Can't find Sources in QUCS

[Circlotron]

Version 0.0.20
There are tabs for Projects, Content, Components, and Libraries but nothing to show where AC and DC sources are located.  Where else should I look?

[Circlotron]

Hey, I got it. With the Components tab open there is a drop down box at the top where you can select Sources and various other things.

[Circlotron]

After stuffing around with QUCS and QUCS-s all day on both Linux and Windows 7 my considered opinion is they are both a steaming pile of crap.

So is Linux.

So is Windows.

In Linux the ng-spice executable is meant to be installed in /usr/local/bin/ but nope. It's installed on the system somewhere else because ng-spice actually runs but the file search thing can't find it. Ended up installing a non repository version and it only installed a library in /usr/local/bin/ and what's more you could see it in the file browser but the search thing STILL couldn't find that library...

Also tried compiling ng-spice from source and ./configure couldn't find a library that was actually installed.

And every time I tried to sim a cct with a diode rectifying the output of a transformer QUCS pooped it's pants.

Win7 version was no better. The default simulator .exe that supposedly installs with QUCS-s was nowhere to be found, and try to install ng-spice and it would throw up errors for misslng dlls, the last of which is not available separately from MS. ng clearly stands for No Good.

 
End of story.

[RoGeorge]

QUCS and SPICE are different.

It will take more than one day to switch from LTspice to Qucs, and it might seem cumbersome at the beginning.  Best intro will be to follow the QUCS manual and examples found in the QUCS user manual PDF.
https://qucs.github.io/docs/tutorial/getstarted.pdf
http://qucs.sourceforge.net/docs.html

Beware that there are many simulators called Qucs, and they can be very different:  https://www.eevblog.com/forum/eda/qucs-qucs-s-and-qucsstudio-simulators-are-not-the-same-thing/



I've found the most useful and easier to use to be LTspice and QucsStudio.  Both can be run under Linux, Windows, or FreeBSD, and probably Mac too (I didn't try on Mac).



Why do you need the sources, and for what type of Qucs exactly?
What are you trying to do exactly, what's the goal?

[Circlotron]

Well, thank you very much RoGeorge. 
Now I'm starting to get somewhere. With QUCS I was starting off with an AC source feeding a simple half wave rectifier with a choke input filter and a catch diode. QUCS would throw errors like "Jacobian circular" something... totally unfamiliar with the terminology so I can't accurately remember. Lots of others had the same problem and supposedly it was something to do with the simulator not keeping up with the sampling rate. Really can't remember. The computer screen had bullet holes in it by that time anyway.

But this QucsStudio looks like it might work, especially seeing it is still in ongoing development. The fact that it is in an all-in-one package makes a big difference too.

Again, Thanks RoGeorge.

[ve7xen]

In Linux the ng-spice executable is meant to be installed in /usr/local/bin/ but nope. It's installed on the system somewhere else because ng-spice actually runs but the file search thing can't find it. Ended up installing a non repository version and it only installed a library in /usr/local/bin/ and what's more you could see it in the file browser but the search thing STILL couldn't find that library...

The normal location for system-managed binaries on a typical Linux system is /usr/bin. /usr/local is for administrator-installed stuff. So if you installed it from the package manager, as would be typical, it was probably in /usr/bin. If you can run a command from the command line, you can use `which <command>` to get its path.

Also tried compiling ng-spice from source and ./configure couldn't find a library that was actually installed.

This is often because distributions often separate the development headers / static libraries from the dynamic libraries needed at runtime into separate packages. The headers and static libs are often in a separate package, usually with a name like `libsomething-dev` or `libsomething-devel`. Though obviously there are frequently other quirks like major version discrepancies, weird paths, etc...

Once I got the hang of it, I found qucs-s to be nicer to use than LTspice on my Linux system. It took some 'getting the hang of' though, for sure, though I actually think the UI is somewhat less cumbersome than LTspice that seems to have been designed before any modern UI conventions were established. If you prefer Windows, you may find the same with QucsStudio.

[Circlotron]

Righteo, got some results for a current doubler rectifier. also trying to sim a 3-phase version but it doesn't seem to like having three AC sources connected in delta. Voltages are equal and phases are 0, 120 deg, 240 deg. "Error in 'TR1': No DC convergence reached after 501 iterations. (1)"
Curent doubler and screenshot attached.

[Whales]

Transient simulation in QUCS can be a PITA.

"No DC convergence" suggests it can't find steady-state DC conditions to use as a starting point for the simulation.  Perhaps try adding some high value resistors in parallel with the two ideal diodes?

"Jacobian singularity" will be the bane of your existence.  Sometimes changing the timestep to be smaller can fix it, sometimes it means your circuit is unstable (oscillators in QUCs transient sims are hard unless you intentionally limit them!), other times changing the "minimum step size" in the transient sim's settings can avoid it.  You are given NO context as to what part of your circuit is causing the issue, you have to guess.

EDIT: I've only ever had luck with vanilla QUCS.  The other variants have never compiled or worked properly for me.

[RoGeorge]

Congrats!   

That warning lets you know it might be something to do about it, yet that is not a problem as long as the simulation finishes without errors.  Since the charts are in the schematic page, you can go in File -> Documents Settings and untick the "open data display after simulation" checkbox, then save, so it won't open an extra tab after each run.

it doesn't seem to like having three AC sources connected in delta

You didn't attach the trifazic schematic, so not sure what is that about.  Series/parallel ideal sources might give infinite currents or voltages.  Try adding the internal resistance of each AC source by drawing a small series resistor with each AC voltage source.

[Circlotron]

Turns out you can interconnect three AC sources in delta. Just make sure the voltages are equal and the phase shifts are correct. I ended up using star connected AC sources and three 1:1 transformers with the secondaries in delta. They are unnecessary though. This is the 3-phase rectifier that won't converge.

Edit -> there is no label on the output wire.

[RoGeorge]

You may try to allow for the minimum time step to go smaller than 1e-16.  I've made it 1e-19 and it runs (to change it, right click on "Transient simulation" then in the properties tab change the parameter).  Also added series resistors.

Changing the integration type or the max allowed steps might also help.  In "Help" -> "Technical Documentation" PDF at page 21 are listed a few workarounds against convergence problems.

If you add a small series resistor with each AC source, it should work connecting the sources directly in delta, without transformers.  Adding series resistors is a must in such a schematic.

Attached is your schematic tuned to run, and a second one running without transformers.