| ??? 06/16/04 15:05 Read: times Msg Score: +1 +1 Informative |
#72597 - RE: ADC input Error logging Responding to: ???'s previous message |
erik malund wrote:
------------------------------- By the way, the above remind me, I have seen ever so often designs where the ADC was used full scale i. e. max normal input = max value. That is a BIG mistake. In such a design there is no way to detect an error causing the analog input going wild because of an open or short. That reminds me of a not-so-funny story. In a first version of a product using a 16-bit ADC to read temperature via RTDs wired through a 1K pull-up resistor, the system would periodically select a channel with a hard-wired 1K pull-down resistor. If the channel read 2.5 Volts (8000H), the system determined that all was good. Well, it turned out that the ADC could lose its calibration data and then when it would take a reading, it would take the reference (which was 8000H) multiply it by the gain (which defaulted to 1) and add the offset (which was then 0) and come up with 8000H. The moral of the story is to be careful when you choose you error checking. It helps to try to predict (or simulate) the method of failure when you design your error checking. -- Tom Bachmann |
