FlexRMS thoughts & comments

5 posts / 0 new

Last post

Wed, 02/08/2017 - 08:06

johnson

FlexRMS seems to assume a sine wave in the calculation.

I ran the following 60 Hz PWM with 50% duty cycle out of FIO0:

MB.W(44900, 0, 0)

MB.W(44901, 0, 1)

MB.W(44904, 1, 1333333.3333333333)

MB.W(44900, 0, 1)

MB.W(44000, 1, 0)

MB.W(44100, 1, 0)

MB.W(44200, 1, 0)

MB.W(44300, 1, 666666.6666666666)

MB.W(44000, 1, 1)

So that's a nice balanced ~3.3 volt square wave. The RMS voltage of a square wave is equal to the peak to peak voltage:

https://en.wikipedia.org/wiki/Root_mean_square

However, the values returned are Vrms = 2.302 and Vpk-pk = 3.250. Note that they are exactly different by a factor of sqrt(2), the amount one would expect to see for a sine wave RMS to peak voltage.

Other thoughts:

The frequency measured with a Fluke DMM is 59.98 Hz.

The Period never seems to get updated in Kipling? I haven't tried through the AIN_EF_READ_D registers yet in software, just in the Kipling GUI.

I've tried a few different combos of Num Scans, Hysteresis and Scan Rate; the default seems fine for 60 Hz signals.

I'm using 3 channels at the moment with FlexRMS enabled, what is the limit and how does this processor load translate into command response time?

Overall I think that FlexRMS is a fantastic extended feature and I would love to see a bit more development and documentation. I've yet to play with the RMS_Auto.

Thanks,

- Jason

Thu, 02/09/2017 - 16:39

LabJack Support

We managed to repeat the

We managed to repeat the issue you are seeing. The firmware guys are looking into it.

Wed, 02/15/2017 - 08:12

johnson

Noticed that if I use a

Noticed that if I use a differential input it seems to give a correct reading (also noticed that my first comment on square wave RMS was a bit incorrect as written above).

I'm curious what the allowable crest factors would be in a 60 Hz application and how accuracy relates to sampling rate, number of samples and hysteresis. I use a lot of SSR based phase angle power controllers driving transformers which can end up with some pretty ugly waveforms...

Wed, 02/15/2017 - 10:56

LabJack Support

I finally figured out this

I finally figured out this issue. It is not a problem of calculation, but rather classification.

The PWM generated by the DIO_EF is a Pulse train, not a square wave. To be a square wave it must have positive and negative amplitudes of equal magnitude.

The T7 calculates RMS using a discrete integration, so the results should be good regardless of the details of the waveform.

The wiki page that you linked seems to agree.

Thu, 02/16/2017 - 11:13

johnson

Thanks, that makes sense now.

Thanks, that makes sense now. Especially with the differential input reading ending up at what I thought I should get initially