I stated:
"Testing only proves that the circuit with those particular components work under those conditions that day. The only way to make a reliable thing is by design."

Kai Wrote:
Don't insult me! I never stated that testing should be done INSTEAD of developing.
Reliability test is recommended for oscillator circuits, which were developed by carefully obeying instructions of according datasheet and application note, as it is the obligatory procedure for professionals. It's suggested to people, which carefully designed their oscillator circuit but still wonder, whether it will actually work under extreme conditions. And I named those critical extremes, namely limits of temperature range and long power-up time.
Kai, no intent of insulting you. Maybe I am psychologically bruised by the many times I have heard (and read here) "it is not my design, the prototype worked". To many participants in this forum "design" means "make a schematic" WRONG. Design means calculate all signal paths with minimum Tpd on one and maximum Tpd on the other, analyze all connections with maximum load from inputs and stray capacitances together with minimum drive from the output, all done with the temperature taken into account.

Ever lab and hobby room should have a large poster:
Succesful testing does not prove the abscence of errors, it only proves the abscence of known errors.


Raghu wrote
There must be more behind the above statement which is not obvious

Now the million dollar question :
Allow the product to ship as it has passed testing
(OR)
Wait for design validation of the change that was done by the Chip manufacturer ?

Going by your statement, the second option is the way to go. But this as you know will not happen if there are no big volumes / OEM application etc..

Obviously you have to ship when the product has passed testing. What I am trying to say is "design a product so that it will pass the test" do not "'design' a product and hope it will pass the test."

Erik