# Noise Filters

Noise is a problem that can occur in any electronic system. In general, it is the presence of any unwanted signal that can affect the overall operation of a system. It can come from a power source (60-Hz hum), from feedback networks, from mechanical systems connected to electrical systems, from stray capacitive and inductive effects, or possibly from a local signal source that is not properly shielded—the list is endless.
The manner in which the noise is eliminated or handled is normally analyzed by someone with a broad practical background and with a sense for the origin for the unwanted noise and how to remove it in the simplest and most direct way. In most cases the problem will not be part of the original design but a second effort in the testing phase to remove unexpected problems.
Although sophisticated methods can be applied when the problem can be serious in nature, most situations are handled simply by the proper placement of an element or two of a value sensitive to the problem.
Fig. 1: Noise reduction in a tape recorder.
In Fig. 1 two capacitors have been strategically placed in the tape recording and playback sections of a tape recorder to remove the undesirable high-frequency noise (rushing sound) that can result from unexpected, randomly placed particles on a magnetic tape, noise coming down the line, or noise introduced from the local environment.
During the record mode, with the switches in the positions shown $(R)$, the 100-$\mathrm{pF}$ capacitor at the top of the schematic will act as a short circuit to the high-frequency noise. The capacitor $C_{1}$ is included to compensate for the fact that recording on a tape is not a linear process versus frequency. In other words, certain frequencies are recorded at higher amplitudes than others.
In Fig. $2$ a sketch of recording level versus frequency has been provided, clearly indicating that the human audio range of about $40 \mathrm{~Hz}$ to $20 \mathrm{kHz}$ is very poor for the tape recording process, starting to rise only after $20 \mathrm{kHz}$. Thus, tape recorders must include a fixed biasing frequency which when added to the actual audio signal will bring the frequency range to be amplified to the region of high-amplitude recording. the phrase normal bias is used. Even after you pass the bias frequency, there is a frequency range that follows that drops off considerably.
Fig. 2: Noise reduction in a tape recorder.
Compensation for this drop-off is provided by the parallel combination of the resistor $R_{1}$ and the capacitor $C_{1}$ mentioned above. At frequencies near the bias frequency, the capacitor is designed to act essentially like an open circuit (high reactance), and the head current and voltage are limited by the resistors $R_{1}$ and $R_{2}$. At frequencies in the region where lower reactance level and reduce the net impedance across the parallel branch of $R_{1}$ and $C_{1}$.
The result is an increase in head current and voltage due to the lower net impedance in the line, resulting in a leveling in the tape gain following the bias frequency. Eventually, the capacitor will begin to take on the characteristics of a short circuit, effectively shorting out the resistance $R_{1}$, and the head current and voltage will be notch filter so that the original sound is not distorted by the high-frequency signal.
During playback $(P)$, the upper circuit of Fig. $1$ is set to ground by the upper switch, and the lower network comes into play. Again note the second 100-pF capacitor connected to the base of the transistor to short to ground any undesirable high-frequency noise. The resistor is there to absorb any power associated with the noise signal when the capacitor takes on its short-circuit equivalence. Keep in mind that the capacitor was chosen to act as a short-circuit equivalent for a particular frequency range and not for the audio range where it is essentially an open circuit.
Fig. 3: Noise generation: (a) due to a car alternator; (b) from a push-pull amplifier.
Alternators in a car are notorious for developing high-frequency noise down the line to the radio, as shown in Fig. 3(a). This problem is usually alleviated by placing a high-frequency filter in the line as shown. The inductor of $1 \mathrm{H}$ will offer a high impedance for the range of noise frequencies, while the capacitor $(1000 \mu \mathrm{F}$ to $47,000 \mu \mathrm{F})$ will act as a short-circuit equivalent to any noise that happens to get through.
For the speaker system in Fig. 3(b), the push-pull power arrangement of transistors in the output section can often develop a short period of time between pulses where the strong signal voltage is inductive effects, sees an unexpected path to ground like a switch opening, and quickly cuts off the speaker current. Through the familiar relationship $V_{L}=L\left(d i_{L} / d t\right)$, an unexpected voltage will develop across the coil and set a high-frequency oscillation on the line that will find its way back to the amplifier and cause further distortion. This effect can be subdued by placing an $R-C$ path to ground that will offer a low-resistance path from the speaker to ground for a range of frequencies typically generated by this signal distortion. Since the capacitor will assume a short-circuit equivalence for the range of noise disturbance, the resistor was added to limit the current and absorb the energy associated with the signal noise.
Fig. 4: Regulator: (a) effect of spike in current on the input side; (b) noise reduction.
In regulators, such as the 5-V regulator of Fig. 4(a), when a spike in current comes down the line for any number of reasons, there will be a voltage drop along the line, and the input voltage to the regulator will drop. The regulator, performing its primary function, will sense this drop in input voltage and will increase its amplification level through a feedback loop to maintain a constant output.
However, the spike is of such short duration that the output voltage will have a spike of its own because the input voltage has quickly returned to its normal level, and with the increased amplification the output will jump to a higher level. Then the regulator senses its error and quickly cuts its gain. The sensitivity to changes in the input level has caused the output level to go through a number of quick oscillations that can be a real problem for the equipment to which the dc voltage is applied: A high-frequency noise signal has been developed. One way to subdue this reaction and, in fact, slow the system response down so that very short interval spikes have less impact is to add a capacitor across the output as shown in Fig. 4(b).
Since the regulator is providing a fixed dc level, a large capacitor of $1 \mu \mathrm{F}$ can be used to short-circuit a wide range of high-frequency disturbances. However, you don't want to make the capacitor too large or you'll get too much damping, and large overshoots and undershoots can develop. To maximize the input of the added capacitor, you must place it physically closer to the regulator to ensure that noise is not picked up between the regulator and capacitor and to avoid developing any delay time between output signal and capacitive reaction.
In general, as you examine the schematic of working systems and see elements that don't appear to be part of any standard design procedure, you can assume that they are either protective devices or due to noise on the line that is affecting the operation of the system. Noting their type, value, and location will often reveal their purpose and modus operandi.