At the risk that I'm not the first asking here: do graphs in xschem with a log x scale work? I seem ...

Christoph Maier

04/09/2024, 12:28 AM

At the risk that I'm not the first asking here: do graphs in xschem with a log x scale work? I seem to be able to plot g_m/I_d vs. I_d on a linear scale, but not on a log scale.

Christoph Maier

04/09/2024, 2:48 AM

I can work around the buggy log x scale, but is that really necessary, or am I just missing something basic?

Stefan Schippers

04/09/2024, 7:05 AM

Usually when you switch to x log scale do a 'full zoom' (f in the middle of the graph area) to rescale.

Christoph Maier

04/09/2024, 2:37 PM

Wors sometimes … but as I'm trying to plot noise spectra and dc sweeps, xschem graph plotting seems to get confused … or I am confused about the finer points of scaling. log log output noise with

set sqrnoise

results in confusion. With gaw, I get meaningful graphs for both dc sweep and noise, although not simultaneously.

Stefan Schippers

04/09/2024, 8:54 PM

@Christoph Maier can you please share the raw file and the variables you need to plot? So I can do a test. There are gazillion plot types ngspice sets in the

Plotname:

line and may be this is yet a new plot type I need to add.

Christoph Maier

04/09/2024, 11:16 PM

@Stefan Schippers, will this do? https://github.com/tatzelbrumm/sky130_cm_ip__biasgen/blob/main/xschem/pfet_g5v0d10v5_noise.sch https://github.com/tatzelbrumm/sky130_cm_ip__biasgen/blob/main/xschem/simulation/pfet_g5v0d10v5_noise.raw (two committed versions, with sqnoise enabled and disabled … both should work)

Stefan Schippers

04/10/2024, 8:50 AM

Thank you will look asap

Stefan Schippers

04/10/2024, 11:01 AM

I have tested your example with set sqrnoise enabled. I have added some options to increase the precision since you are dividing by derivatives of almost constant signals (vg) I have added in the schematic the same graphs plotted in ngspice plus the spectrum graph. Added an image in the schematic of the ngspice plots, they seem to match. This example was useful to do a small change in xschem since due to random rounding errors in the ngspice generated data it did not correctly recognize wrap arounds (see following image) DC plot is still visible though.

pfet_g5v0d10v5_noise.sch

Stefan Schippers

04/10/2024, 11:01 AM

2.png

4 Views

Open in Slack

Previous Next

What is Linen?

Linen is a search-engine friendly community platform. We offer integrations with existing Slack/Discord communities and make those conversations Google-searchable.