Circuits that shouldn't work...?

Thye do work but you need to supply an impulse to get them going.

Tam

I've never needed to use an 'external stimulus' to start an oscillator in LTSpice. It always delivers the goods.

It appears to have some kind of noise stimulus built in. Some other spice variants (notably, Circuit Maker) seem not to do this, and end up with circuits balanced on the head of a pin, so to speak.

Yes, that is what I meant. An unstable equilibrium. Thanks for the clarification.

If you want a circuit that should not work, how about an emitter follower with a voltage gain of almost two...

You mean the "point" of the pin...

What makes you think there is any correllation whatsoever with simulations and real world performance?

Simulations are valuable, provided you already understand the circuit well enough that you don't really need to simulate it.

John

And already know where the critical node is that needs the ".IC=0V" :)

Err... 1000s of 1,000 transistor ic designs that work first time, and are made in millions per month. Modern ic design wold be quite impossible without simulation.

This *is* the reality mate. Sure, simulations *can* fail if you don't model correctly, however, they also demonstrable work as a matter of routine.

Kevin Aylward snipped-for-privacy@anasoft.co.uk

So does LTSpice use a different or improved engine, but works with ordinary Spice models?

All I ever use is the good old original free Spice on Unix, and I always thought that the derivatives (like PSpice) used the identical engine with some added GUI sugar.

Clarification welcome.

robert.

Hello Robert, LTspice is originally based on SPICE3F4, but then it has been practically fully recoded by Mike to make it better and also to overcome the bugs in the original code. It has also many new intrinsic models required for fast simulations of SMPS regulators.

LTspice is 99% compatible to PSPICE syntax for the analog elements of SPICE. You can practically use any 3rd party opamp model in LTspice. It's just a very few mouse clicks and a ".include modelfile" SPICE-line.

Digital components are of course very different, because there is no accepted SPICE-standard for digital elements. Every SPICE-vendor "cooks its own soup" for it. LTspice has the A-elements which are very powerful models for digital and mixed-signal components.

See my comments above. Linux users can run LTspice under Wine.

LTspice is already replacing PSPICE in education, because LTspice is fast, very stable, unlimited and freely available. What can we wish more?

There is a very strong user group for questions regarding LTspice.

Best regards, Helmut Moderator of the LTspice Yahoo group

There are several classes of circuits that often do not model well in Spice, various (often simple like Hartley or Colpitts or three logic inverters and a crystal) oscillators is a typical example. What irritates me is that LTSpice badly fouls up on simple "brute force" power supplies. No other Spice that i have used fouls up in this way.

Well? It's "faster" ;-)

...Jim Thompson

It seems to work just fine here. Where exactly does it foul up?

Jeroen Belleman

What are "good results" for that (useless?) circuit?

Don

In article , JosephKK wrote: [...]

Fouls up how?

I've modeled:

---->!---- ! ! (V) --- 0.1F ! --- GND ! GND

With no problem.

I've also modeled:

K1 ------ -------------+-----------+--------- ) ( L2 ! ! ) ( ! --- C1 ) ( ! --- L1 ) +-- GND ! ! ------ ( ! GND ( ! ( ! -------->!---

Where (IIRC):

L1 = 100mH + 2R L2 = 1mH + 0.1R K1 = 0.95 C1 = 1000uF + 0.01R

... and got good results.

try a standard full bridge from 120 60 Hz into a 5000 uF cap and 240 ohms load.

Looks quite reasonable to me. What *is* your problem? Give us something to sink our teeth into. Vague statements that something doesn't work aren't productive.

Jeroen Belleman