I am trying to develop an Irms measurement device for the amps used for compressors and fan motors like those used on Air Conditioning package units (PU) or Roof tops units ( RTU). I am using a 0-30 Amp CT for the measurement .

For development purpose I am using a small fan to measure its amps.

The hardware is an 16 bits Delta Sigma ADC converter with sampling rate of 10,000 sps.

Before the ADC I have added an Low Pass filter( LPF) like anti aliasing.

The code implement the classical equation for RMS value which is :






I was wondering if the LPF insertion can degrade the purpose of the measurement which is get the RMS amps.

My understanding is that because inside of a RTU there are at least one compressor, two fan motors and a couple of contactors there are a high probability to get a noisy signal. This is due to switching compressors fans motors, pick up noise from contactor switching inside the unit and from other RTU units.
So that the anti aliasing filter is necessary but it can avoid sampling the real waveform that the different motors are really using.
I will appreciate some commentary about that.
Thank you very much in advance.

 

With a 10ksps sample rate, only noise above 5kHz will be aliased, so the filter just has to suppress noise above that frequency.
It's likely that you need no more than the 10th harmonic or 500-600Hz for an accurate RMS calculation, so using a 1-pole LP rolloff at about 1kHz will give you a good signal bandwidth, while still suppressing noise above 5kHz (-14dB @ 5kHz).

And spike noise from contactor switching, etc. is generally infrequent and of short-duration so if you do a running average of many cycles to get the RMS average, then that will also suppress the effect of such noise on the measurement value.

 

I am trying to develop an Irms measurement device for the amps used for compressors and fan motors like those used on Air Conditioning package units (PU) or Roof tops units ( RTU). I am using a 0-30 Amp CT for the measurement .

For development purpose I am using a small fan to measure its amps.

The hardware is an 16 bits Delta Sigma ADC converter with sampling rate of 10,000 sps.

Before the ADC I have added an Low Pass filter( LPF) like anti aliasing.

The code implement the classical equation for RMS value which is :


View attachment 122834



I was wondering if the LPF insertion can degrade the purpose of the measurement which is get the RMS amps.

My understanding is that because inside of a RTU there are at least one compressor, two fan motors and a couple of contactors there are a high probability to get a noisy signal. This is due to switching compressors fans motors, pick up noise from contactor switching inside the unit and from other RTU units.
So that the anti aliasing filter is necessary but it can avoid sampling the real waveform that the different motors are really using.
I will appreciate some commentary about that.
Thank you very much in advance.

Interesting! Is that precision really required? What are you using as a voltage source that is equally stable? Is a regular clamp-on Amp meter not sufficient?

 

Interesting! Is that precision really required? What are you using as a voltage source that is equally stable? Is a regular clamp-on Amp meter not sufficient?

Hi hp1729

Thank you for your commentary.

Can you please expand it?

.

 

With a 10ksps sample rate, only noise above 5kHz will be aliased, so the filter just has to suppress noise above that frequency.
It's likely that you need no more than the 10th harmonic or 500-600Hz for an accurate RMS calculation, so using a 1-pole LP rolloff at about 1kHz will give you a good signal bandwidth, while still suppressing noise above 5kHz (-14dB @ 5kHz).

And spike noise from contactor switching, etc. is generally infrequent and of short-duration so if you do a running average of many cycles to get the RMS average, then that will also suppress the effect of such noise on the measurement value.

With a 10ksps sample rate, only noise above 5kHz will be aliased, so the filter just has to suppress noise above that frequency.
It's likely that you need no more than the 10th harmonic or 500-600Hz for an accurate RMS calculation, so using a 1-pole LP rolloff at about 1kHz will give you a good signal bandwidth, while still suppressing noise above 5kHz (-14dB @ 5kHz).

And spike noise from contactor switching, etc. is generally infrequent and of short-duration so if you do a running average of many cycles to get the RMS average, then that will also suppress the effect of such noise on the measurement value.

Hi crutschow

Thank you very much for your answer.
Your answer confirm my thought about use a anti alias filter and I just need to adjust the cut-off as you suggest me.
Kind regards,