0
\$\begingroup\$

Is there any workaround to force LTspice integrate backward in time from the \$t_0\$ to \$-t\$, for some \$t_0,t\in\mathbb{R}_{+}\$?

\$\endgroup\$
4
  • 1
    \$\begingroup\$ Are you, perchance, seeking to implement a convolution integral? If yes then you can simply use an input signal and an appropriately modelled kernel (filter) and simply run the simulation. If not, what is it for? \$\endgroup\$ Commented Jul 17, 2022 at 7:53
  • 1
    \$\begingroup\$ No, nothing like this. My aim is to check how the system evolves toward unstable invariant manifold \$\mathcal{W}_\mathrm{u}\$ as time \$t\rightarrow -\infty\$. It is worth mentioning that I already used some robust integration schemes in Matlab but time reversal failed terribly. \$\endgroup\$
    – struct
    Commented Jul 17, 2022 at 8:17
  • 1
    \$\begingroup\$ In that case, I'm afraid I have some bad news... at least as far as LTspice (and SPICE, in general) is concerned. \$\endgroup\$ Commented Jul 17, 2022 at 9:04
  • \$\begingroup\$ If the problem space is LTI, then simply inverting the poles/zeros does the job. \$\endgroup\$ Commented Jul 17, 2022 at 16:54

1 Answer 1

1
\$\begingroup\$

This is not possible directly. You could change your circuit to make +t emulate -t by using negative components or controlled sources etc. However this is likely going to be difficult to achieve convergence.

For instance, a capacitor has i=C.dV/dt; You could use a negative value of capacitance C (SPICE allows this !), to emulate the behaviour of a positive capacitance in the -t direction.

\$\endgroup\$
1
  • \$\begingroup\$ this sounds really clever, thank you I will give it a serious try \$\endgroup\$
    – struct
    Commented Jul 19, 2022 at 15:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.