[Solved] Issue with simulations nightly 7.99

Hello everyone, great thanks for the huge improvement made in the simulation part of Kicad 7.99. FFT, measurement built-in in the new simulation box, search y at x with the cursor, fourier analysis pressing a button, bias point directly on the circuit, all it’s awesome and time saving.
I know that it’s still early, all in development, but I noticed some issues and I don’t know if it’s my fault or not. I am knew to Kicad, but not to spice.
My version of the nightly release is of 11/08/2023, I am on Windows 11 (AMD Ryzen 5700u, GPU Radeon).

  1. Sometimes when tuning a component, I can’t see the cursor in the cursor box. Actually the box is somewhere in the plot section (Solved)

  1. During AC analysis of a simple resonant RLC circuit, I can’t see any current in the signal panel and in the user signal section. But I found strange to see that in the user signal section there was a “c1#branch” (and “r1#branch”, “l1#branch”) that gave me exactly the current that i need (i verified it in orcad capture pspice) but the unit of measure are all in voltage (and dbV), also for measurement. (See answer below also)

  1. In the user signal section, the signal literally disappears and the graphics here is continuously flickering (maybe the guilty one here is the GPU) (Only on Windows, not WSL GUI)

  1. When selecting AC analysis I cannot see the dialog to choose between “linear”, “decade” or “octave”.

  1. There isn’t anymore the “custom” spice directive in the analysis panel. I don’t know if it is intended since a really good number of simulations are resolved using the GUI in Kicad 8 (again, thanks for improving it) (Solved)

  2. Is there a way to change the plot axis between db and linear in AC?

  3. Trying OP, TRAN, FFT, POLE-ZERO, I found Kicad a really good GUI for ngspice, but I have problem with the dc transfer function (.tf), I can’t see any results in the output (maybe I am wrong, but I remember that Kicad 7 could print the results).

Sorry for the english and for the long explanation. Could you give an help where there are my mistakes please?
Thanks for the attention.

(1) should be fixed in the next nightly simulator, TUNER_SLIDER: fix incorrect wxWindow parent. (f0d18adf) · Commits · KiCad / KiCad Source Code / kicad · GitLab

1 Like

I’d like to add my opinion that the usage of “dBV” in the AC simulation tab is not helpful from a user standpoint. I understand that the ngspice simulation routine seems to use 1V input for AC, but it does so automatically without the ability for the user to choose another input level, and the point of the AC simulation presentation is not to report absolute signal levels in the circuit (dBV means decibels relative to 1V, thus a value in unit dBV is an absolute signal level), but to report the relative level of signal in different parts of the circuit, where the appropriate unit is “dB”.
I would therefore prefer that the unit “dB” instead of “dBV” were used in the AC simulation tab.


(5) I am stupid, I just needed to write on the schematic, and instead of clicking on the new simulation profile, I should have clicked on the manage current simulation button. Sorry for that.

The only problem that remains is with the .tf spice directive, which does not output anything. However, when using external ngspice/ltspice, I am able to see the results.

In ac simulation you may simply change the ac voltage, e.g. by the command ‘ac 2’ instead of ‘ac 1’. However, this does not matter: The circuit is linearized around its operating point, if you double the exciting voltage, all resulting node voltages from the ac simulation are doubled as well.

Ok, I see what you say. ampl=1 is for TRAN and ac 1 is for AC analysis.
Then it follows, that the X axis title should not be “Gain” but “Level” or a suitable synonym.

You have gain and phase, both relative to the input. If AC=1, it is gain (output/1) by the number.

Today I delved a bit deeper into the issue.

(3) On Windows Subsystem Linux with GUI, there isn’t any flickering/disappearing problem in the user signal section.

(2) I am able to select the current flowing in the RLC series, either via the probe or the user signal section writing “i(c1)” (same with r1 and l1), but the plot it’s still in dbV.
Now I have two signal that measure the same thing, c1#branch and i(c1), but if I make some measurement on c1#branch I will have Voltage for both the current and the frequency, while on i(c1) I will have Ampere on current and seconds on frequency.

Finally I noticed a problem with measurement of voltages, there’s not consistency between looking at the plot with the cursor and make spice measurement (these results are also wrong)

Maybe it’s something trivial that I can’t perceive at the moment, do you experience these problems?

Yes, the numbers presented in the AC plot coincide between output level and gain when ac=1, but for any other number, but for any other ac=x the presented numbers will not coincide with the gain, and therefore the label should be “Level” instead of “Gain”.

My main point is that the labeling of the X-axis should be corrected.