DDC264: Data retrieval

That's interesting question... I can think of two things... 1/ There is a max integration time (1s if I recall right). Not sure what breaks afterwards, so, I wouldn't exceed that, although probably nothing really breaks. It is just that beyond that, the result of that integration probably wouldn't be very accurate, but as it is for something that your customer wants to trash, may be ok. Still, if they can avoid it, better... 2/ As big of a concern is that the device will still integrate during B and if the integrator gets saturated, the input (which traditionally is virtual ground) will start drifting (changing voltage level). That results on a charge accumulation at the input that gets dumped on the next integration cycle (which is the one that customer really cares about).

So, whatever scheme they use they probably need to take these two into account. Nevertheless not sure I can find a solution if there is no indication on when the next sample is going to come. Basically you would like to be ready on the dummy state but not so early that 1/ or 2/ hit you just waiting for the good sample. Without having an idea on when the next sample is going to come, I can only see some strategy but not 100% proof. I.e., say the worst case of 1/ and 2/ is 1ms maximum time. Then you would probably be integrating at 1ms rate, jumping from A to B and back, throwing all those dummy samples, waiting for the good one... But you always run the risk that if the sample that you care for comes just few microseconds after you toggled the dummy sample, then you actually can't toggle there to capture the good sample (as the dummy didn't finish the minimum integration time). Can't see how to design a 100% proof method without having more information about the sample to sample interval, but probably I am missing something...

Regards,
Edu