| ??? 08/15/06 17:36 Read: times |
#122300 - Indeed! Responding to: ???'s previous message |
Erik said:
That someone obsoletes a chip is bad, but if you can not react to that, your design is way too locked. It was not that we could not react to it, but that we had to deliver and had not enough time for a revision! Since this time I never rely on any exotic chip any longer and may it be the famoust chip of the world. (Or by other words I try to avoid MAXIMs' chips...) Erik said:
The question here is: "was the chip then not within what was stated in the datasheet, or this a result of relying on so called testing?" Refering to charge injection at that time you never got a maximum specification, only typicals. So, all you could do, was comparing switches by their typical data and choosing the one series which showed the best results. But then we found out that from one day to the other the charge injection increased astronomically, which wasn't bad for them, because they didn't specify the upper limit, but bad for us, who heavily suffered form the increase. Finally, in a later databook we found a photo of a new die and some hints in the datasheet, that they now used a "better" methode to bury layers in order to enhance the latch-up immunity. Well, the designers weren't telling the truth, when we tried to contact them. Erik said:
How many salespeople have you caught lying? Exactly! That's the point! But it's not a "proprietary Atmel feature". Erik said:
just two examples of unadmitted chip malfunctions Could tell you some more, from other companies... Kai |
