[SOLVED] Saving curve accuracy in virtuoso

[MahmoudHassan]

Hi
I need a help in cadence virtuoso
I was trying to extract curve data using save the table in virtuoso to a CSV file so that I can plot it Matlab
but the problem is that the table generated is rounded and so the curve is so inaccurate.

I am attaching the curve and the table generated



Best Regards,
Mahmoud

 

[erikl]

the problem is that the table generated is rounded and so the curve is so inaccurate.

Only a very tiny part of the curve is shown in the table:


... and that in logarithmic partition. May be the table is much longer?

And you could choose the same partition type (logarithmic) for the x-values (IL) in the curve representation.

 

[MahmoudHassan]

Aha thanks for your note
I am attaching the complete CSV file
I am using a linear dc sweep with a step of 1u (not logarithmic -- this curve is for load regulation the X axis is the load current and the Y axis is the output voltage )

I couldn't upload the file to edabroad so I used my google drive (CSV file)
https://drive.google.com/file/d/1eMUoEgfOQD7kZPaNQ1GDP3W9jtKXxGxm/view?usp=sharing

Here I am trying to use a logarithmic scale current and had the same problem for the reported output voltage
https://drive.google.com/file/d/1vU4Xj1gHrKZOyN0QS9UQF-aO9X-Fnvnn/view?usp=sharing

and the logarithmic response:

 

[Dominik Przyborowski]

In the table, you are able to change formating and set number of significant digits - by default it is 4, so your numbers are rounded to 1.1.

To change it you should to mark column in which you would like to change format, and from the context menu (under right mouse click) should be field format.
However, I have not an access to virtuoso so I am not sure, whether this option is not under menu Tool.

 

[dick_freebird]

Do you really believe that the display "inaccuracy" is more
significant than the rollup of model fidelity / process
variation / omitted parasitics?

What is "good enough"?

 

[erikl]

I am attaching the complete CSV file
I am using a linear dc sweep with a step of 1u (not logarithmic -- this curve is for load regulation the X axis is the load current and the Y axis is the output voltage )

It really looks like the CSV export doesn't report the y-values with the required resolution. Perhaps there is an accuracy setting possibility per axis before exporting?

A possible work-around: Try and subtract a fix value (e.g.) 1.09975 from the y-values (instead of V(net15), let V(net15)-1.09975 be shown). Don't know, if this could work.

 

[dick_freebird]

Have you tried using the Calculator "print" rather than CSV export?
I used to do that a lot, copy/paste the X window contents to a
text file and import to Excel. I never had the problems you describe.
Maybe this is the way to go.

 

[MahmoudHassan]

Do you really believe that the display "inaccuracy" is more
significant than the rollup of model fidelity / process
variation / omitted parasitics?

What is "good enough"?

I am also doing corner analysis with temperature variations and supply voltage too
I am trying to just measure a specification of the LDO I am designing
But also I will provide a tabulated corner analysis

- - - Updated - - -

In the table, you are able to change formating and set number of significant digits - by default it is 4, so your numbers are rounded to 1.1.

To change it you should to mark column in which you would like to change format, and from the context menu (under right mouse click) should be field format.
However, I have not an access to virtuoso so I am not sure, whether this option is not under menu Tool.

IT WORKED
Yes I Found it I will provide photos to show it

- - - Updated - - -

It really looks like the CSV export doesn't report the y-values with the required resolution. Perhaps there is an accuracy setting possibility per axis before exporting?

A possible work-around: Try and subtract a fix value (e.g.) 1.09975 from the y-values (instead of V(net15), let V(net15)-1.09975 be shown). Don't know, if this could work.

I think it will work if the problem is in formatting. I tried another solution to take the DB of the output and then convert it back in Matlab but resolution was also limited

- - - Updated - - -

Have you tried using the Calculator "print" rather than CSV export?
I used to do that a lot, copy/paste the X window contents to a
text file and import to Excel. I never had the problems you describe.
Maybe this is the way to go.

I didn't know how to do print from the calculator. can you provide me a photo where to find it may be I will need it in future?

- - - Updated - - -

It worked by trying to find out the formatting option in the table
I am providing the steps required for anyone may face this problem
The problem was in the number of significant digits in exporting







Thanks a lot for your help really