Hardware Design in Advanced Company

Thread Starter

eepty

Joined Sep 11, 2012
12
I am a design electronics engineer. When my team design a new circuit, we always base on our knowledge and experience to draft the circuit and layout the PCB. And then test it by "trial and error". For example when designing the EMI filter, we will just estimate the capacitor and inductor value by our experience, and then test it after the prototype is made. If it goes wrong, then we change the value and test again.

I just feel that this method is kind of rough. It is difficult to answer when somebody (for example the customer) asked why we use this value. Will the big, sophisticated, advance organizations, such as the companies making aeroplane or military parts, NASA, use another method? For example will they use simulation software for the hardware design?
 

shteii01

Joined Feb 19, 2010
4,647
You just described how you chose the value and then in next paragraph you tell us that you can not explain how that value was chosen. Seriously?
 

cmartinez

Joined Jan 17, 2007
6,621
Doesn't everyone use the SWAG method over the much inferior WAG method?
In the words of Richard Feynman... "first we guess" ... it's a valid approach to a completely unknown subject... But I would've thought that by now there would be enough knowledge accumulated so as to capture it into software. Instead of still relying on old fashioned intuition gained from experience.
 

GopherT

Joined Nov 23, 2012
8,012
I am a design electronics engineer. When my team design a new circuit, we always base on our knowledge and experience to draft the circuit and layout the PCB. And then test it by "trial and error". For example when designing the EMI filter, we will just estimate the capacitor and inductor value by our experience, and then test it after the prototype is made. If it goes wrong, then we change the value and test again.

I just feel that this method is kind of rough. It is difficult to answer when somebody (for example the customer) asked why we use this value. Will the big, sophisticated, advance organizations, such as the companies making aeroplane or military parts, NASA, use another method? For example will they use simulation software for the hardware design?
Maybe you could do a calculation first, then fine tune it for your needs with a prototype. Do you know how to calculate the filter's cut-off frequency?
 

crutschow

Joined Mar 14, 2008
23,742
I worked in Aerospace and this is generally the process:
  1. Analyze and develop the requirements until they are thoroughly understood and make engineering sense.
  2. Mathematically develop the design to meet the requirements.
  3. Simulate the design on a computer to see if it meets the requirements.
  4. Built and test the design, and determine why it doesn't meet all the requirements.
  5. Redo steps 2, 3, and 4 as needed.
 

AnalogKid

Joined Aug 1, 2013
8,221
Will the big, sophisticated, advance organizations, such as the companies making aeroplane or military parts, NASA, use another method?
Yes, rigorous engineering calculations and datasheet analysis.

Two examples. First, when you are building 10,000 cell phones *every day*, you can bet there is a firm mathematical foundation for every component down to the smallest resistor. A friend worked Lucent (ATT), and spent *months* qualifying a new 0.1 uF decoupling capacitor. At that time they were using 100,000 per week.

Second, when you are building *1* Voyager probe or Mars Rover, there is a solidly documented reason for every technique, design, and component.

In both cases, a circuit's performance over the range of initial component tolerances, temperature drifts, aging, etc. can be calculated using design equations and spreadsheets. The pc board layout can be ported to electromagnetic field analysis software to evaluate noise, crosstalk, impedance issues, etc. All before anything is actually built.

Powerline filters are like RF circuits in that the mechanical arrangement among the components sometimes is almost as important as the component values in terms of electrical performance. Still, if the first prototype is worse that 5% off design, work harder at the design.

ak
 
Top