This problem actually already shows up with the ps output so apparently that means a frontend bug? pstoedit -f ps -page 1 2023-12-31.ps test.ps
DXF output contains very jagged text edges
sharedspice vector saving behaviour unclear
I think there are two different viewpoints here. The physicist viewpoint and the business viewpoint. I agree from the physics viewpoint that if the error margin on the results is within the error margin of your modeling parameters, you're doing pretty well. The business viewpoint is that in a chip design where millions are at stakes, the question you will get is not if your simulator produces physically accurate and meaningful results, but if it EXACTLY matches what they were using before. So the...
From my understanding of discussing the issue with the Xyce developers, the CDC tried to standardize device models, and for a commercial simulator to pass the test it needs to be within 1e-6 relative error. So it appears that pretty much the whole industry has standardized on the constants that spice3 used, weather they are exact or not. Existing models have been characterized with those "wrong" constants, so changing the constants later breaks existing models. I expect you will find that using the...
[PATCH #41] breaks model compatibility with spice3
Hm I wanted to test this fix, but the latest git version doesn't install headers for me using this package https://aur.archlinux.org/cgit/aur.git/tree/PKGBUILD?h=ngspice-git I'll debug that another time, but I assume it'll work.
Ah we were writing at the same time. Great!