| ??? 11/18/09 20:11 Modified: 11/18/09 20:15 Read: times |
#170944 - Thorough testing? Often an illussion! Responding to: ???'s previous message |
Richard said:
Too many people think that having it do what they believe it should is sufficient indication that it, the device in question, "works." Until "works" is clearly defined in writing, nothing about it is known, aside, perhaps from its cost. Richard, reality is far away from this "ideal-yes-no-world". A product can have been thoroughly tested and nonethelss suddenly fail from one batch to the next. It can have been thoroughly tested and nonetheless suddenly fail, when changing the cabling or surrounding equipement. You cannot test something, that isn't specified. As long as uncertainties, which you don't exactly know, have influence on your product, testing is never 100% reliable. I can write an endless list of examples, where the manufacturer has changed the design of a chip, without informing the customer via datasheet: I remember an analog switch from Analog Devices, where the charge injection suddenly increased by a factor of about 50 and the product failed from one batch to the next. Years later, they mentioned in the datasheet to have "improved" the design. I remember the ultra low input bias current OP297, which from one batch to the next changed the performance, when overdriving the inputs. A fomerly totally quiet OPamp suddenly showed heavy oscillation at the output. Even today, not even a word about this strange performance in the datasheet. And and and... Or, how many times we see a device stopping working properly, only because we have changed the connected equipement or the cabling. Think only of those many contradictory and incompatible design practises (or should I say philosophies?) of how to handle the cable shield, or of how to handle signal ground. A product, that is assumed to have signal ground disconnected from protective earth, will have earth loop issues if connected to it and can suddenly fail. Again, there's an endless list of incompatibilities, which can make a fully tested device fail from one moment to the next. Testing is alright. But to think, that a thorough testing will automatically result in a reliable product is a painful illusion. You need to repeat this testing over and over again. And you must be damned lucky to have a non failing product! From my experience, it's about 50% testing and 50% hoping, when selling a new product. Make no mistake, there's always a huge risk when producing or even inventing something new. And you must can stand this... Kai Klaas |
