| ??? 01/22/04 15:01 Read: times |
#63129 - RE: reliability vs cost Responding to: ???'s previous message |
Erik
I somehow get the impression that a lot of projects get made to drive cost to the very bottom with no concern for reliability. Jez I think you have to work hard to produce even a home made project which is simply going to stop working once it does work You should add to the above ... as long as it stay in the same environment. I can design something that works beautifully on my bench and will carp out immediately when mounted in a vehicle Michael If your power supplies were awarded their MTBF figure by the manufacturer, it might just be smoke and mirrors. Absolutely; however I get a much better result: abt one in 2000 per year Russell Making devices more 'reliable' generally involves extra cost. even then you sometimes have to debate the use of protecting against a probable but unlikely circumstance. Let me restate "reliable": Reliable in the environment it is to be used in. I do not advocate optocouplers in the interface in a toy, but consider it reason for dismissal to drive a uC pin directly from a sensor in an automotive application. Steve or in 2.6 million units one will fail in an hour ? Absolutely, I thought everybody knew that. Raghu And the rules of the game vary with the domain - what goes in consumer field may be a non-starter in the industrial field and if you enter aerospace or medical diagnostics - most of the components that are normally used may not even qualify. See reply to Russell above Patrick It's not as black and white as you state. I think that one has to make the equipment as reliable as possible Absolutotally agree within the limits laid out by the customer. If the customer has expectations that are set too high, for a given design/cost then it's up to the designer/engineer to warn the customer about his unrealistic expectations. And, if safety is involved, walk away if reliability can not be achieved at the customers cost. Even if something is only used within it's design specifications, even then things can break. And, according to Murphy, it will. Of course, you can not design anything that will NEVER break. What you can do is to design so that it is unlikely that it will. It scared the socks off me when someone in a past post suggested that for an automotive application it was too expensive to protect the inputs against transients. Erik |



