In an ideal world there would be 100% signal and no noise or interference. In real life, all signals come with added noise. Here are some common sources of noise.
- Magnetic tape hiss caused by random variations in the magnetic material coating the tape.
- Thermal noise. This is caused by random movements of electrons because they are warm. It is most obvious when a radio receiver is not tuned into any station. You can hear the hiss generated by the first amplifier stage in the receiver.
- Carbon resistor noise. The noise is caused by fluctuations in contact resistances between the carbon granules.
- Lightning strikes. These cause short intense bursts of radio noise. (Impulsive Noise)
- Radio noise from many sources such as industrial machinery and circuits being switched on or off (Impulsive Noise).
- Big Bang background radiation. This can be detected with very sensitive receivers and radio telescope dishes.
- Solar / stellar noise. The sun and stars radiate random noise radio energy. This can be detected with something as simple as an analogue TV receiver. It appears as a snow storm effect of the screen. With careful measurement, it is possible to detect more noise when the TV antenna is pointing at the sun.
Analogue Systems
When noise is added to analogue signals, it usually sounds like background hiss. Such noise can not be removed so the original clean signal can not be re-created. Techniques such as Dolby noise reduction make the noise less obtrusive but do not remove it. Faint analogue signals can disappear into the background noise.
Digital Systems - Regenerator
When noise is added to a digital signal, it is often possible to regenerate the original digital signal perfectly. This means that signals can be transmitted without any noise being added. This is why a phone call from across the world can often sound as clear as one from next door. The digital data crosses the entire network without being damaged.
The diagram above shows noise added to a digital signal.
Using a normal comparator to regenerate the digital signal, there would be errors. If the reference voltage is the gray line the noise is large enough cross the reference line. This causes the errors.
Using a schmitt trigger helps a lot because the signal has to rise above the red line for a ONE to be registered and go below the blue line for a ZERO to be registered. The spikes in the noise are no longer big enough to create errors. Of course, really bad noise would still cause errors but this circuit is much better than a simple comparator.
Digital noise immunity explains why digital phones, radio, TV and CD music all perform so well. On an analogue system this amount of noise would sound very unpleasant. The digital system is able to remove the noise.
If there is too much noise in a digital system, this causes effects like a TV picture breaking up into rectangular blocks. Audio output can be damaged too. Symptoms include gaps, squeaks and gurgles where a sound is repeated because the new data is missing. This last effect is less obtrusive than a gap or a squeak.
Decibels and Signal To Noise Ratio (S/N or SNR).
For voltage measurements ...
SNR = 20 log10(VS/VN)
For power measurements ...
SNR = 10 log10(PS/PN)
- SNR is the signal to noise ratio
- VS is the signal voltage
- VN is the noise voltage
- PS is the signal power
- PN is the noise power
- SNR is measured in dB (Decibels)
For noise free TV reception (no snow on the picture) the SNR needs to be 50dB. Speech becomes unintelligible if the SNR is less than about 10 dB.
Signal to noise ratio is measured in decibels. This is because received signals can be 1,000,000,000 times weaker than transmitted signals. The decibel scale is ideal for representing such a huge range of values. The decibel scale is logarithmic. A huge range of numbers is reduced to small manageable range. Base 10 logarithms work like this ...
Number |
Log 10 |
0.001 |
-3 |
0.01 |
-2 |
0.1 |
-1 |
1 |
0 |
10 |
1 |
100 |
2 |
1000 |
3 |
10000 |
4 |
100000 |
5 |
1000000 |
6 |
Example 1
- Signal level is 50µV
- Noise level is 1µV
- SNR = 20 log10(50/1)
- SNR = 20 x 1.7
- SNR = +34dB
- This is a nice strong signal with a little noise in the background.
Example 2
- Signal level is 1µV
- Noise level is 1µV
- SNR = 20 log10(1/1)
- SNR = 20 x 0
- SNR = 0dB
- This is a very weak signal with noise almost completely masking the signal.
Example 3
- Signal level is 0.1µV
- Noise level is 1µV
- SNR = 20 log10(0.1/1)
- SNR = 20 x -1
- SNR = -20dB
- The negative SNR indicates that the signal is completely swamped by the noise.
Interference and Electromagnetic Compatibility
Noise shows up as a background hiss. Interference comes in many forms.
Devices should be designed to perform their intended function without accidentally behaving in other ways. Here are some example of what can go wrong.
- Fluorescent lights radiate random radio interference modulated at 100Hz (Europe) or 120Hz (Americas).
- Cathode ray tubes in TV sets and computer monitors can radiate energy that damages radio reception.
- Strong radio signals can leak into more sensitive systems. On your HiFi or sound recording equipment, you might hear mobile phones using maximum power to connect to a base station.
- Microwave ovens leak radiation which can show up on TV receivers.
- CB, police, fire, amateur radio and other transmissions can leak into poor quality audio equipment in a similar way.
- If you live near a powerful broadcast transmitter, you might pick up the transmissions on unexpected devices like hearing aids, false teeth or even from your bath drain!
- Radar transmissions sometimes show up on HiFi or audio recording equipment. This has a very short distinctive buzz repeated every time the radar dish is pointing at your equipment.
- Cheap audio amplifiers pick up radio transmissions. These signals ought to be filtered out using low pass filters.
- Parasitic oscillations. This is when an amplifier (or voltage regulator) accidentally oscillates.
- Motors transmit radio noise pulses caused by the sparks where the motor brushes connect to the commutator.
- Computer networks can leak radiation that damages radio reception.
- The sparks from car/bike/lawn mower/etc petrol engine ignition systems can cause serious interference.
Cables are often fitted with a ferrite sleeve near the connector. This makes the cable very inductive and it behaves as a low pass filter blocking radio frequency radiation.
Signal Conditioning - regeneration
When a noisy signal is cleaned up with a Schmitt Trigger, this is called signal conditioning. There are other examples of signal conditioning ...
- pulses can be re-timed to remove jitter (the pulse arriving too soon or too late).
- statistical techniques can be used to re-construct data from a very noisy signal. Many samples are taken and averaged to see if the pulse should be a zero or a one.
- Dolby noise reduction is a kind of signal conditioning. When recording to tape, the high frequencies are exaggerated. When playing back, the high frequencies are reduced back to their correct level. High frequency background noise from the tape is reduced at the same time.
