MSP430F67641A: SD24_B: how much accuracy is proven when the readings are accumulated? (not single reading)

Hello,

Hideaki Nambu said:

Are there evidences which prove higher ENoB or effective bits by the MSP430 SD24_B and firmware post calculation?
We would be glad if there are proven cases which realized, like 18bit, 20, or more accuracy.

While I'm not sure 18- or 20-bit accuracy can be achieved, you can improve the performance by performing software oversampling (or averaging) across a number of samples. This acts like a low pass filter and can improve performance post sampling. More details on how to do this can be found in the following thread. The trade-off here is that you have a slower effective sampling frequency but if your application can handle that, it's fine. The effective sampling rate with software oversampling would be the hardware sampling rate (Fmodulation/OSR) divided by the number of samples averaged in software before reporting the value.

[FAQ] MSP430I2020: Measuring SD ADC noise

Hideaki Nambu said:

The SD24_B has a bit SD24SNGL.
I expected we can clear the bit for "continuous", accumulate many readings, and obtain a better ENoB,
like 20bits, although I could not find any articles which encourages such the accumulation.

Continuous sampling may improve the results very slightly since the decimation filter will be continuously full and you won't need to wait for the 4th sample for the results to be accurate.

Regards,

James