| ??? 11/19/09 06:16 Read: times |
#170957 - Agreed, with caveats Responding to: ???'s previous message |
Erik Malund said:
From my experience, it's about 50% testing and 50% hoping, when selling a new product.
'hope' is not a design parameter. But to think, that a thorough testing will automatically result in a reliable product is a painful illusion BINGO!!! Many do testing to "see if it works" that is a TOTAL mistake, testing shold only be done to "catch mistakes in implementing a carefully crafted design" If the design is not "test proof" it serves no purpose to test. That's the whole point behind "design for test." You can't test for potential failures that you haven't considered. The goal of a rigorous test regimen is to verify the behavior of your functional device under the range of conditions over which it is characterized, and, to verify that it does what you predict when those conditions are exceeded, within limits, of course. It's up to you to characterize it, too. Of course, you should test, but if you have not designed the thingy to work unconditionally it serves no purpose to test.
Erik Succesful testing does not prove the absence of bugs it only proves the abence of known bugs. There's more to testing than looking for bugs in the firmware, not that searching for and finding/remediating bugs isn't a vital objective. How does the hardware behave when the power supply drops below the rated value but doesn't reach the brownout trip-point? How does it behave when the supply goes a bit too high? Does that foul up the data acquisition? How does the system behave when the suppy has 200 mV of random noise on it? What if that random noise wachses and wanes at regular intervals? How does it respond to shot noise? What about temperature changes? ... Do the relays behave as you wish when the PCB is subjected to a 3G 1.5 mm displacement (shock) at 200 millisecond intervals in bursts of 2.5 second duration every 90 seconds? Does the PCB retain its integrity? ... you get the picture ... RE |
