I am sick of running into MultiSim's bugs

[Euler's_Identity]

I was introduced to MultiSim many years ago via Electronics Workbench.  I had the opportunity to get in on the ground floor, so to speak, as, because I was a student (and through my professor), I was able to purchase the instructor's (not education/student!) edition of the MultiSim/Ultiboard package for a reasonable price. Since then it's been one bug after another.

Now, I'm years out of school, can afford the huge pricetag, have the lastest version via grandfathering, but I don't want it, because I don't trust it any further than I can throw it. (I swear I'd buy the professional if it worked, as expensive as it is.) I've bought the service agreement, now that National Instruments got involved, but I don't believe I'll ever get anything but the latest bug-ridden kluge.

I've had a bug under investigation for two weeks now and, yeah, maybe it'll be resolved, but, honestly, it won't matter.  Why?  Because I'll just turn up another one and go through this mess again.  Why?  Because I've been doing this for years, nearly a decade.  You get what you pay for.  If you want something that works, you have to pay for it, and MultiSim isn't it.  I just feel for those who actually shelled out the money for this latest version that I got for pennies.  But I swear I'd buy the thing if it worked. It just doesn't work.

Sorry, no time for proof read.

[Euler's_Identity]

With the above said, I wanted to come back and clarify my intention here.  I want MultiSim to work.  It would be so great if it did.  It just never does.

The power of a simulator is the ability to blow up stuff without cutting into the budget for the parts.  It's about saying, "what if I do this?" in a situation that'd be risky in reality.  But then you don't take out an expensive component or entire circuit in the process!

That's why I hate to see MultiSim having these bugs like it does.  It's such an easy program to use -- it's much like playing at a bench in many ways.  Now if it'd just work!

But it's like core pieces are flawed in some way, some way that doesn't become evident to MultiSim's creators right off the bat.  How will I know whether my results are true or just another MultiSim bug?  Sometimes it's readily apparent when MultiSim's burped, but what about when it isn't?  What about when that current is possible or that voltage could've been produced?  It's this constant uncertainty that enters that makes one want to scream sometimes.

So, lest anyone out there have convinced him or herself I'm just a troublemaker who wants to bad mouth MultiSim, I assure you I am not and do not.  I so much want to be able to trust MultiSim.  It is the best simulator, if it would only work.

[Andrew_Boniface]

I have had some experience with handholding Multisim to get it to work.

 Initial conditons seem to be a problem (and seem to be a problem you have with a circuit you posted about involving flip-flops).  I have read something about how to set initial conditions in multisim, but it seems to be much more work than some other simulators I have used (Microsim, I think).  Usually in the circuits I have used, the best method I have found is jury-rigging the circuit to cause Multisim to choose useful initial conditions (Phase shift and other positive-feedback oscillators in my case).

The other problem I have seen, is that, since Multisim uses very general numerical methods, circuits which require a more specific approach (say, switch-mode power, or amplifiers involving choppers) tend to perform very poorly.  This is a problem I am working on.  I expect success for methods custom-designed for a single circuit.  I think I can make some of these methods adapt to a certain range of circuits, in fact my eventual plan is a wizard permitting design of a certain family of circuits, which desings the simulation method in tandem with the circuit.

To take a "random" (not of a limited family) circuit, and recognize that general methods fail, is already a difficult pattern recognition problem (for a computer, the operator usually recognizes the "freeze" easily enough).  An attempt at an automated fix is harder yet, even for a small subclass of problems.