How do engineers know how to build stuff?

Ever look inside a VCR? or a TV? or any other type of electronic device? Look at the circuit board. How the heck does an engineer know what to put where and how many? If I wanted to build a new VCR, where do I start designing it?

Since you are designing a device that already exists you have to have a better idea or a new twist on an old one.
Montgomery Ward made money for a long time by using Zenith and Rca parts in their Admiral TVs.My guess was that was a legal thing. Someone had to design the chassis so that the different circuits would have the correct power to work. Also safety was a concern. At 25,000 volts X radiation exists. It must be safely contained.
More to the point they have the knowledge and most of the time the experience in a particular "science"to create.

>> How the heck does an engineer know what to put where and how many?

That’s what they teach you in engineering school. Or do you think the purpose of college is just partying?

Actually, that’s what I learned as an entry-level engineer (learning how to design that is, not partying :)). What I learned in school were the tools needed to learn how to design, not actually how to design anything.

Anyway, you don’t design something complex like a VCR all at once. It’s broken down into functional blocks, which are further broken down until you get to small, manageable chunks, some of which can even be done by entry-level learning-on-the-job engineers (the senior engineers are the ones who can tell the difference between the easy and hard tasks- it’s not always obvious!) Also, you never “re-invent the wheel”- if a functional block has been designed before, you adapt that design to meet your needs instead of starting from scratch. Finally, experience speeds things up greatly. If you’ve designed similar products before, you can often develop a good intuition for how to proceed on new products without a lot of guess-work and backtracking.

Actually, that’s pretty much how software is written too.

Arjuna34

I would say pretty much that is how anything is done. When you encounter a problem for the first time, the first thing to do is to ask yourself “how did other people deal with this in the past?” You do not reinvent the wheel is someone already did… you try to build on what others have done, whether you’re designing a VCR or installing a new roof.

A Big Book of Pre-Invented Wheels is a pretty standard item for design engineers of any stripe–it’s full of things (circuits, gearboxes, what have you) that have been used before and found to work. You take a power supply here, a servo control circuit there, a microprocessor from the next chapter and stick them together. Then you tweak the design until it does what you want it to. You don’t generally learn this in school. You just learn the stuff that lets you understand the designs, so you can modify them effectively. I sometimes design circuits from scratch (when I want to do something so weird that it doesn’t show up in the books). These circuits are much simpler than (for example) an entire VCR, and they still take a lot of time and effort.

I think Arjuna34 explains it rather well; when one starts out (like I did a year ago), it’s relatively easy to look at a finished product and understand what’s where and why it’s there. It’s slightly harder to figure out how’d you want to set it up if you made your own, and the most difficult to come up with something not done before.

As to the actual process, I can give an overlong description : Someone comes up with an idea of what to make, or someone comes up to you and says, “I want a phylotonic pulse confibulator that can gleep burkles, too.” So someone (or several people look at existing pulse confibulators, and existing burkle-gleepers. What they get from this is the general idea, to make a functional block design. (When beginning to design a bridge, engineers don’t look at existing bridges and say, “we’ll need 35 miles of 1-foot thick steel cable, 2 million lbs. of concrete, etc.” they say, “we will put a deck, two or three posts here in the middle and anchor it in the rock at the south end.”) Sometimes you have to figure out how to add burkle-gleeping without modifying the confibulation part too much. And even if it’s never been done before, you know how to produce this pulse or that timing circuit or whatever to get the effect you want.

And often enough, the blocks themselves are simple and what you’ve learned about in school. Maybe you can re-use existing parts easily, or maybe you need better performance so you spend some tedious hours searching for newer components. The World Wide Web makes this task much much simpler. Then you draw up the schematic, which describes the functional blocks with individual components, plus all the connections between them. At this stage, you maybe build parts of it to test, especially if you’ve got a newer design.

Then, if you have a computer, you convert the schematic’s symbols into to-scale representations of the physical board and parts. (Note that a schematic is unlike a house plan in that it does not detail where everything goes, it only shows how they’re hooked together.) Each part you then put in a place that makes sense for it. There’s different criteria for what’s good for what, but mostly you want to make the connections short and simple. The computer (or you, or both) draws out the connections in lines which will be the copper traces seen on the circuit board. (I’m being facetious when talking about the optionality of the computer. It saves literally months of work with tape and paper (yes, in the past they used tape to draw out the circuit board traces.))

You get it made, test it, redo it because it didn’t work or you made a mistake, make another one, test it, and ship because it was due three months ago and hope the errors aren’t too great.

In short, just like a lot of other work.

panama jack


I have a B.S. degree!

engineers are gods !!! (small ‘g’ gods mostly, but some get capitalized)

Indeed. You may arise now. :smiley:

Sometimes marketing departments seem to believe this, but it doesn’t stop them from risking our wrath. Shame I can’t fire off a few thunderbolts around all this computer equipment.

And don’t forget the engineers in the R&D departments who come up with new widgets or improved materials that can then become part of the Book of Pre-Invented Wheels for other engineers to incorporate into their designs.

I agree with the other descriptions of the engineering process, but I’d add that (at least among the good engineers) there is also an understanding of how the components work…it’s not just sticking together black boxes in a new order. Without that understanding, things fall apart. And of course, engineers work in specialized teams so that the different components (electrical, mechanical, etc.) can be understood as a whole.

aside…
as Carl Sagan said, these are the people who should be on our stamps :slight_smile:

Can I attack this question from the point of view of one who has been doing mechanical and structural design for some years?

Sadly, many engineers do NOT know how to build stuff. They know what they want and they know what they want it to do, but they don’t necessarily know the smartest way to get there. Either through arrogance or inexperience, they design things that can’t be built, or they make things absurdly difficult for the machinists/mechanics/tradesmen who must take the design from paper to product.

A good engineer will set aside the books from time to time and wander out to the production floor and watch how things are made. A good engineer will consider the end-user and the maintainers who will keep the widget running for years to come. A good engineer isn’t afraid to learn from the experiences of other engineers, technicians, welders, tool makers, carpenters, and on and on. A GREAT engineer will have done manual labor at one point and will truly appreciate what it takes to make a good design work.

…and some of us just get by on our looks…

All of this has been extremely accurate and to the point, but I would like to add a few bits:

Computers are used more and more to do the grunt work and check out the design. Often the entire design is loaded into the proper software, whether it is a bridge, a VCR or a car, and a simulation is run to make sure that everything is put together properly. While it is not perfect (The computer generally does not know yet that because I put this component here it generated interference with that component there.), they can spot major DOH!s.

This is especially evident in microprocessors. The first ones, with a few hundred transistors, were actually built up out of individual components on a breadboard in order to test the design. Once it was working, All of the mask layers were cut and inspected by hand. This would clearly be impossible with the latest generation of mp’s. Not only would it be too big with millions of discrete transistors, but they use some devices, like dynamic memory, that have no discrete conterparts.

Also, while a lot of the circuits are cookbook, generally you are trying to design something that is different in some marketable way. For example, you may want to make a VCR smaller. The engineer has to then look at the circuit, either the building blocks, or as a whole, and try to figure out how he can modify the cookbook design to make it fit his new constraint and still work reliably.

Engineering is a cumulative effect. We know much because much has been done before, and documented.

We know the science to go where things have not been done.

How would I balance where my engineering knowledge came from? About 30% in school, 70% on the job - the school of hard knocks. But then, consulting engineering is different then what they teach you in most Engineering schools.

Una Persson,
P.E.

I’d like to emphasize the importance of the machinsts and tradesmen who actual build the “thing.” I’ve been on construction sites where it is obvious that the engineer and designer are bound and determined to do it one way and the guy with the hammer is bound and determined to do it another. While there’s usually a compromise that can be reached, if time (and, by extension, money) are of the essence (and they almost always are), the guy with the hammer is going to win.

Okay, here’s my take on the engineering process:

First, engineers learn how to reduce problems to their essential bits. This is an important skill in life, and non-engineers could use a few lessons in how to do this effectively.

For example, let’s say I want to build a radio. As an engineer, I’d start with the basics. First, I have to understand the general principles of radios. We’ll use an AM radio for a sample. An AM radio receives waves of a certain frequency, which have a signal modulated on them by varying amplitude. So, to make an AM radio we need to:

[ul]
[li]pick up the signals from the air[/li][li]tune out the frequencies we don’t want[/li][li]convert the changing amplitudes of the waves into a signal we can use[/li][li]amplify the result, and play it through a speaker[/li][/ul]

Now, we start at the first bit. We need an antenna. If we had to build one from scratch we’d refer back to electromagnetic theory and know that it needs to be a good conductor, and the length should cut to a multiple or fraction of the wavelength. But since AM radios have been around for a long time, we’ll just use some other engineer’s work. I’ll go to my parts catalog, look up antennas, and pick one that matches the specs I’ve set up.

Now we’ve done that, we need to filter out the frequencies we don’t need. We do that by building a band-pass filter. In the old days, we might have had to design a filter from scratch. This involved tuning capacitors and inductors together, and required a fair bit of math. Chebychev polynomials, that sort of stuff. But again, we don’t need to do this, because some bright guy already designed one for us and encapsulated it into a nice IC that we don’t have to understand. All we need to know is that if we use this Integrated circuit and put a 20pf capacitor between pins 2 and 3, it will filter what we want. The formulas for selecting the capacitors etc are on the data sheet for the thing.

Now we need a detector. Again, we don’t have to design one, because some company makes one we need. Pick the part.

Same with the amplifier. In the early days of transistor radios, you used to see little emblems like, “8 transistor amplification!”. Those engineers actually had to build the amplifiers out of discrete transistors, biasing each one separately and configuring them for the specification they needed. Nowadays, you simply spec out the IC you need and buy it. Here’s one that takes a 20uA-40uA input and converts it to a 20-30 mA output, with no more than .2% total harmonic distortion. Great.

Now buy a speaker. Take this bag full of components, and connect inputs to outputs. After you’ve built a schematic diagram, you convert it to a PC board layout, etch your PC board, and connect it all up.

The devil here is in the details, and here is where experience comes in. You prototype your radio, and you get a gawdawful hum in the speaker. Oops. Got some unductive leakage between PC board traces. Add a capacitor here and there to cut out the DC signal that’s leaking through. Put a tiny filter here or there to improve performance. Now you need a metal shield to prevent the IF stage from radiating and messing up your TV. That sort of stuff.

Anyway, the message here is that encapsulation is what allows more and more complex things to be built. It used to take a whole design division to design an AM radio - now a hobbyist can do it on his desk. More complex computers can be built, because they are just a collection of black boxes strung together. If a process has specific inputs and desired outputs, then if I can encapsulate that process I can hide the implementation from others. That way, they don’t have to understand how it works to use it. That’s how you reduce complexity in design, and we do it that way in software as well as hardware (this is the whole point to ‘object oriented programming, btw’. If I have a device that allows me to input two numbers from one to 100, and the output will be the natural log of those two, that’s all I need to know. Inside that black box may be thousands of transistors arranged in a hellishly complex pattern. If I had to understand it to use it, I’d be in big trouble. But I don’t. I just need to know the relationship between inputs and outputs, and I’m happy.

Some of these ‘black boxes’ are just documentation. If I build an airplane, I don’t have to do wind tunnel testing of airfoils, because NASA has already done it for me and documented it. I can go get the information for a NACA-245 airfoil, and it’ll tell me that if I build it at this size, then at this speed it’ll generate X amount of lift. When designing my airplane, I just flip through catalogs of airfoils until I find one with the characteristics I need. The shape of it will be documented as well, and I just need to duplicate it. Now I need to build a wing, and I have to make sure the spar is strong enough. Well, I can order metal of a certain guaranteed strength, and my engineering manuals tell me that an I-Beam made out of this metal will withstand X pounds of force before distorting. Again, I didn’t have to learn that myself.

And so it goes… Sometimes, the original work hasn’t been done already, and R&D comes into the picture. Perhaps my airplane is going to go faster than any published data available, so I have to build test planes and go collect new data. Or my new computer is so fast that previous engineering knowledge breaks down. Time to do research.

The vast majority of products on the market do not break new ground, and represent different collections of off-the-shelf black boxes to solve different needs. But even within those products there is room for creativity and innovation. That’s where intelligence and skill and experience comes in.

An answer to your post
Thats why god invented Technicians.
Takes deep bow.

Oops that was a note to FairyChatMom

Well,what can I say?
When I hit refresh part of my post wasn’t there.

From plnnr

A sincere thank-you for recognizing us lowly steel cutters…its very much appreciated.

And yes, I still have all of my fingers. They’re cut up a bit, but they’re all still there (unlike my father, missing half an index finger…).

Whew! Thanks for the info. Next time you see me open something electronic, smack me!