Semi OT: engineering education

I can't speak to the ratio of "gets design work" vs. "flips burgers", but in the engineering departments that I worked in there was maybe a 2:1 ratio between software developers and software QA, and maybe one or two outside support staff who were of engineering but not actively working in new designs.

And the 2:1 developer:QA figure in software is deceiving: the Software QA guys are generally doing exceedingly important work, and in a perfect world there'd probably be a 1:1 ratio and product design would probably be cheaper overall if you did it that way.

My guess (based on close to 57 years in the biz) that 99% can't find their ass without a GPS, a flashlight, and an assistant to hold the flashlight.

I'm this close to dumping on a client whose engineers can't even comprehend what a spec says... reading comprehension = big fat zero. ...Jim Thompson

I have no real numbers, but I estimate that a small fraction of EE grads ever design electronics at the component level. 10%, maybe less? Maybe much less.

I see a lot of "software engineers", project engineers, QC people, specifiers, test engineers, system engineers, logistics people, managers.

Many really big companies don't design their own electronics any more. They buy it or pay consultants to design stuff.

I think your estimates are generous. You can find people who know how to do stuff. The problem is finding people who can figger out WHAT to do. Even rarer is the manager who can appreciate what his brightest engineers are telling him.

I had an excellent engineering education. Even more important was that my first industrial position presented me with a really good mentor. I learned more practical stuff the first week on the job than in 5 years of college. He taught me how to create an electrical solution from three transistors, a water polo analogy and some lemon pudding. MacGyver had nothing on him.

When I interviewed applicants, I learned to quickly verify what they knew, the proceeded to investigating their capacity for extrapolation. I'd let them tell me about their favorite design. Then, I'd introduce a twist and asked how they would have changed the design to accommodate it. Most fell flatly on their face when presented with something new.

Some percentage (in any field of endeavor) know how to operate an appliance, like a simulator, but have no clue what to simulate. They can't extrapolate what they know, or create anything new.

Those same people become customers. They think they know what they want, but can't put it into a useful format.

I want to see the acceptance criteria and complete test specifications that they are going to use to verify the design meets their expectations.

If it's a product, I want to see the user manual. YES, you CAN write the user manual before you build the product. And that process works extremely well to constrain the task.

If you know what the end user wants to see and the test criteria for the design, you have half a chance of making it work.

Much less. 90% (or more) do software. Subtract managers and there are far fewer hardware jobs. Fortunately, there are even fewer who can do it. ;-)

I don't see a lot of that in large companies, anyway. If there is a specialty they need but don't have (or don't want to have) in-house, sure but they normally want their bread and butter stuff covered by regular employees. Of course they fill gaps with contract help, too.

I'll never forget when asked to do the first design in labs as an undergrad, only 2% of us knew what to do and got on with it, the rest looked like rabbits caught in headlights.

How much and in what way people understand the course material given that shortcoming in understanding real world physicals I don't know - I'm inclined to be doubtful but who knows.

NT

I had the benefit of growing up in a Radio/TV repair shop... with attached hardware store ;-)

So, by the time I got to MIT, I knew what I didn't know, and what I needed to know to understand how those kits I built were designed. ...Jim Thompson

I got my start with electronics via model trains. We weren't particularly wealthy, and if my power pack went on the fritz, well, I couldn't just bug Dad for a new one. Someone was on the hook for replacing that bad electrolytic...

Nowadays with trains you have all sorts of cool technologies, like digital command control that lets you operate some absurd number of locomotives on the same piece of track independently with no special wiring. When I was 10, if you wanted more than one train on the same layout it required a complex arrangement of block switches and relays.

Which is fun for the consultants. ;)

Cheers

Phil Hobbs

And fiber pins connecting rail sections ;-) ...Jim Thompson

Absolutely. There are zillions of smallish science-related companies that need help with combined physics and electronics. It's impressive how many scientists are helpless with electronics.

I think consultants get a wider exposure to measurement and signal processing problems than in-house engineers, more cross-fertilization. We want to manufacture stuff, but we get a lot of that same exposure. Even the failed projects are educational.

Yeah! I remember that. You had to gap the rail junctions on the blocks..

And most modern systems will work how well without software or a unified system design? I've had the misfortune of working on big, complicated systems that were designed with a philosophy that started by saying that designing the whole system from the get-go was a waste of time and energy. It's amazing and appalling how much time and energy goes into making such systems actually work in the field.

Bad for them, good for me (for the time being).

When I was a kid my bedroom was the attic... with a built-in 4'x8' model train layout. It was Lionel O-gauge... couldn't afford any of that fancy HO stuff.

When I got to MIT they had a model train club, layout occupying several rooms in Building 20... you crawled under to get to operating stations. ...Jim Thompson

I think MIT still has the model train club. At least, they did last I looked. They did it in a very MIT way (i.e. extremely technologically sophisticated.)

By the time I got into trains, HO was the standard "consumer" grade (along with the somewhat less popular N scale.) At that point O gauge Lionel was (and still is) the gauge for collectors and enthusiasts with big bankrolls.

The third power rail down the center never appealed to me.

The Franklin and South Manchester railroad is one of the best examples of what can be accomplished with HO scale in a relatively small space:

Nice layout! And nice videography! ...Jim Thompson

Just like real trains. ;-)

Even at HO (1/87th) scale, the big issue with modeling is always space compression. Even large basements can barely accommodate much more than a couple scale miles of track, and forget about accurately doing 3% grades on hills...