RMS Voltage measurment using MSP430FE4232

Chinmay Gadre said:

1 reading / sec is not acceptable to me when SD16OSR = 256 fM = 1.048576 MHz

With this setting you should get 4096 conversion/s.

Clockign the SD16 with 1.048576MHz results in 1048576 comparisons and as many generated output bits per second. These are fed into the digital filter and with OSR256, after 256 bits a conversion result is produced -> 4096 rsults per second. Not enough for RMS?

Anyway, the OSR is reduced by writing a different value for SD16OSR into the SD16 config register. It's that simple. The SD16 will then present you a different range of result bits (the meaning of the result value changes as it represents a different window of the filter output data). With an OSR of 32 (the minimum) the usable bits are ~11 (the rest is noise), while with OSR = 1024, >16 bit can be significant (depending on the external hardware). However, the samling frequency reduces to 1024 samples/s

Jens-Michael Gross said:

With an OSR of 32 (the minimum) the usable bits are ~11 (the rest is noise)

How did you calculated 11bits for OSR = 32 ?

 

what will be rms voltage parameter update rate for OSR = 32 ?

Chinmay Gadre said:

How did you calculated 11bits for OSR = 32

My fault, this was calculated for a gain of 32, where the SINAD is >62dB which results in 10.x usable bits. Still better than with a gain of 1 where it is >83dB = 14 bits. but for 1/32 of the signal = 5 more MSBs always zero.
But then, the SINAD over OSR curve was almost identical, so with OSR32 you only have a SINAD of >60dB too.

It a bit complex, as there are many different things to be considered, such as dynamic range of the input signal - I think it is 24 bit, but many of these bits, MSBs or LSBs, meaningless for the different OSRs, the SINAD which is relative to the signal I think, and also absolute restrictions.

'usabel bits' does not mean the absolute number, but the number of bits you can trust from the highest '1' bit down. All other bits below are most likely noise or at least suspicious.

So if you have a 16 bit result and 11 usable bits, the result 0xf0f0 for an input voltage could as well be 0xf0e7 or 0xf0f3. But if it's 0x07e4, then it's 0x07e4 and not 0x07e5 or 0x07e3.

The trick of the SD16 (or any delta-sigma converter) is that it gets the more precise on small voltage changes the smaller the absolute input voltage is (down to the point where you don't have a usable resoluton anymore) It's completely different to 'linear' ADCs liek the ADC10 or ADC12. Those are much faster, though.