The following doe NOT happen at gain of 1:
When using a high gain to study a sine wave input, the gain can amplify the signal beyond the +/-Vref.
If the gain is set ONLY ONCE before the scan starts, the readings wrap around instead of saturating at the rails In fact, if the highest gain is used, it wraps around many time
If I set the same gain every time ISR is invoked to fetch the data, the reading is perfect, saturating at the rails. Of course, this action totally screw up the sample rate. See http://ez.analog.com/message/89811#89811
Message was edited by: MMA - this is not a BUG must be a error in configuration and/or SW