Issue with Skywater 130 Monte Carlo Mismatch Analysis

What model files are you including in your simulations. From what I can see, the models have statistics blocks, but they are commented out. I can't see how Monte Carlo would even run... (I am assuming you're using Spectre). How did you set up the Monte Carlo run?

Regards,

Andrew

Hi Andrew, Thanks for the prompt response.

To run the monte-carlo analysis I am including the following files:

i) montecarlo.spice

ii) Additionally another spice file named critical.spice, contains some definitions needed for montecarlo.spice, so I am including that too.

Both montecarlo.spice and critical.spice are present in the parameters directory.

Hi Andrew,

I just wanted to follow up on my previous message. Any update or insight from your end would be really appreciable.

Apologies - I didn't get any email notification from your previous append where you answered my question.

However, I tried doing this - this is what I have in my spectre netlist (created from ADE):

include "$KITHOME/models/sky130.lib.spice" section=tt
include "$KITHOME/models/parameters/montecarlo.spice"
include "$KITHOME/models/parameters/critical.spice"

The first is the default model file that gets auto-added in ADE (I changed the path to have $KITHOME rather than the normal full path, as this makes it easier for me to switch kit versions).

With the addition of the second two files (actually just adding the second causes this), a single (non-monte Carlo) run fails with:

Reading file: /export/home/user/kits/sky130_release_0.1.0/models/sonos_e/begin_of_life.pm3.spice

Error found by spectre in `sky130_fd_bs_flash__special_sonosfet_star', during circuit read-in.
  ERROR (SFE-1016): "/export/home/user/kits/sky130_release_0.1.0/models/parameters/../sonos_e/begin_of_life.pm3.spice" 30: Model `sky130_fd_bs_flash__special_sonosfet_star__model.0', the expected number of value followed by `l*w*mult' is 1, but found 0.
  ERROR (SFE-1016): "/export/home/user/kits/sky130_release_0.1.0/models/parameters/../sonos_e/begin_of_life.pm3.spice" 178: Model `sky130_fd_bs_flash__special_sonosfet_star__model.1', the expected number of value followed by `l*w*mult' is 1, but found 0.
  ERROR (SFE-1016): "/export/home/user/kits/sky130_release_0.1.0/models/parameters/../sonos_e/begin_of_life.pm3.spice" 326: Model `sky130_fd_bs_flash__special_sonosfet_star__model.2', the expected number of value followed by `l*w*mult' is 1, but found 0.

Reading file: /export/home/user/kits/sky130_release_0.1.0/models/sonos_p/begin_of_life.pm3.spice

Error found by spectre in `sky130_fd_bs_flash__special_sonosfet_original', during circuit read-in.
  ERROR (SFE-1016): "/export/home/user/kits/sky130_release_0.1.0/models/parameters/../sonos_p/begin_of_life.pm3.spice" 30: Model `sky130_fd_bs_flash__special_sonosfet_original__model.0', the expected number of value followed by `l*w*mult' is 1, but found 0.
  ERROR (SFE-1016): "/export/home/user/kits/sky130_release_0.1.0/models/parameters/../sonos_p/begin_of_life.pm3.spice" 178: Model `sky130_fd_bs_flash__special_sonosfet_original__model.1', the expected number of value followed by `l*w*mult' is 1, but found 0.
  ERROR (SFE-1016): "/export/home/user/kits/sky130_release_0.1.0/models/parameters/../sonos_p/begin_of_life.pm3.spice" 326: Model `sky130_fd_bs_flash__special_sonosfet_original__model.2', the expected number of value followed by `l*w*mult' is 1, but found 0.

Also, looking in these files, there are no statistic definitions - there is in the "critical.spice" file but it's all commented out. I can't see how this would be working with Monte Carlo for you.

If I omit the "montecarlo.spice" then the simulation runs (I leave the critical.spice included) as a single run. 

If I run with Monte Carlo, Spectre fails with:

************************************************
Monte Carlo Analysis `mc1': iteration = (1 -> 5)
************************************************

Error found by spectre during Monte Carlo analysis `mc1'.
  ERROR (SPECTRE-16012): mc1: Attempt to run Monte Carlo analysis with process and mismatch variations, but no variations were specified in the statistics block.

Analysis `mc1' was terminated prematurely due to an error.

Can you please share the model setup from the top of the input.scs? Have you modified any of the model files?

Also, where did you get the PDK from? I got it from the Resources->PDK on http://support.cadence.com - i.e. sky130_release_0.1.0 (is that the version you're using?)

Andrew