Using a nodeset in a pre-run calibration

Dear david73,

david73 said:

My first thought was to use a nodeset to help for IC, but I am a little unsure, because every iteration of the simulation changes a control, which makes a (significant?) change to the circuit, and I don't want to bring the circuit to a bad operating point because of it.

Using a nodeset command on a net, unlike an ic command, does not force a node voltage to be the value you choose for the nodes s the DC operating point voltage for the node. It only serves as a value for the start of the DC operating point analysis. The concept is that by using a voltage value for a net "close to" the final value, the number of iterations required to compute the DC operating point will be minimized. 

Cadence On-line support has an Application note, in case you have not seen it, discussing the differences between the use of its nodeset and ic commands at its URL:

https://support.cadence.com/apex/ArticleAttachmentPortal?id=a1O0V0000091CAfUAM&pageName=ArticleContent

with the mechanics as to how spectre treats a nodeset and ic in an initial DC operating point analysis at URL:

https://support.cadence.com/apex/ArticleAttachmentPortal?id=a1Od0000000nbGzEAI&pageName=ArticleContent&id=a1Od0000000nbGzEAI&oMenu=Documents%20which%20may%20be%20similar%20to%20this%20document

Do these help answer your question david73? I hope I understood your question correctly!

Shawn

I understand the difference between an initial condition and a nodeset, though I will read the app note. Both of the main stated uses are to help a circuit with more than one DC solution choose the "correct" one. My question is whether I can use it just to speed a simulation, and whether it is correct to use if a control of the circuit has changed. For example, if the control of a delay line has changed, will the nodeset with the old control hurt the solution?

Dear david73,

david73 said:

I understand the difference between an initial condition and a nodeset, though I will read the app note

Great! I think the second URL I provided may help to answer your second concern - but I will respond to it anyway.

david73 said:

Both of the main stated uses are to help a circuit with more than one DC solution choose the "correct" one.

In general, I tend to think that both are not usually employed to choose between two valid DC operating points - that certainly is one use. I might suggest the dominant use of the nodeset command is for exactly what you want - to expedite the time to complete an intial DC operating point. I think the dominant use of an ic command is to force a specific node to have a specific value after a DC operating point analysis.

david73 said:

My question is whether I can use it just to speed a simulation, and whether it is correct to use if a control of the circuit has changed. For example, if the control of a delay line has changed, will the nodeset with the old control hurt the solution?

If you are referring to a nodeset command, my response to your question is yes, you can use a set of nodeset commands (one for single node or N for N nodes) "just to speed a simulation". The amount of time their use might save is circuit dependent and hence will vary.

As to whether a nodeset is "correct to use ff a control of the circuit has changed", in my opinion, in general it should not impact the final DC solution (unless your circuit has multiple valid DC opearting points), but may result in a longer time to find a DC operating point than without the nodeset. From the second URL I provided:

"Nodesets on a node cause a current source set to the nodeset value in parallel with a 1 ohm resistor to be applied to the node. The DC algorithm iterates to a solution, then the resistors/current sources are removed. The simulation starts from the converged solution, and iterates (hopefully) to convergence."

Hence, the "starting" point for a final DC operating point is influenced but the value(s) of nodeset(s) you choose.

I hope this provides a bit more insight david73!

Shawn