Harmonics filter capacitive or inductive

Hi,

something is not matching:

"sine wave measured with 3rd and 5th harmonic", this is usual and if disturbing to be filtered.

You did not state if you see these harmonics in current or voltage or both?

The source impedance for any harmonic may be capacitive, resistive or inductive, so to look at the zero crossing of the fundamental which phases have the harmonics will be a good idea. But where is the fundamental if you have a DC winder-motor?

Usually there is an input AC voltage, this rectified (3 phase to minimise necessary filtering) giving an intermediate DC.

This intermediate DC is switched by a pulse-width-modulator giving a DC component of any desired voltage and a lot of harmonics.

Then there is a filter to remove most of the harmonics as sharp voltage steps will shorten the lifetime of the insulation and any current harmonics will generate ripple in the torque of the DC-motor and unwanted power-losses and heat in the motor.

So if you see residuals of the switching frequency of the PWM and its harmonics (that vary widely with on/off relation of the PWM) there is most often the need for a filter.

If you think about a L-C combination as filter, L and C in parallel inserted into the hot side of your "DC" voltage, then depending on the resonance of this filter and the frequency of the PWM this will be inductive if PWM-frequency is below filter-resonance and capacitive if above and pure resistive at resonance.

You are right that the peaks of the filter are responsible for suppressing unwanted harmonics - but what happens if the peak is not at the wanted frequency?

Then the amplification (suppression is an amplification with below 1 as gain) and phase of the filter will work according to the values of its gain,phase versus frequency plot.

But in power application you cannot stay with this simplified thinking that is assuming decoupled source, filter, output stages.

The source impedance of your power supply will be part of the filter. So if you introduce a pure capacitance lengthwise in your DC power line you will have an L-C blocking filter made by the internal L with your added C. If you tune the capacitance so there is resonance at the capacitive harmonic you will attenuate this to a level defined by the filters Q-factor. But you will attenuate the capacitive component only according to the ratio of the capacitive divider (one C internal the other one introduced).

So now introduce an additional L to play the same game with the capacitive component: this will change the internal L as it is in effect parallel to the internal L. (Same with the C). So you need to resize the values of L and C to have the resonances at the frequencies of 3rd and 5th harmonic and the added third resonance somewhere where it does not disturb.

This tuning in total is compensating automatically any L or C component in the source impedance.

You asked:Is it possible to have a phase change between harmonics and the fundamental frequency?

Make a Fourier with arbitrary harmonics and change their phases, this will change the voltage/time plot - so where is the problem with the phases of the harmonics?

And also, do filters get design so specifically to filter out a capacitive and inductive component or is it only the magnitude that counts.

Input values times magnification (complex) yield output values. So you are right with your doubts. But I assume that you have no decoupling at the filter-input so you have to include the internal L,C-values into the filter design and this is naming your mentor as designing to suppress inductive or capacitive behaviour.

The real situation is often still more complex as the switching frequency and its harmonics vary. Then you may need a switched capacitor filter as a comb filter that is synchronised in its notches to the operating frequency of the PWM.

RHABE

Some info is here & link the articles from Further reading

Yes! It is possible in "De-tuned" Harmonic Filters. They are designed with a certain Resonance Frequency. For all harmonic frequencies less than this resonance frequency, the filter will act inductive and block the harmonics and above this it will act capacitive and permit the harminics to pass thro it.

For example, a 7% de-tuned filter has a resonance frequency of 189 Hz and thus would block all harmonics above 3rd (i.e.) 5th, 7th, 9th & so on.

A 14% de-tuned filter has a resonance frequency of 134 Hz. and thus would block 3rd harmonics upwards.