DC operating point for different analyses

Dear Banai,

Fid you happen to see the skipdc option in the documentation or alternately in the Application note concerning initial conditions and node sets at URL;

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

Shawn

Shawn (sorry to contradict you a little here), skipdc is not the right thing here, and should be used with extreme caution as it can certainly lead to the wrong result if not used very carefully. There are times when it's useful (mainly to aid with transient start up), but I'd be very wary of using it with a small-signal analysis that has no chance to recover from an incorrect starting point (it's the same reason that initial conditions are normally ignored for small-signal analyses).

Was about to answer the original question (day job got in the way).

Anyway, in general, each analysis will run a DC analysis at the beginning in order to ensure that it either can do the small-signal analysis around the operating point, or for large-signal analyses, it has a starting point for time-varying behaviour to work with.

Now whilst each analysis does its own operating point analysis, it will restart from the previous solution - and assuming the condition it's analysing is the same, it should reach convergence very quickly and not be expensive (it's likely to converge in an iteration if nothing changed from the previous analysis). If however you're running under a different condition (e.g. an ac analysis where you're sweeping some other parameter which affects the operating point, or you're running the ac/noise after a DC sweep say). Also, the DC run at the beginning of a transient may be different because the time-zero value are used compared with DC which might be using the DC value of a source (if that's different).

Each small signal analysis does have a prevoppoint=yes option that you can specify, which can attempt to use the previous operating point rather than doing it again (if there was one).

You can also for ac/noise use the oppoint parameter on the options form for that analysis to save the operating point to a file, or rawfile if you want to look at it in ViVA, so you could use that to look at any differences (or at least confirm to yourself that there are none).

Regards,

Andrew

Dear Andrew,

Andrew Beckett said:

Shawn (sorry to contradict you a little here), skipdc is not the right thing here, and should be used with extreme caution as it can certainly lead to the wrong result if not used very carefully.

Please - never feel that way and to be honest I totally agree with you!

It is FAR more important that you alert other Forum members of your wariness concerning my suggestion than to worry about contradicting me! Had you not, this additional valuable information might not have been clear to other Forum members!

The best use of this feature is for a transient analysis when there is a bit of time after the initial time point for the netlist transient solution to recover from any slight differences in the DC operating point you choose and that it would normally compute at its first time point. I certainly have observed cases in a small-signal analysis where a small difference in DC operating point can make a very significant difference in its resulting small-signal simulation result. This is especially the case, in my experience, for small-signal analysis of transfer functions of circuits whose large signal performance differs significantly from their small-signal performance (for example, a ring oscillator or DC/DC converter, etc).

In fact, page 6 of the Application Note to which I provided the URL notes that by default the use of .ic commands in analyses other than a transient is not allowed in the following paragraph:

"By default, Initial Conditions are honored in transient (and other time domain analyses), but not in AC or DC. The "ic" parameter on a transient analysis lets you turn on Initial Conditions in transient, and the "force" parameter on AC or DC lets you enable them for AC and DC analysis. The main use of Initial Condition however is for transient or time domain analyses.
Use Initial Conditions with caution, because they can lead to misleading results if you have forced the circuit into a non-realistic Initial Condition. With Nodeset, the effect of an incorrect initial Nodeset tends to be reduced by the completion of the DC operating point anyway and hence, it is less critical."

The only other comment I might mention with respect to the question for small-signal analysis is to potential use a nodeset for some or all nodes with a set of nodal voltages that are close to the desired DC values - this will only serve to guide the DC solution. Perhaps this might be of value to Banai?

Once again, thank for to your insight and willingness to correct my overly broad initial response Andrew!

A humbled Shawn

Gentlemen,

Thank you for your kind replies.

Andrew, your detailed description was very useful indeed. I was able to speed-up my simulations significantly and have started verifying the operating points of the different analyses.