Is there a significant difference between the SPICE simulation engines such as Spice 3F5, 2G6, Spice B2, XSPICE, TopSpice, and other Spice engines? Do you think that one engine is better than all the others?
I have read on some commercial websites for third party Spice simulators that some Spice engines have better convergence than others. Is this true and/or much of a factor in the price for the program?
I have read on some commercial websites for third party Spice simulators that some Spice engines have better convergence than others. Is this true and/or much of a factor in the price for the program?
The "engine" is the same in all of them. It's that stuff that's wrapped around it that varies from one vendor to another. Yes, some vendors optimize their flavor of spice for one thing or another, so depending on what you want to do, one may be a little faster or may converge more reliably than another.
I_F
I_F
==> The "engine" is the same in all of them
Really? I have read on some of these websites (such as B2Spice and Simetrix) that their simulator program handles switchmode circuits better, or has convergence where other simulators would fail, or faster convergence, etc.
I guess I shouldn't believe all the marketing stuff I read. . . .
Thanks for the info I_Forgot!
Really? I have read on some of these websites (such as B2Spice and Simetrix) that their simulator program handles switchmode circuits better, or has convergence where other simulators would fail, or faster convergence, etc.
I guess I shouldn't believe all the marketing stuff I read. . . .
Thanks for the info I_Forgot!
The basic simulation algorithm is the same in all versions of spice - it is the Gear Trapezium algorithm.
But the details of the implementations have changed over the years; in particular the error estimation and convergence has been tweaked.
For well behaved circuits, all versions are OK. The differences matter for cases where the time scales in the circuit are very different (this means the equations are technically "stiff"), or when discontinous elements like switches or PWL sources are used.
The later versions of Berkely Spice (like 3f5 etc) are significantly better behaved than the 2 series; I don't know what tweaks, if any have been applied to the commercial variants.
But the details of the implementations have changed over the years; in particular the error estimation and convergence has been tweaked.
For well behaved circuits, all versions are OK. The differences matter for cases where the time scales in the circuit are very different (this means the equations are technically "stiff"), or when discontinous elements like switches or PWL sources are used.
The later versions of Berkely Spice (like 3f5 etc) are significantly better behaved than the 2 series; I don't know what tweaks, if any have been applied to the commercial variants.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.