How filter works?

Thread Starter


Joined Dec 17, 2009
Hallo guys..

I hv very basic questions which need your help to clarify them..

Let say that I hv a LC filter, tuned at 8kHz with THDv and THDi limit to 5%, configured for 3-phase system (400Vrms,562VAC,50Hz,10A)

a) if there is a spike of i.e. 60V, how does the filter actually deal with it (absorbed, suppress, cutoff,??)..

b) if absorbed, is it means only 562V is allowed on all the time? how about spike less or more than 60V?

c) what is the gain of filter in bode plot tells us since the gain is negative in the working frequency (8kHz)? is it means it will suppress extra voltage on the system (noise/spike) ? how about if it only able to suppress only 5V at certain frequency but the spike is actually 10V?

d) if the filter is design to specific THD limits, is it means that it will suppress harmonics starting from 2nd/3rd harmonic number to maybe the 17th or only works on certain harmonics number only

e) is it ok to limit the THDv and THDi < 5% ? how about make them much much lower i.e. 0.05% ?

I hv googling a lot but still cannot reach to any concrete answer..

Thanks for reading those questions and many thanks for answering them!!


Joined Apr 18, 2009
Lets explain it this way: The way a filter attenuates a signal depends on the frequency the signal that is propagating trough the filter.
If you have a lowpass filter then a 60V spike will get attenuated by some amount which depends on the frequency of the spike and the design of the filter. If the frequency of the spike is at the cutoff the spike will get attenuated only by -3db. If its much higher then it might easily be attenuated by 80-99%.

How the filter deals with the spike depends on the filter. It can either block it by representing a big impedance against it. Or it can short the spike to ground.

The size of the spike does not matter in any case when it comes to filters as long as they can withstand the spike (speaking of hardware). A 20V spike at 50hz and a 70V spike at 50hz will get attenuated by the same amount if they go trough the same obviously what's left of the 70V spike is always going to be larger then the 20V spike.

The gain in the bode plot tells you how much a signal at frequency x gets attenuated. For pass filters you have a gain of 1 for some range of frequencies which you want (that is they pass trough without a change) but preferably want a gain of zero for those frequencies you do not want. An ideal filter would exhibit this behaviour but in reality this is not possible do to causality reasons. Some of the frequencies you actually want get attenuated and those that you dont want have nonzero gain.

If the filter specs say that the filter only handles a 5V spike at certain frequencies then you do not use it in applications where 10V spikes do occur frequently.
Last edited: