<@U01SBMJT5BP> It looks like there are a few more ...
# xycem
Matthew Guthaus
06/21/2021, 11:03 PM@User It looks like there are a few more model errors that didn't show up until I started simulating other (memory) devices:
Copy code
Netlist warning in file
/software/PDKs/sky130A/libs.tech/xyce/../../libs.ref/sky130_fd_pr/spice/sky130_fd_pr__special_nfet_pass.pm3.spice
at or near line 34
No model parameter LDIF found for model
SKY130_FD_PR__SPECIAL_NFET_PASS__MODEL.0 of type NMOS, parameter ignored.
Netlist warning in file
/software/PDKs/sky130A/libs.tech/xyce/../../libs.ref/sky130_fd_pr/spice/sky130_fd_pr__special_nfet_pass.pm3.spice
at or near line 34
No model parameter HDIF found for model
SKY130_FD_PR__SPECIAL_NFET_PASS__MODEL.0 of type NMOS, parameter ignored.
Netlist warning in file
/software/PDKs/sky130A/libs.tech/xyce/../../libs.ref/sky130_fd_pr/spice/sky130_fd_pr__special_nfet_pass.pm3.spice
at or near line 34
No model parameter RD found for model
SKY130_FD_PR__SPECIAL_NFET_PASS__MODEL.0 of type NMOS, parameter ignored.
Netlist warning in file
/software/PDKs/sky130A/libs.tech/xyce/../../libs.ref/sky130_fd_pr/spice/sky130_fd_pr__special_nfet_pass.pm3.spice
at or near line 34
No model parameter RS found for model
SKY130_FD_PR__SPECIAL_NFET_PASS__MODEL.0 of type NMOS, parameter ignored.
Netlist warning in file
/software/PDKs/sky130A/libs.tech/xyce/../../libs.ref/sky130_fd_pr/spice/sky130_fd_pr__special_nfet_pass.pm3.spice
at or near line 34
No model parameter RSC found for model
SKY130_FD_PR__SPECIAL_NFET_PASS__MODEL.0 of type NMOS, parameter ignored.
Netlist warning in file
/software/PDKs/sky130A/libs.tech/xyce/../../libs.ref/sky130_fd_pr/spice/sky130_fd_pr__special_nfet_pass.pm3.spice
at or near line 34
No model parameter RDC found for model
SKY130_FD_PR__SPECIAL_NFET_PASS__MODEL.0 of type NMOS, parameter ignored.
Netlist warning in file
/software/PDKs/sky130A/libs.tech/xyce/../../libs.ref/sky130_fd_pr/spice/sky130_fd_pr__special_nfet_pass.pm3.spice
at or near line 34
No model parameter NQSMOD found for model
SKY130_FD_PR__SPECIAL_NFET_PASS__MODEL.0 of type NMOS, parameter ignored.
Netlist warning in file
/software/PDKs/sky130A/libs.tech/xyce/../../libs.ref/sky130_fd_pr/spice/sky130_fd_pr__special_nfet_latch.pm3.spice
at or near line 34Matthew Guthaus
06/21/2021, 11:56 PMLooking at this, am I understanding that maybe these are not valid bsim level 49 spice parameters? I see this:
http://literature.cdn.keysight.com/litweb/pdf/ads2001/spicent/snt0816.html
which describes the level 49 parameters. It has an "x" for each of them:
Ldif, Hdif, Rd, Rs, Rsc, Rdc, Nqsmod
Does that mean they AREN'T supported?
Matthew Guthaus
06/21/2021, 11:58 PMThey are included in other models such as the ASU PTM models
Matthew Guthaus
06/21/2021, 11:58 PM(except for nsqmod)
Matthew Guthaus
06/22/2021, 12:26 AMIf I comment those out, it seems to get further.
Matthew Guthaus
06/22/2021, 12:32 AMI do still get some warnings that I'm not sure about:
Xyce sh_stim.sp
Copy code
Netlist warning in file
sky130A/libs.tech/xyce/../../libs.ref/sky130_fd_pr/spice/sky130_fd_pr__pfet_01v8__tt.pm3.spice
at or near line 33
Device instance
XSKY130_FD_BD_SRAM__OPENRAM_DFF:X1020:MSKY130_FD_PR__PFET_01V8: Source
conductance reset to 1.0e3 mho
Netlist warning in file
sky130A/libs.tech/xyce/../../libs.ref/sky130_fd_pr/spice/sky130_fd_pr__pfet_01v8__tt.pm3.spice
at or near line 33
Device instance
XSKY130_FD_BD_SRAM__OPENRAM_DFF:X1020:MSKY130_FD_PR__PFET_01V8: Drain
conductance reset to 1.0e3 mhoe
Eric Keiter
06/22/2021, 2:56 PMHi @Matthew Guthaus; missing parameter errors are usually reported as warnings rather than fatal errors, so these might be no big deal. We are only able to implement the public domain versions of most compact models. So, if a parameter is present in the original Berkeley version of the BSIM3 or BSIM4, then we support it. But if the model has been extended by a CAD company, then we often don’t have an easy way to support those extensions. It depends on if those extensions are well documented or not.
Eric Keiter
06/22/2021, 2:58 PMI will take a look at the files you just attached.
Eric Keiter
06/22/2021, 3:04 PMI will also double-check the ngspice source. Possibly they figured out how to support those parameters. But in my experience they usually implement the Berkeley versions of these models, rather than attempt to reverse engineer proprietary extensions. If you’ve been running successfully with ngspice, and (it turns out) that they don’t support these params either, then I think those missing param warnings can probably be ignored.
Eric Keiter
06/22/2021, 3:04 PMBut, if ngspice does support those params, then it won’t be hard to add them to Xyce.
Eric Keiter
06/22/2021, 3:36 PMI just browsed the ngspice source code (version 34). It looks like they have added several ACM parameters to the BSIM3. There is a block of them in the b3.c file (which is where all the model parameters are handled). So, the bad news is that we haven’t supported those parameters in our BSIM3 implementation. The good news is that since it is in ngspice, it is easy enough to mimic what they did.
m
Matthew Guthaus
06/22/2021, 3:38 PMCool. Yeah I've not gotten into level 49 models at all so this is fairly new to me. Thanks.
e
Eric Keiter
06/22/2021, 3:40 PMThese params are new to me as well. I implemented our version of the BSIM3 back in 2001, 20 years ago (!) It looks like these parameters are related to a fancier way of computing the source and drain resistances. If the ACMMOD parameter is set to 0, then they aren’t used, FYI.
Eric Keiter
06/22/2021, 3:44 PMActually it looks like it is a bit more than that; there is an ACM method for source/drain resistance, another one for saturation current and another one for junction capacitance. It seems (at first glance) like all of them are modifying the original equations with more detailed geometric factors, which are specified in part by params like LDIF and HDIF.
m
Matthew Guthaus
06/22/2021, 4:36 PMThat makes sense that these were specifically for HSpice. It's interesting that these parameters were only in the SRAM transistor models. Perhaps because they were characterized with a tool that output them?
e
Eric Keiter
06/22/2021, 5:20 PMMaybe that is what happened. Interestingly, these ACM extensions have only been added (in ngspice) to the BSIM3 and not the BSIM4.
m
Matthew Guthaus
06/22/2021, 5:21 PMBTW, it is FAST 😄
e
Eric Keiter
06/22/2021, 5:21 PMcool!
m
Matthew Guthaus
06/25/2021, 10:58 PM@Eric Keiter I realized that the errors in the dff_demo above are still present. That demo appears to "run" but not sure what impact it has on results.
e
Eric Keiter
06/27/2021, 11:38 PM@Matthew Guthaus do you mean the “conductance reset” errors?
m
Matthew Guthaus
06/27/2021, 11:46 PM@Eric Keiter yeah
Matthew Guthaus
06/28/2021, 11:09 PMI upgraded Xyce to the newest release with your parser changes and I get some additional output from the dff test:
Copy code
BSIM4::Instance::processParams l = 1.5e-07
BSIM4::Instance::processParams w = 3e-06
Netlist warning in file
sky130A/libs.tech/xyce/../../libs.ref/sky130_fd_pr/spice/sky130_fd_pr__pfet_01v8__tt.pm3.spice
at or near line 33
Device instance
XSKY130_FD_BD_SRAM__OPENRAM_DFF:X1020:MSKY130_FD_PR__PFET_01V8: Source
conductance reset to 1.0e3 mho
Netlist warning in file
sky130A/libs.tech/xyce/../../libs.ref/sky130_fd_pr/spice/sky130_fd_pr__pfet_01v8__tt.pm3.spice
at or near line 33
Device instance
XSKY130_FD_BD_SRAM__OPENRAM_DFF:X1020:MSKY130_FD_PR__PFET_01V8: Drain
conductance reset to 1.0e3 mhoe
Eric Keiter
06/28/2021, 11:30 PMAh. The extra output.
Copy code
BSIM4::Instance::processParams l = 1.5e-07
BSIM4::Instance::processParams w = 3e-06m
Matthew Guthaus
06/28/2021, 11:31 PMFigured. But at least we can confirm this wasn't due to the initial scaling problem
e
Eric Keiter
06/28/2021, 11:31 PMYes, I agree.
m
Matthew Guthaus
06/28/2021, 11:32 PMI am getting reasonable results though, for whatever reasonable is. 🙂
😂 1
3 Views