why var-R MOS oscillate at "accurate=1" in tran. s

[paulux]

As I am using variable resistance nature of MOS as a tunable resistor in trans. sim, the oscillation occurs when the setting in Hspice "accurate=1" is enabled. When I do not set any option or just use "method=gear", the output just shows few ringings and then stable. I have checked all the op. points during the changing transient. They are all stable(at least 30deg phase margin)....Since there is no DC current flowing through this var-R MOSFET and the gm is found to be zero when .op is used and only rds of several hundreds kΩ is found, is this problem related to MOSFET model inaccurate parameter extraction problem or spice simulation accuracy problem? Which result should close to the real situation? The model I use is level49 MOSFET model. Please kindly help and comment about this.
Or any others comments are highly welcomed

 

[braudelk]

As u have mentioned, "method=gear" does matters.
This issue is related to the arithmetic hspice used. All tools has its limitation.

 

[kamalakkannan]

Re: why var-R MOS oscillate at "accurate=1" in tra

hi..
try with this..
.option method=gear accurate=1
basically this option sets [max dv/dt] between the internal time points which simulator kernal uses on its own,internal time step,accurate integration mthod

you are suppose to get reliable results in this option.but simulation time will be long after adding this option.

--
kamal.

 

[paulux]

Re: why var-R MOS oscillate at "accurate=1" in tra

Thanks for your reply.
After I set ".option method=gear accurate=1", the result is the same as "accurate=1". And I checked with Hspice menu, only the last option "accurate=1" is executed in Hspice. In fact, how could I determine which way (1)"method=gear" or without any setting, or (2) accurate=1 could provide realistic result before the fabrication? Or any other ways or suggestions? THX SO MUCH.

 

[MLR67]

Re: why var-R MOS oscillate at "accurate=1" in tra

When you use Gear , convergence is achieved with an "integration" method.
So sometimes real oscillations in silicon vanish on spice or other tool.
Be carefull with gear , trap and etc. , try to perform AC simulations to verify stability