Myauk said:
Is there any one who could explain to me the basic calculation scheme
for the RLC components in Line Filtering?
Is there any site explianing more details about the Power Line
Filtering, I mean, for example,the output filtering for inverter
designs?
My problem is especially with inductors, how can I design and construct
an inductor for line filtering?
I found many links but most of them explain about signal filtering,
antialiasing, and some DSPs but not the topic I am having problems
with.
A line filter is the same as any other signal LC filter, but with the
requirements that it handle much more current and much higher voltage than
typical signal filters. To design a filter, you first need to decide on the
response shape. If you are doing this professionally, you should avail
yourself of a filter design book - - the most useful of them being "Handbook
of Filter Synthesis" by Anatol I. Zverev. Then you need to decide what
frequencies to attenuate. Obviously the frequency to pass from your inverter
is 60Hz, but a filter to eliminate all frequencies above the fundamental
would certainly require rather large inductors and capacitors. I don't know
much about how you create the 60Hz sine wave in your inverter, but if you
can allow frequencies of 50KHz and below to pass through the filter
unattenuated, then the component values will be much more reasonable in
size. Lets say you need 60dB attenuation above 500 KHz, so the hash in your
source can't interfere with AM broadcast band reception, and a 3-pole filter
is all you have room for. Select a Butterworth response shape with cutoff at
50KHz since that will reach 60dB at ten times the cutoff frequency. Now we
have a bit of a problem - - filters normally have to be terminated with a
specific resistive source and load to behave according to the typical
response curve. It is also possible to terminate the filter at one end only,
with the other end either a "zero impedance" voltage source or an open
circuit. These are called extreme termination designs. The impedance of the
source in a line filter may be assumed to be close to zero, let's say, and
since the load voltage and current is known, the filter can be designed
around that single resistive termination. The normalized component values
for a 3-pole Butterworth singly terminated for a zero impedance source are
(from Zverev) L1 = 1.500 henries for the source end, C2 = 1.3333 farad, and
L3 = 0.5000 henry at the load end, where the load R = 1.000 ohm. To
de-normalize the design to your desired cutoff frequency and load
resistance, calculate a value for w = 2 x pi x Fc, where Fc is 50KHz in this
example, so that w = 314,159.27. Each inductor is de-normalized by the
formula L = L' x (Rt /w) where L' is the normalized value, and Rt is the
actual termination resistance. Each capacitor is de-normalized by the
formula C = C' x (1 / (w x Rt)) where C' is the normalized value. In this
example, lets say your load pulls 2 amps at 120 volts AC, so the Rt is 60
ohms. De-normalized values for 50KHz and 60 ohms are L1 = 286 microhenries,
C2 = 0.0707 microfarads, and L3 = 95.5 microhenries. These may still be too
large inductance values to be reached with open core ferrite inductors, so
you may need to consider ferrite EI core or pot-core or toroid inductor
design, or re-think the cutoff frequency or the number of poles in the
filter. If you use more poles, you can reach 60dB attenuation with a higher
specified cutoff frequency, and the components, though more numerous will be
smaller. Or you might decide 40dB attenuation is sufficient, which similarly
will allow you to raise the cutoff frequency and use smaller components.
Good luck
Chuck
