12 responses to “The Model Engineer”

  1. Fluxor

    RE: Chris not using a network analyzer in school. Why is this a shock? It’s certainly not as ubiquitous as an oscilloscope.

    Anyhow, on the topic of modeling. I never physically build my own circuits. It’s not possible. So in a way, all I do is simulations with models. I then give instructions to the test team on what to look for. I rarely “get my hands dirty” unless something is really really really wrong. But my type of modeling is pretty close to reality (most of the time).

    There are some I work with that do all their modeling with Matlab at a high level of abstraction with larger systems. Not all of them are well grounded in reality. For example, one of them told me a certain signal path requires a 12-bit digital-to-analog converter. The model tells them it’ll give them an optimal solution. What they don’t realize is with many of their requests is the cost involved — in power, in physical size, in design time, and in design risk. All they see is what Matlab spits out. 12-bits may be optimal, with plenty of built in margin. But is it necessary? Is it a wise trade-off versus other competing specs? The bigger picture is often lacking. That’s where teamwork and the ability to be flexible is important. If the guy doing the modeling is a obstinate mule, then what ends up happening is a lot of heated arguments and finger pointing about how “the other guy” doesn’t know what he’s talking about.

  2. paul hopwood

    Sometimes engineers forget how much of what we do on a daily basis is actually just a model or an abstraction of an idea, we just apply a new model & go about our daily business.

    Ohms law, the formula known by all ee’s isn’t actually guaranteed its just a model for ohmic materials … And not all materials & conditions are ohmic.

    On the flip side, I know there’s stuff I’d like to look at in way more detail to optimise & get the best solution but I guess most engineers work for commercial companies where money & profit are the religion!

    1. Stanley Ma

      Re: Ohm’s Law. Good point. Even if a material is consider “ohmic”, V=I*R is better modeled as a stochastic process of electrons flowing through an idealized plane due to an electric field. But then, the stochastic process is simply a model for us to get a handle around seemingly random events. That’s how we end up being able to model thermal noise of a resistor.

  3. GEARS

    I agree *somewhat* with Miss Outlier’s sentiment. That is, modeling when you have no idea what you’re modeling and it’s easier to built it in the lab is pointless. As systems get more complex, and costly, modeling some aspects becomes an inevitability. Plus, this is all highly field dependent. Specifically when dealing with things like microfab.

    With that said, I don’t think the stereotype is nearly 100%/0% experimental vs theoretical (or vice versa). Rather it’s a 67%/33% depending on your flavor. I prefer experimental problems but they’re going to need some modeling aspect either for optimization or simple verification.

    Oh, and I just looked at my papers and 30% of them are based mainly on modeling complex systems… and I consider myself highly experimental. Go figure.