The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> charge injection
https://designers-guide.org/forum/YaBB.pl?num=1360936885

Message started by rajdeep on Feb 15^th, 2013, 6:01am

Title: charge injection
Post by rajdeep on Feb 15^th, 2013, 6:01am

Hi All,

How good is Cadence Spectre for simulating effects of charge injection assuming the device models are of standard industry practice? I think the models are similar to BSIM3 models.

Thanks,
Rajdeep

Title: Re: charge injection
Post by Geoffrey_Coram on Feb 18^th, 2013, 12:26pm

I don't think the BSIM3/4 class of models are any good for charge injection, regardless of simulator. You need PSP.

Title: Re: charge injection
Post by rajdeep on Feb 19^th, 2013, 4:39am

Sorry for my ignorance. What is PSP? Google says Penn State Philips models. Can we simulate such models using Cadence Spectre?

Also, assuming I will not get any PSP model for our devices, is it possible to model the effect of charge injection for worst case scenario i.e. let all the charge be injected on to the node that affects the design most. For example, I have a scenario where a PMOS (5u/500n) switch turns off, and I suspect that the charge injection can lift the drain of the PMOS, and this can cause some issue in the design as the net connected to the drain of the PMOS is a critical net. To model the adversity that charge injection may cause I have added an extra cap (like 100fF) between the gate and drain of the PMOS. But I understand this could be an overkill. But I blv I can create a pessimistic model for worst case charge injection very crudely by adding such coupling capacitor. Is this a practical method designers (e.g. sw-cap ckt designers) can adopt?

Many thanks for the reply.
Rajdeep

Thanks,
Rajdeep

Title: Re: charge injection
Post by rajdeep on Feb 19^th, 2013, 2:41pm

Hi All,

Would like to share this just for completion.

I came to know that BSIM3 models are ok to model charge injection for lengths < 2uM. Since the switch I am concerned with has 500nM channel length, that should be alright.

Thanks,
Rajdeep

Title: Re: charge injection
Post by Geoffrey_Coram on Feb 20^th, 2013, 8:37am

BSIM3 and BSIM4 have the parameter XPART which you can use to decide how to handle the charge partitioning between drain and source. It's not physical, but it would allow you to set it to the value that gives the worst case for your particular simulation.

I don't know how that length < 2um is relevant. It seems to me that the ratio of the MOS capacitance to that of the rest of the circuit would be important, so at least the width should also be important.

Title: Re: charge injection
Post by rajdeep on Feb 21^st, 2013, 2:46am

What I understood was the shorter the channel length better is the approximation! That is to say, for longer channels the way the residual charge recombine with the substrate or find its way thru source or drain become more complicated and it does not happen instantaneously, as predicted by the BSIM models. They use some 40-60 rule i.e. 40% goes to the source and 60% to the drain. I think the 2uM number was more of a thumbrule, and may vary from process to process. This is what I was told by one of our process guy.

Best Regards,
Rajdeep