Your opinion about Amplifier sim.

[Dori]

I try to simulate an amplifier. The simulation works but I have a note that mention warning :

Note: starting Gmin stepping
Warning: Gmin step failed
Warning: Gmin stepping failed
Note: starting source stepping
Note: One successful source step
Note: Source stepping completed

I suppose this note is normal because I have made an Operating Point analysis, so for me Gmin Step can't be proceed. Because Gmin is a small signal transconductance.

Are you agree with me or you think I'm wrong ?

[destroyer X]

You have a bad contact...can you believe those things....nodes connection opened...and line is there...everything saying it is connected.

A bad solder joint even in virtual...this is too much realism...to turn users crazy.

reconnect....will work!

Carlos

[jleaman]

What is the program your using ? Got a link ?

[Dori]

Hi, jleaman.

The programm is protel v. 99SE
But I think a lot of CAD as Pspice, Cadence and so on use this kind of step analysis.

I find in help menu that but I am french and I am not sure about the meaning :
GMINSTEP :Sets the number of steps in the GMIN stepping algorithm. When set to 0, GMIN stepping is disabled, making source stepping the simulator’s default DC (operating point) convergence algorithm. Default=10 steps.

So I repeat my question :
Operating Point analysis => Gmin Step can't be proceed. Gmin is a small signal transconductance.

Are you agree with me or you think I'm wrong ?

[Giaime]

Hi, I'm having these problems too with Multisim. I found that it should be related to two things:

1) first of all, the lack of processing power on my PC, because I've tried with others and the same circuit worked well, even with the same simulation settings.

2) it has to be something related the calculation with a node short circuited: for example if I put a potentiometer, if I set it to the full 100% position or the 0% position, the simulation stops saying those kind of messages.

I just don't know why. I'm not an expert...

[jacco vermeulen]

I am not an expert either but it is logical that any program uses an algorithm for simulation calculations.

These algorithms are not that difficult to write, and one program or the other needs to have the same algorithm to come up with the same answer, right ?

For simulations to work the algorithm can be self-resolving.
Meaning: after one calculation the endresult is put back into the algorithm for a second calculation untill the set parameter is reached.
Another option is to set the number of calculations, the algorithm will then choose the size of the calculation step based on the number of steps.

For an operating point analysis a calculation step can not be made, so step number is 0, just the one calculation is done to get the operation point number.
If you do not set it to 0, default is 10, so it will make 1 step and then terminate.
(standard error with finite element algorithms)

I have no trouble with EW MultiSim2001
(running XP Pro)

[wimms]

Originally posted by Dori
I try to simulate an amplifier. The simulation works but I have a note that mention warning :

Note: starting Gmin stepping
Warning: Gmin step failed
Warning: Gmin stepping failed
Note: starting source stepping
Note: One successful source step
Note: Source stepping completed

I suppose this note is normal because I have made an Operating Point analysis, so for me Gmin Step can't be proceed. Because Gmin is a small signal transconductance.

Are you agree with me or you think I'm wrong ?

Wrong. Failures in gmin stepping hints that there is something fishy in the circuit and mathematical methods used in the spice are unable to achieve converging results (it is reaching contradictory state or insufficient precision) within constraints configured. In short: you can't trust the results. May be indirect warning that some element would fail. Most probably occurs when currents/voltages in circuit tend to reach levels outside of model meaningful behaviour (like 100Amps through BC547). That includes, but is not limited to, short circuits, unconnected nodes, nonlinear feedbacks, instabilities within selected timestep precision, etc.

Source stepping is a "backup" method, and it succeeded. Still I'd suggest looking into the circuit and try until it resolves without warnings.

http://www.elecdesign.com/Articles/I...leID=8282&pg=2
Taking A Peek Under The Hood Of Your Spice Circuit-Simulation Engine
Understanding Spice numerical methods, convergence issues, and innovative simulation techniques can rev up your circuit design expertise.

That was the very first result google offered when searching for "gmin spice"

[Dori]

Thanks for your information wimms.

I had search for Gmin step in google and read about this interesting article. But in fact, when you are working on a complex project alone (and I am also searching for a job, in France, it is very difficult !!), you don't have a lot of time to understand how a simulator works in detail, and how it creates the impedance matrix or things like that.
That's why I'd want to know other opinion to save time.

Good night folks !!

[jacco vermeulen]

c'est difficile a ce moment en France, pour les ingenieurs ?

j.

[Dori]

Ben ouai, ou alors je n'ai pas le comportement qu'il faudrait avoir en entretien pour inspirer confiance aux recruteurs. Ca fait un an et demi que je recherche (spécialité Microélectronique), j'ai eu une douzaine d'entretien, mais toujours sur le carreau.
Et je suis pas le seul dans ce cas la. C'est super chaud pour les jeunes diplômés la france dans mon domaine. D'ailleurs si vous avez des contacts ou des tuyaux pour travailler à l'étranger, je suis preneur.