Referring to ADE7953 datasheet pg. 28 section titled **"Reactive power calculation" **I wanted to know whether the algorithm mentioned works for power electronic loads which draw non-sinusoidal currents.

Referring to ADE7953 datasheet pg. 28 section titled **"Reactive power calculation" **I wanted to know whether the algorithm mentioned works for power electronic loads which draw non-sinusoidal currents.

Dear Dlath,

I am attaching the voltage and current waveforms for

a 32W CFL sampled by a 10 bit ADC. As you can

observe the current is non-sinusoidal. Using your

algorithm for reactive power calculation I find the

reactive power taken by the 32W CFL is 'capacitive'

whereas in the literature it is clearly mentioned that

such kinds of loads are 'inductive' in nature. Please clarify the discrepancy.

Please let me know how to send you the ADC data of the waveform. The forum

does not allow text files to be attached.

Hi,

If you go into "Use advanced editor" you should be able to attach a spreadsheet of your data.

I would recommend that you check the polarity of your current sensor to make sure it is connected correctly, from your waveforms it does look like they are in phase. An inductive load would mean that it contains both positive active and positive reactive energy, are you seeing that the reactive is negative?

Regards,

Dlath

Dear Dlath,

The polarity of the current sensor is correct. I checked out your algorithm for loads that draw sinusoidal currents and they work perfectly well.

However, for non-sinusoidal loads, there seems to be a problem. I am attaching a text file containing the actual samples of both current and

voltage waveforms for a 32W Phillips CFL. And yes, for this load your algorithm for the reactive power computes a negative value as is

seen in the instantaneous reactive power graph attached with this message.

Hi,

I looked at the waveform again and looking at your reactive power waveform it actually does look to make sense. CFL bulbs typically generate reactive power and this means it is a capacitive load. This is reinforced by the fact that the current waveform you show is leading the voltage waveform, this points to it being a capacitive load and therefore positive active power and negative reactive power.

Looking at the waveform is the most basic way to understand whether it is capacitive or inductive and as long as the waveform is ok then the power being calculated is also accurate. The one factor that could affect the waveform is if a CT is used for measuring the current and that CT has a large phase error but if you are using a shunt then it should not be a problem. Looking at some papers, the quality of the CFL also looks to matter in terms of the power factor of the load.

Even waveforms which seem non-sinusoidal will have a fundamental frequency at the line frequency but then a lot of harmonics distorting the signal. As long as the signal is contained within the 1.23kHz bandwidth of the ADE7953 then the reactive algorithm mentioned still holds true since the calculation is performed over the full bandwidth of the part.