| ??? 10/03/07 20:05 Read: times |
#145325 - it's just as I said ... YOU guess and ship ... Responding to: ???'s previous message |
Erik Malund said:
Richard Erlacher said:
so YOU really haven't ever looked at any of them? sure I did, but the statisticl probability of a problem that rondomly show up, say, in any 5 of 10,000 units, showing up in the one(s) you look at is infitessimally slim. Thus ihave never 'observed' thge problem. It's not infinitessimal if it's 5 out of 10000. Thus the approach is to, based on experience, modify the design, distribute the fix and wait and see. All this shows is that I was right ... You guess and ship. It's not that uncommon. However, if a system has to be power-cycled multiple times, the smart fix is to add a reset switch. I would like to know what approach you would take to handle a problem you can not reproduce If you can not suggest an accptable approach other than the one I describe above SHUT UP! It's simple. You get the 5 units out of the 10000 fielded units, replace 'em with known functional units, and set up a test for those 5 units, cycling power (you said the problem was that they had to be reset several times to start properly, right?) for, perhaps 10 million cycles, and record the system response with your logic analyzer, examining the signals and comparing with what they should be. If after 10E6 cycles, which can be run automatically, you haven't found the "problem" then it's likely you don't have one. Erik If you wanted to do your job, Erik, you'd find the problem and fix it. I'd venture that the "real" problem is too much capacitance on Vcc and too little heft in the off-board power source. Now, I haven't observed the problem either. I bet you don't know what the risetime of Vcc at the MCU is, either. RE |
