Temperature sweep of a chopped bandgap in MonteCarlo

Hi again,

I believe that of the usage of the additionalParams approach suggested here is the right way to go. Including the parameter file so that temperature is changed dynamically over time allows me to observe whether the curvature compensation is effective for each of the Monte Carlo samples (after trimming, of course). I hope this can help someone else in the future.

Regards,

Lewis

Hi Lewis,

Actually the tran analysis in current versions (this changed in an IC617 hot fix version some time back) allows the dynamic parameter set to be set up on the transient analysis form:

If you're not varying two parameters at the same time, then you have been able to do this via with the Dynamic Parameter capability on the tran analysis form for many years, so I'm sure you don't need to use additional parameters to achieve what you're doing in that case.

Be aware though - making too large a step in the temperature within a transient is likely to lead to convergence problems as otherwise you've introduced a sudden discontinuity. You can take advantage of the "param_step" option (see "spectre -h tran" on the UNIX command line for more details) to break down the step into smaller pieces (param_step is the time-step between each discrete change). That does still have to be entered via the additional parameters field.

Regards,

Andrew.

Hi Andrew,

Thanks for your remarks. Let me see if I got this right. In my particular test I'm using 5 degree steps every 1ms. I do observe some overshoot in the reference signal and indeed I presumed it was due to the discontinuity introduced with the temperature change. If I were to set param_step option to 1ms, would this mean that the 5 degree change is done gradually over 1ms (let's say linearly)? How does the tool do this? Does it take the timestep to gradually increase the temperature?

Additionally,

Andrew Beckett said:

That does still have to be entered via the additional parameters field.

You mean I have to enter param_step=1m in the additionalParam field as well as paramset=myset and load my paramset.scs file or simply add param_step=1m and use the dynamic parameter form from ADE-L to change the temperature?

Thank you for helping us all get the most out of the Virtuoso tool time and again.

Regards,

If you are changing the temperature in the table every 1ms and set param_step to 1ms, then it will be no different. The parameter change is always discontinuous - if you set the temperature change to be (say) 5 degrees every 1ms, and you set param_step=0.1m then it would change the temperature by 0.5 degree every 0.1ms. In essence it will do smaller discontinuous steps without you needing to specify each intermediate step in the paramset table. So if the first temperature was temp1 at time1 and the second was temp2 and time2, it would update the temperature every param_step to be param_step*(temp2-temp1)/(time2-time1).

For your second question, if you've specified the paramset information through the ADE UI as I've shown above, or you're using just a single parameter being changed in older versions, the only "additional parameter" you need to specify is the param_step - no need for external files in that case.

Regards,

Andrew.