Theory vs. Simulation

Discussion in 'Off-Topic' started by Nanophotonics, Apr 8, 2010.

  1. Nanophotonics

    Thread Starter Active Member

    Apr 2, 2009
    365
    3
    Hello,

    Many of us make use of circuit simulation software packages to confirm our theory after we've derived the component values and drawn the schematics, whether it's a student working on a new concept or an expert design engineer working on complex circuits in industry.

    How would you guys deal with inconsistencies between theory and simulation results? Which one do you think gives more realistic results? Given your maths is correct.

    Thanks.
     
  2. steveb

    Senior Member

    Jul 3, 2008
    2,433
    469
    It would generally depend on how sophisticated the theory is; however, I think most people try to use as simple a theory as posible to capture the important effects and develop an intuitive understanding, as well as simple design formulas. Simulations generally take theory to a higher level by number crunching a more detailed model that would be too confusing for a person to deal with. For this reason, my opinion is that the simulation gives more realistic results. However, it does not give a better understanding, which is why both are important tools.

    I think the way to deal with inconsitencies, assuming they are enough to bother you, is to develop a more detailed theoretical model. The final step is (of course) to build the real system because there will definitely be some inconsistencies between reality and theory/simulation. If there doesn't appear to be any inconsistencies, it just means you haven't looked hard enough.
     
  3. Nanophotonics

    Thread Starter Active Member

    Apr 2, 2009
    365
    3
    It's interesting to know that the simulation software works with a higher level of the theory.
     
  4. studiot

    AAC Fanatic!

    Nov 9, 2007
    5,005
    513
    Calibration against reality - As with models in any branch of engineering.

    Neither.

    They are both only as good as the supplied information

    GIGO

    Garbage in = garbage out.
     
    Last edited: Apr 11, 2010
  5. JoeJester

    AAC Fanatic!

    Apr 26, 2005
    3,373
    1,159
    Realize there are errors. I've seen simulations of predictive transmitter coverage areas go bizare results which was a result of the mathematical process at that particular point in the area.

    Everything is taken with a grain of salt. Some grains are alot larger than others.

    Models are just that, models. Nothing is preventing you from modifying the model to closely replicate what you've seen in the lab over a few thousands of actual measurements.
     
  6. retched

    AAC Fanatic!

    Dec 5, 2009
    5,201
    312
    That about sums it up.

    Using models will give you a good starting point for the design phase of most any circuit. However there are always unknowns and "fidgitty" parts that you just have to know. If you spend a year on a model, right next to the physical circuit, you can get pretty good. But then someone comes along and plugs a radio in on the same power source and things go haywire in real life.. But the model keeps chugging along..no problems..

    You can consider models like an engine diagram for a car. You can read about, and learn how to change the spark plugs.. but you will never have to bandage your knuckles with a model.
     
  7. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    My simulator is between my ears. It occasionally backfires, but mostly works OK.
     
  8. Blofeld

    Active Member

    Feb 21, 2010
    82
    18
    There are some problems where I would trust the simulation, because the algorithms that have to be used are relatively simple. One useful distinction is between linear and nonlinear problems - the linear ones are generally the easy ones to simulate.

    Consider an arbitrary number of R, L and C and connect them in any way you like. The resulting network might not be easy to analyse for a human on the first sight. But there exists an algorithm to generate a set of (complex) linear equations from it. Now during several centuries the math guys have developed accurate and efficient methods to solve sets of linear equations. So people know how to program this very well. Apart from stupid bugs in the program, I would trust a simulation that only has to take into account such linear elements.

    On the other hand, consider some highly nonlinear devices that are connected in an arbitrary way. Now there is no guarantee that the simulation can solve the nonlinear equations of the resulting circuit in such a way that the results accurately describe the real circuit. Not even under the (highly optimistic) assumption that you have accurate models of each individual device. The simulation might send you a warning message if it has problems analyzing the circuit, but even this is not guaranteed. So if in this case there is a fundamental discrepancy between what I expect from theory and the output of the simulation, I would probably trust the "simulator between my ears" as Bill Marsden said it.
     
  9. retched

    AAC Fanatic!

    Dec 5, 2009
    5,201
    312
    You cant beat experience. Not with a stick, not with a brick, and not with a simulator.

    Now, HAVING the experience along with a simulator, ala SgtWookie, is a very powerful tool. If you know the non-linear actions or reactions of a component, you can simulate other parts of the circuit with those in mind. That ultimately gives your "Between the ears" simulator another tool for design. Especially when you are dealing with a large component count or high voltages. I find it nice to use discretes to simulate a function of a uC. This allows for using a simulator like a breadboard, or any other design tool.

    If you dont know the limitations of the tools you are working with, ESPECIALLY your brain, you will run into problems.

    A bench with a breadboard, simulator, and BOOKS is a good design approach.
     
  10. R!f@@

    AAC Fanatic!

    Apr 2, 2009
    8,754
    760
    I like the Sgt tool part...
     
  11. steveb

    Senior Member

    Jul 3, 2008
    2,433
    469
    Be careful how you say that. :p
     
  12. R!f@@

    AAC Fanatic!

    Apr 2, 2009
    8,754
    760
    I am ....:(....he has a big gun and he knows how to use it.
    I don't want to be at the other end when Sgt gets charged :D
    donno how I will end up :p
     
  13. Nanophotonics

    Thread Starter Active Member

    Apr 2, 2009
    365
    3
    I like that approach. Students now have to learn all sorts of simulating tools. I can only understand their growing importance and the ease they provide in running a quick design check and grasp the essence of fundamental concepts.
     
  14. Nanophotonics

    Thread Starter Active Member

    Apr 2, 2009
    365
    3
    I use simulation a lot, but there's something I've learnt, GIGO, as studiot said.
     
Loading...