| February 2006 |
| By Glen Chenier |
There are many subsets of Electrical Engineering and many specialties within those subsets. The term “Hardware Design” encompasses various different areas - digital and analog on silicon, FPGA logic, imaging, RF and microwave, control systems, audio, telecommunications, optics, chemistry, and power to name but a few – even these can be broken down into many more subsets and areas of expertise. Most hardware designers will specialize in one or two of these areas, and if necessary be willing to learn others for a new project or career change. Those who wish (or are fortunate enough) to remain in the same discipline for years, continue to hone their skills in that discipline. Usually they stay within their choice of specialty because they enjoy it immensely and have become very good at what they do.
Me, I went from telecommunications to power through a forced career change and the help of Dice. I had to quickly learn new concepts, but electrons work the same everywhere so it was not too hard.
My current project is a high reliability power system for high end business servers that must perform with a bare minimum of 'unscheduled' down time. Parallel and redundant DC to DC power converters supply high currents at various voltages to the power-hungry high speed digital electronics that do the server work. If any one of these power converters fail, the power control system must immediately determine which is at fault and remove it from service within a millionth of a second so that the power rails to the server are not glitched or otherwise corrupted. The design goal is that no matter in what manner a converter fails, the server should not even hiccup – instead, it should keep on running normally and flag an alarm so that a human can later perform a graceful repair without traffic disruption. My company, my team, and myself take a lot of pride in designing the best possible circuitry to do this.
There are several different phases to a hardware development project. First, the functions of the sub-product and how it fits into the overall system must be clearly defined and understood. The designer then creates, examines, and determines which of several different approaches for many more sub-circuits are most likely to work together with a minimum of ‘gotchas’. (There will always be ‘gotchas’ on the first spin of the design – the trick is to minimize them ahead of time). Most of this work is done on paper or white board, then transferred to a word processor, and yes, sometimes even on napkins at lunch with your colleagues when a new circuit concept hits.
Then the hardware designer uses CAD tools to create a multi-page drawing – the Schematic Diagram – which shows the thousands of interconnections between thousands of electronic parts. During this time the designer must also evaluate the pros and cons of various components and chose which will perform each particular function best under all environmental conditions. Pick lists of component libraries allow selection of the desired parts, and ‘wires’ are drawn between them with the mouse to create the circuits. Multiple mathematical calculations are required to determine optimum component values. Typical components are resistors, capacitors, inductors, transformers, diodes, transistors, and a plethora of integrated circuits.
In many cases computer simulation is performed (exercising the circuit design using software tools with the schematic) to determine if there are any glaring errors. This works best on digital designs which are fairly cut and dried with standard logic functions and critical but predictable timing parameters. Analog circuits can be simulated too, but these simulations are nowhere near as predictable as to how the final real product will perform. Too many unknowns and circuit interactions in the analog world.
When the schematic diagram is mostly complete, a design team review of the entire concept takes place. This is where errors large and small get flushed out and corrected, and colleagues will suggest improvements. Then a specialist in printed circuit board layout uses more CAD tools to turn that schematic into artwork patterns that will eventually become the interconnecting copper etch of a multi-layered printed circuit board (pcb). During this time the electrical designer and the CAD pcb designer must work very closely together to control mechanical properties that affect impedance, heating, current carrying capacity, and unwanted electrical noise. Often the electrical designer is still making small changes to the schematic design and the artwork designer must continually adjust the layout to match.
When the artwork is done, another team design review is held and many days are spent by several individuals examining the artwork to catch any possible errors. Then the artwork design files are sent to a printed circuit manufacturing house, the bill of materials is given to a production department for parts procurement and limited-run manufacture. Eventually several prototypes of the circuit board are built and returned to the designer.
While waiting for the prototypes to be returned from manufacturing, the hardware designer keeps busy by writing test plans and figuring out all the ways of stressing the design to the max so as to flush out any and all lurking bugs. Nobody wants to overlook a hidden bug that waits until delivery to a customer before manifesting itself. The wise designer will already have designed in test points and methods to make debug easier and time efficient.
Prototype debug is where the fun really begins. The designer must now make the product work – and work flawlessly. The ease of this activity is highly dependent on the effort and expertise that was put into the design in the initial development stages. The success depends on the designer’s experience with making electrons do what the designer wants them to do, not what the electrons themselves want to do.
The first poke at the design takes place in a lab, usually with the pcb mounted in a test fixture that allows injecting multiple stimulus inputs from various pieces of test equipment, and open access to all areas of the board for connection of oscilloscope probes. After the initial ‘smoke test’ (self-explanatory), the designer measures every circuit function with voltmeters, ammeters, oscilloscopes, and depending on the type of product may need to use many other types of test equipment.
Throughout the entire debug cycle the designer keeps a detailed test log. I am fortunate in this particular job – my oscilloscope includes a word processor that lets me combine waveform recordings with text commentary and is on the company network for file transfer and edit on my office computer. A real luxury!
This test log should describe all problems and peculiarities, measurement recordings, state of any hardware modifications, potential fixes, fixes that worked, fixes that did not work, fixes that worked but caused new problems somewhere else, new problems noted while debugging known problems, and occasionally something that actually did work as intended. Some problems may not be overly serious – is it worth fixing? Usually this question is put to the team, and a group decision is made. In most cases, if there is a fix to be tried, then do it. There is no such thing as a ‘little problem’ – however small it may appear now, later it may become a monster.
Discovery of unexpected and surprising bugs means modifying the test plan on the fly. New tests must be conceived and performed to get a circuit function working properly. Until this is done, testing to the the original plan cannot proceed.
Some fixes can be as simple as changing a component value slightly – unsolder the old one and solder in a new one. Other fixes to cure totally unforeseen problems can require drastic measures – hand build an entire new circuit that you think will cure a problem on a ‘breadboard’ and wire it into your pcb. Today most electronic parts are incredibly tiny – hand building a breadboard to test a fix concept means using tweezers to hold parts and soldering under a microscope. But it can be done.
Eventually, with a great deal of perseverance and a bit of luck, all the bugs will be rectified and it will be time to ‘respin’ the pcb – draw all your new circuit changes and fixes into the schematic and repeat the entire process. This time it is usually successful. That is the big reward and satisfaction of this type of work – creating a perfectly functional product in only two attempts. Often it does take three or more stabs at it to get all the bugs out.
Industry culture? Hardware designers are some of the best colleagues to work with. Common goals, common problems, common solutions, and always willing to lend expertise where needed. Teamwork runs rampant in hardware groups. Fun times too - the company I now work for had a talent show last Christmas; our group dressed up as “The Power Pirates” and sang sea shanties in front of the whole company.
For those considering this type of work, a bit of advice. You must enjoy it – if you do not enjoy it, you may not last very long, and you will be very unhappy in your chosen career. Often having a hobby in the same general field (ham radio, for instance) can be a good indication as to whether or not you might enjoy this. Sometimes technical problems crop up that appear to have no solution; these can be very frustrating, sleep disturbing and does not sit well with the boss. You must be willing to persevere and keep trying in spite of everything looking bleak. You will mentally bring problems home with you, then you may undergo the glorious experience of suddenly hitting on the solution while in the shower. And then the even better experience of breadboarding the shower solution the next day and finding that yes, it really does solve the problem nicely.
You probably won’t get rich either. But you will have the satisfaction of career enjoyment, creative accomplishment, and knowing that your designs have made some impact on society.
If you would like to submit an article providing an insider’s view of your tech or engineering job, please send your article to us at ITtrenches@dice.com.
Comments on this article? Share your feedback on our discussion forum, Dice Discussions.
*Please note, you must be a registered job seeker in order to submit your question to Dice Discussions.
|
|
|
