.pl?num=1218645658

Message started by monte78 on Aug 13^th, 2008, 9:40am

Title: PHY_RES SPECTRE
Post by monte78 on Aug 13^th, 2008, 9:40am

Hello everybody,

I have a question about the phy_res model in spectre. I am designing a resistive DAC using n-diffusion resistors, but I am seeing some problems about the non-linear voltage dependence of these resistors.

In particular I noticed that the problem is due to the fact that the behaviour of the resistor is asymmetric with respect to the terminals Vt1 and Vt2, in the sense that if I put the resistors in one way I obtain a certain INL, while putting them in the opposite way the INL change (the resistor is perfectly symmetric from a layout point of view).

This is obviously a non-sense, so my questions are:

1) Can I use the phy_res model in the case that I am not sure about the more negative terminal of the resistor?
2) Why the phy_res model doesn't take into account the absolute value of the voltage across it? (polyarg=diff)
3) Is there a possibility to have the absolute value take into account on these resistors?

Thanks, Best Regards,

Monte

Title: Re: PHY_RES SPECTRE
Post by Geoffrey_Coram on Aug 13^th, 2008, 10:45am

Isn't this a question for the foundry modeling team? polyarg=diff gives them the ability to make a symmetric device, and they chose not to!

Title: Re: PHY_RES SPECTRE
Post by monte78 on Aug 14^th, 2008, 1:11am

Hi,

The problem is the polyarg=diff! If I have the plus terminal more negative than the minus terminal the model is not correct. :(

Geoffrey_Coram wrote on Aug 13^th, 2008, 10:45am:

Isn't this a question for the foundry modeling team? polyarg=diff gives them the ability to make a symmetric device, and they chose not to!

Title: Re: PHY_RES SPECTRE
Post by Geoffrey_Coram on Aug 15^th, 2008, 11:17am

Hmm, I read the Spectre documentation that said that the resistor is symmetric wrt the two resistor terminals when polyarg=diff. It's when polyarg=sum that you get asymmetry.

However, now that I look at the equation, it's only symmetric if the odd coefficients are zero, since there's no absolute value for V = V(t1) - V(t2).

Title: Re: PHY_RES SPECTRE
Post by monte78 on Aug 20^th, 2008, 5:58am

You are right, this is basically my problem. I will change model!

Thanks, Bye,

Monte

Geoffrey_Coram wrote on Aug 15^th, 2008, 11:17am:

Title: Re: PHY_RES SPECTRE
Post by jbdavid on Aug 27^th, 2008, 1:23am

for a diffusion resistor, the Phillips RDIFF model is probably a good bit better than the phy_res..

I think its built into spectre (and spice) these days..
If _I_ had to create a diffusion resistor model for a pdk, its what I'd use.
G'luck

Title: Re: PHY_RES SPECTRE
Post by sheldon on Sep 13^th, 2008, 5:57am

You might also want to look at r3, the Compact Modeling Council
three terminal resistor model. It has also available and intended
for modeling the non-linearity of three terminal resistors.

Best Regards,

Sheldon

Title: Re: PHY_RES SPECTRE
Post by rf-design on Mar 29^th, 2009, 5:42am

I read the McAndrew paper.

Is the "r3" which implement depletion, velocity and self heating somewhere implemented?

Title: Re: PHY_RES SPECTRE
Post by Geoffrey_Coram on Aug 11^th, 2009, 10:09am

The r3 model was standardized by the Compact Model Council; it's available in recent versions of Spectre, and I believe the Verilog-A is available from the CMC web site.