Filters » Resonant filters

Resonant Filters

So far, the filter designs we've concentrated on have employed either capacitors or inductors, but never both at the same time. We should know by now that combinations of L and C will tend to resonate, and this property can be exploited in designing band-pass and band-stop filter circuits.

Series LC circuits give minimum impedance at resonance, while parallel LC (tank) circuits give maximum impedance at their resonant frequency. Knowing this, we have two basic strategies for designing either band-pass or band-stop filters.

For band-pass filters, the two basic resonant strategies are this: series LC to pass a signal or parallel LC to short a signal. The two schemes will be contrasted and simulated here:

FIGURE

Series LC components pass signal at resonance and block signals of any other frequencies from getting to the load.

FIGURE

There is virtually no signal attenuation within the "pass band" (the range of frequencies near the load voltage peak), unlike the band-pass filters made from capacitors or inductors alone. Also, since this filter works on the principle of series LC resonance, the resonant frequency of which is unaffected by circuit resistance, the value of the load resistor will not skew the peak frequency. However, different values for the load resistor will change the "steepness" of the Bode plot (the "selectivity" of the filter).

The other basic style of resonant band-pass filters employs a tank circuit (parallel LC combination) to short out signals too high or too low in frequency from getting to the load.

FIGURE

The tank circuit will have a lot of impedance at resonance, allowing the signal to get to the load with minimal attenuation. However, under or over resonant frequency, the tank circuit will have a low impedance, shorting out the signal and dropping most of it across series resistor Ri.

FIGURE

Just like the low-pass and high-pass filter designs relying on a series resistance and a parallel "shorting" component to attenuate unwanted frequencies, this resonant circuit can never provide full input (source) voltage to the load. That series resistance will always be dropping some amount of voltage so long as there is a load resistance connected to the output of the filter. It should be noted that this form of band-pass filter circuit is very popular in analog radio tuning circuitry, for selecting a particular radio frequency from the multitudes of frequencies available from the antenna. In most analog radio tuner circuits, the rotating dial for station selection moves a variable capacitor in a tank circuit.

The variable capacitor and air-core inductor of a simple radio comprise the main elements in the tank circuit filter used to discriminate one radio station's signal from another.

Just as we can use series and parallel LC resonant circuits to pass only those frequencies within a certain range, we can also use them to block frequencies within a certain range, creating a band-stop filter. Again, we have two major strategies to follow in doing this, to use either series or parallel resonance. The series variety bad-stop filter is shown in the figure below.

FIGURE

When the series LC combination reaches resonance, its very low impedance shorts out the signal, dropping it across resistor Ri and preventing its passage on to the load.

FIGURE

Next, we will examine the parallel resonant band-stop filter.

FIGURE

The parallel LC components present a high impedance at resonant frequency, thereby blocking the signal from the load at that frequency. Conversely, it passes signals to the load at any other frequencies.

FIGURE

Once again, notice how the absence of a series resistor makes for minimum attenuation for all the desired (passed) signals. On the other hand, the amplitude at the notch frequency is very low. In other words, this is a very "selective" filter.

In all these resonant filter designs, the selectivity depends greatly upon the purity of the inductance and capacitance used. If there is any stray resistance, this will diminish the filter's ability to finely discriminate frequencies as well as introduce anti-resonant effects that will skew the peak/notch frequency.

A word of caution to those designing low-pass and high-pass filters is in order at this point. After assessing the standard RC and LR low-pass and high-pass filter designs, it might occur to a student that a better, more effective design of low-pass or high-pass filter might be realized by combining capacitive and inductive elements together.

FIGURE

The inductors should block any high frequencies, while the capacitor should short out any high frequencies as well, both working together to allow only low frequency signals to reach the load. At first, this seems to be a good strategy and eliminates the need for a series resistance. However, the more insightful student will recognize that any combination of capacitors and inductors together in a circuit is likely to cause resonant effects to happen at a certain frequency. Resonance can cause strange things to happen.

FIGURE

What was supposed to be a low-pass filter turns out to be a band-pass filter with a peak somewhere around 526 Hz! The capacitance and inductance in this filter circuit are attaining resonance at that point, creating a large voltage drop around C1, which is seen at the load, regardless of L2's attenuating influence. The output voltage to the load at this point actually exceeds the input (source) voltage! A little more reflection reveals that if L1 and C2 are at resonance, they will impose a very low impedance on the AC source, which might not be good either. We'll run the same analysis again, only this time plotting C1's voltage, vm(2), and the source current, I(v1), along with load voltage, vm(3):

FIGURE

Sure enough, we see the voltage across C1 and the source current spiking to a high point at the same frequency where the load voltage is maximum. If we were expecting this filter to provide a simple low-pass function, we might be disappointed by the results.

The problem is that an L-C filter has an input impedance and an output impedance which must be matched. The voltage source impedance must match the input impedance of the filter and the filter output impedance must be matched by "rload" for a flat response. The input and output impedance is given by the square root of (L/C).

Taking the component values from the following figure, we can find the impedance of the filter, and the required Rg and Rload to match it.

In the figure, we have added Rg - 316Ω to the generator and changed the load Rload from 1000Ω to 316Ω. Note that if we needed to drive a 1000Ω load, the L/C ratio could have been adjusted to match that resistance.

FIGURE

In the following figure, the "flat" response of the L-C low pass filter when the source and load impedance match the filter input and output impedances.

FIGURE

The point to make in comparing the response of the unmatched filter to the matched filter is that variable load on the filter produces a considerable change in voltage. This property is directly applicable to L-C filtered power supplies. The power supply voltage changes with a change in load, making it undesirable.

This poor load regulation can be mitigated by a swinging choke. This is a choke, inductor, designed to saturate when a large DC current passes through it i.e., the DC current creates a "too" high level of flux in the magnetic core, so that the AC component of current cannot vary the flux. Since induction is proportional to d<t>/dt, the inductance is decreased by the heavy DC current. The decrease in inductance decreases reactance XL.

Decreasing reactance, reduces the voltage drop across the inductor; thus, increasing the voltage at the filter output. This improves the voltage regulation with respect to variable loads. Despite the unintended resonance, low-pass filters made up of capacitors and inductors are frequently used as final stages in AC/DC power supplies to filter the unwanted AC "ripple" voltage out of the DC converted from AC.

Why is this, if this particular filter design possesses a potentially troublesome resonant point?

The answer lies in the selection of filter component sizes and the frequencies encountered from an AC/DC converter (rectifier). What we're trying to do in an AC/DC power supply filter is separate DC voltage from a small amount of relatively high-frequency AC voltage. The filter inductors and capacitors are generally quite large (several Henrys for the inductors and thousands of μF for the capacitors is typical), making the filter's resonant frequency very low. DC has a "frequency" of zero, so there's no way it can make an LC circuit resonate. On the other hand, the ripple voltage is a non-sinusoidal AC voltage consisting of a fundamental frequency at least twice the frequency of the converted AC voltage, with harmonics many times that in addition. For plug-in-the-wall power supplies running on 60 Hz AC power (60 Hz United States; 50 Hz in Europe), the lowest frequency the filter will ever see is 120 Hz (100 Hz in Europe), which is well above its resonant point. Therefore, the potentially troublesome resonant point in such a filter is completely avoided.

FIGURE

With a full 12 volts DC at the load and only 34.12 of AC left from the 1 volt AC source imposed across the load, this circuit design proves itself to be a very effective power supply filter.

The lesson learned here about resonant effects also applies to the design of high-pass filters using both capacitors and inductors. As long as the desired and undesired frequencies are well to either side of the resonant point, the filter will work fine. But if any signal of significant magnitude close to the resonant frequency is applied to the input of the filter, strange things will occur.