This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

MSP430F6723A: SD24 50Hz rejection filter

Part Number: MSP430F6723A
Other Parts Discussed in Thread: MSP-DSPLIB

Hello.

There is a problem with an SD24 rejection filter in MSP430F6723A.

While measuring voltage 1..5mV on sample frequency 1KHz, we can see a presence of 50Hz signal in SD24 results.

As we can see, 50 Hz is suppressed roughly by 30 dB. And we need to suppress this noise at least by 90 dB.

What could you recommend to fix this issue?

  • Sigma-Delta ADCs get their 50Hz rejection by placing a null of their SINC filters at 50Hz. You would have to set the sample rate to 50hz to do that. Or, if your sample rate is exactly 1KHz, average 20 samples.

  • Hi Slava,

    Thanks for posting your question on E2E. I see that David has already chimed in, and he's absolutely right.

    The Sigma Delta ADCs in the MSP430F6723A feature Sinc3 decimation filters, and the first filter notch (or null) occurs at the sampling frequency, Fs, as illustrated by Figure 29-7 from the User's Guide.

    Some devices have sampling frequencies of 50 Hz which would reject 50 Hz noise. However, the MSP430F6723A devices are commonly used in metering applications such as electricity meters and power quality monitors where the nominal high voltage AC frequency ranges from 50 Hz to 60 Hz, depending on the region.

    Now, the SD ADC sampling frequency (Fs) depends on two things: oversampling rate (OSR), modulation clock frequency (Fm)

    Fs = Fm / OSR

    David recommended using a sampling frequency of 50 Hz to take advantage of the first filter notch at 50 Hz to minimize the noise. Assuming that sampling frequency meets your requirements, you could use the minimum Fm = 30 kHz and an OSR = 600 to achieve Fs = 50 Hz.

    If you need a higher sampling frequency, the other options would be adding an external hardware filter or implement a software filter to remove the 50 Hz noise. You could also use the lowest possible sampling frequency to take advantage of the Sinc3 filter roll-off. Lower sampling frequencies mean there's more attenuation at lower frequencies and vice versa.

    Edit: You could also narrow down if the noise is coming through VCC (AC PSRR performance) or through the ADC inputs. Perhaps better filtering on the VCC supply would eliminate this noise.

    I hope this helps.

    Regards,

    James

  • Hi James and David.

    Thank you for your exhaustive answers.

    1. Do I understand correctly,that there is no way to set the null of SINC filter on 50 Hz, while leaving sampling frequency on the level greater than 100 Hz ?

    2. Can we provide some examples of software filter for MSP430?

  • Hi Slava,

    Slava Prokopii said:
    1. Do I understand correctly,that there is no way to set the null of SINC filter on 50 Hz, while leaving sampling frequency on the level greater than 100 Hz ?

    Correct. A sampling frequency of 100 Hz will attenuate 50 Hz noise more than the sampling frequency of 1 kHz because the first filter notch is closer to 50 Hz. This assumes that the input frequency is very low and won't be affected too much by the attenuation. The sampling frequency sets the first filter notch. The second, third and other notches are just integer multiples of the sampling frequency.

    Slava Prokopii said:
    2. Can we provide some examples of software filter for MSP430?

    Take a look at our Digital Signal Processing (DSP) Library for MSP430 Microcontrollers (MSP-DSPLIB).

    Regards,

    James

  • Hi James,

    We have tried to use a software 64-pole filter from MSP-DSPLIB.

    The achieved suppression of 50 Hz noise is only 30dB.

    What can you recommend to increase the suppression up to 90dB?

  • Hi Slava,

    Have they tried configuring the filters using the DSPLib GUI? I spent a few minutes and was able to create a high-pass filter that seems to filter out 50 Hz. If the bandstop filter type doesn't meet their requirements, they can always use one low-pass filter and one high-pass filter to achieve the same thing (or perhaps better suppression than the bandstop filters).

    Regards,

    James

**Attention** This is a public forum