Signal to Noise Ratio

From CCTV Information
Jump to: navigation, search

Introduction

There are many occasions when you see a picture on a monitor and think "that looks pretty lousy picture" The problem is to decide why it is lousy and whether anything be done about it. Some problems can be cured but the one that has always baffled me is the understanding of a ‘noisy’ picture and how it can be more objectively defined. So, as always, if it is a question of transmission or video theory, I called Peter Turnbull of Tecsec Ltd. for some help and a layman’s guide. (Sorry, layperson’s guide!.) I understand noise and what effect it has, but could I obtain some measure of it with simple equipment?


Notes on signal to noise ratio

The arch enemy of picture clarity on a monitor is noise, this is electronic noise that is present to some extent in all video signals. Noise manifests itself as snow or graininess over the whole picture on the monitor. There are several sources of noise; poor circuit design, heat, over-amplification, external influences, automatic gain control, transmission systems such as microwave, infrared etc. The important factor that determines the tolerance of noise is the amount of noise in the video signal, the signal to noise ratio. Note that every time that a video signal is processed in any way, noise is introduced.

The S/N ratio is exceedingly difficult to measure without special (and very expensive) test equipment. For instance to test for S/N ratio could cost in the region of £25,000 for the equipment. A less expensive way is to introduce a special filter to exclude the video signal and measure the remaining noise, from which the S/N ratio can be calculated. However even this filter can cost in the order of £1,000. Neither of these methods are practical in the field on actual installations, even if the equipment could be afforded.

This leaves the problem that when viewing the picture on an installed system, the assessment of the amount of noise is very subjective. One persons idea of a ‘noisy’ picture is not necessarily anothers. The quality of the picture can also be aggravated by other factors as mains hum, transmission losses, etc. These can be generally be overcome by isolation transformers, video line correctors, using twisted pair transmission etc. However the noise cannot be reduced by correction equipment, it is introduced at the source or in transmission systems. A common source of noise is when automatic gain control (AGC) is introduced at a camera in very low light conditions. This is why manufacturers state the minimum sensitivity of a camera with the AGC on but the S/N ratio with AGC off.

The only real way to reduce noise lies in correct system design, selection of equipment and transmission systems. Once it is there, it won’t go away and can only get worse.


Measuring S/N Ratio

There is though, one method of determining the S/N ratio which will give a reasonable guide. The only equipment needed is an oscilloscope with a bandwidth of 10 MHz and a very sensitive millivolt range. Connect the video signal to be checked to the ‘scope via a 75W impedance and view the black level of the video signal. The black level should be at 0.3 volts which is the top of the sync pulse. Normally this should be a thin horizontal line but when noise is present the line will be thicker. Keep increasing the sensitivity of the millivolts reading until the thickness of the line can be read to within 0.1 Mv. Note this reading in Mv, also the peak level of the video signal above the black level i.e. the white level. The video signal is measured above the black level, therefore if the black level is 0.3 v and the video white level is at 1.0 volts then the video signal is

1.0-0.3=.07 v. Signal to noise ratios are calculated from the peak to peak value of the video signal.

The signal to noise ratio is calculated as follows:

Cons73.gif which for a 1volt p/p video signal becomes Cons74.gif dB.

Where R is the signal to noise ratio and the signal and noise are measured in millivolts. The signal to noise is actually based on the RMS value. Therefore, without going into theory, add 3dB to the calculated value. To calculate the noise level in millivolts from the above, the formula can be transposed as below.

Cons75.gif

This gives the UNWEIGHTED value for the S/N ratio. When a filter is used to measure the S/N ratio it gives a WEIGHTED value which is about 8dB greater than the unweighted value. Once again many manufacturers do not state whether the value given is weighted or unweighted. It seems safe to assume therefore that they will show the value that enhances the specification to the maximum, which in this case would be the weighted value. If comparing different specifications, it would be reasonable to deduct 8 dB from a weighted value to arrive at the equivalent unweighted value.

In many cases the actual scene will not contain a great deal of black which can make it difficult to determine the black level. In these situations try to focus on an area with a vertical contrast between light and very dark areas. The best way, is to view a target made up with a vertical line having a black surface on one side and a white surface on the other.

In most common cameras the signal to noise ratio will be in the order of 55 dB, i.e. a ratio of 562 : 1. That is, the signal is five hundred and sixty two times greater than the noise signal. At this ratio the noise will be unnoticeable. The following guidelines interpret some ratios of signal to noise in terms of the subjective picture quality. A S/N ratio of 46dB is generally accepted as the threshold at which noise can be visually seen.

S/N ratio dB S/N ratio:1 Picture quality
60 dB 1,000 Excellent, no noise apparent
50 dB 316 Good, a small amount of noise but picture quality good.
40dB 100 Reasonable, fine grain or snow in the picture, some fine detail lost.
30 dB 32 Poor picture with a great deal of noise.
20 dB 10 Unusable picture.

Note that if the video signal is less than 1 volt p/p and the noise is constant, then the S/N ratio will be less. (i.e. worse.) Some manufacturers specify the sensitivity of cameras using vague terms, such as ‘usable video’, ‘50 IRE units’, ‘50% video signal’, etc. Using the camera at this level of sensitivity will have an adverse affect on the S/N ratio.

To save calculation, some typical values are listed in the following table. The following graph represents the relationship between signal to noise ratio in dB and the noise level in millivolts, for a 1.0v p/p and a 0.5 volt p/p video signal. ( The values have been adjusted for the RMS value of the noise measured.)

Table15.gif

Graph showing the signal to noise ratios for 1 volt and 0.5 volt peak to peak video signal

Cons76.gif


This chapter is supplied by Mike Constant and was originally published in CCTV Today. Mike is the author of 'The Principles & Practice of CCTV' which is generally accepted as the benchmark for CCTV installation in the UK.