Regression/Validation testing with SPICE

[Darryl Dieckman]

Are there any tools out there that allow you to compare the results of
multiple SPICE simulations to verify that your circuit is operating
as expected between different versions of the circuit?

For example, I have revision 1.0 of my design and create a bunch of
test decks to verify its operation. I have the chip fabricated and
then either find a problem that I want to correct or need to adjust
a small portion of the chip's functionality. What is the best way
to rerun my set of test decks to make sure that everything else is
still working and that I did not accidentally break some other part
of the design?

I can certainly rerun the tests and visually inspect the plots between
the two versions of the chip, but I would much rather have this process
automated such that I could detect and flag differences between key
measurement values using some scripts or tools.

I'm using Spice3f5 and/or ngspice so I don't have the "measure" function
that simulators like PSpice have. I'm curious what other methods people
use to validate their designs. I'm sure I could write a bunch of PERL
to compare the results but I'd rather not reinvent the wheel if there
is already a better way to solve the problem.

Thanks in advance

Darryl Dieckman
Senior Engineer
Clifton Labs, Inc.

 

[Charles DH Williams]

In article <549f6ebe.0309081234.40317cc0@posting.google.com>,
ddieckma@cliftonlabs.com (Darryl Dieckman) wrote:

Are there any tools out there that allow you to compare the results of
multiple SPICE simulations to verify that your circuit is operating
as expected between different versions of the circuit?

For example, I have revision 1.0 of my design and create a bunch of
test decks to verify its operation. I have the chip fabricated and
then either find a problem that I want to correct or need to adjust
a small portion of the chip's functionality. What is the best way
to rerun my set of test decks to make sure that everything else is
still working and that I did not accidentally break some other part
of the design?

Try the following:

Use 'write'

http://newton.ex.ac.uk/teaching/CDHW/Electronics2/userguide/sec5.html#5.3.60

to save a cannonical set of results in a rawfile.

Later... run your revised simulation, then use 'load'

http://newton.ex.ac.uk/teaching/CDHW/Electronics2/userguide/sec5.html#5.3.25

to read the cannonical values back and use 'diff'

http://newton.ex.ac.uk/teaching/CDHW/Electronics2/userguide/sec5.html#5.3.12

to to get a report of differences.

Charles

 

[Jim Thompson]

On 8 Sep 2003 13:34:02 -0700, ddieckma@cliftonlabs.com (Darryl
Dieckman) wrote:

Are there any tools out there that allow you to compare the results of
multiple SPICE simulations to verify that your circuit is operating
as expected between different versions of the circuit?

For example, I have revision 1.0 of my design and create a bunch of
test decks to verify its operation. I have the chip fabricated and
then either find a problem that I want to correct or need to adjust
a small portion of the chip's functionality. What is the best way
to rerun my set of test decks to make sure that everything else is
still working and that I did not accidentally break some other part
of the design?

I can certainly rerun the tests and visually inspect the plots between
the two versions of the chip, but I would much rather have this process
automated such that I could detect and flag differences between key
measurement values using some scripts or tools.

I'm using Spice3f5 and/or ngspice so I don't have the "measure" function
that simulators like PSpice have. I'm curious what other methods people
use to validate their designs. I'm sure I could write a bunch of PERL
to compare the results but I'd rather not reinvent the wheel if there
is already a better way to solve the problem.

Thanks in advance

Darryl Dieckman
Senior Engineer
Clifton Labs, Inc.

If you're doing *only* value changes you could run Worst-Case or Monte
Carlo. If the designs are different implementations of the same
function you could use the "Append Waveform" function if you were
running PSpice ;-)

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice480)460-2350 | |
| E-mail Address at Website Fax480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.