SSTV mode information
compiled by G3ZLS
Home ] [ Principles ] RGB Modes ] YUV Modes ] Feedback ]

General Principles

The use of frequency modulation to convey the picture information makes analogue SSTV very immune to impulse interference. Although co-channel amplitude interference will degrade an SSTV picture the FM capture effect means that a strong SSTV signal will tend to override weaker signals. 

The image on the right shows a typical spectrum display for an SSTV transmission.

SSTV spectrumThe left side of the display represents an audio frequency of 1,000 Hz and the right hand side represents 2,400 Hz. The three vertical lines are at 1,200 Hz, 1,500 Hz and 2,300 Hz.

Picture Content

The image information is transmitted between 1,500 Hz and 2,300 Hz.  Black corresponds to a frequency of 1,500 Hz and white corresponds to 2,300 Hz.  The 800 Hz band in between represents a continuous gray scale between pure black and pure white, so that 1,900 Hz represents mid gray.

The number of gray levels which can be transmitted is limited by the ability of the receiving software to discriminate between one frequency and another, but in practice it is possible to display a continuous spectrum between black and white without visible steps.

It should be noted that since only a single audio sub-carrier is used the transmission will only contain a single frequency at any instant.  The above display represents the information gathered over a short period, which includes a segment of image information and at least one sync pulse.  This particular image segment contains some pure black picture content ranging through to mid gray, but there is no pure white content.

Line Sync Pulses

Line synchronization pulses (sync pulses) are transmitted at a lower frequency than the image information, normally at the start of each line so that the start of the line can be clearly identified. They are visible in the spectral display as a peak around 1,200 Hz. (The small peak around 1,350 Hz is probably noise.)

The main function of the sync pulses is to allow an image to be properly synchronized even if the start of the transmission was missed.  Once horizontal synchronization has been established the sync pulses can generally be ignored provided the timing of the receiving equipment is accurately aligned to the transmitting equipment.  This has the advantage that noise and interference will not affect picture synchronization, but if there is any timing discrepancy the picture will be received with a slant.  For this reason, some programs provide an option to make use of the sync pulses to re-synchronize the lines as the picture comes in.  Although this may correct any slant, results can be unreliable under less than perfect conditions.

If the image is sent by SSB and is not tuned in correctly this will cause the spectrum to shift to the right or the left. The signal is correctly tuned when the sync pulses fall at 1,200 Hz.

Decoding Considerations

At any one time during picture transmission the frequency of the SSTV sub-carrier determines the gray level of the image. Thus, picture quality depends on the ability of the receiving equipment to accurately measure the instantaneous frequency of the sub-carrier. For most practical purposes one half cycle is the theoretical smallest quantum of information which can be sent using such a system.

At black level: f = 1,500 Hz, so cycle time t = 666 uS and t/2 = 333 uS

At white level: f = 2,300 Hz, so cycle time t = 434 uS and t/2 = 217 uS

Waveform symmetryThere is therefore little point in attempting to transmit picture information with a pixel time significantly less than 217 uS.

Although digital signal processing techniques have been employed to try and accurately measure the frequency of the received waveform within less than half a cycle with some degree of success, the tried and tested method is to measure the time intervals between the points at which the waveform passes through the zero axis. This has proved to be a simple, accurate and robust technique, although it is subject to one major pitfall.

Consider a constant audio frequency as produced by a uniform gray image. If the signal input circuits are less than perfect they can easily produce an asymmetrical waveform, shifting the waveform above or below the zero axis. This will cause adjacent half cycles to be alternately longer and shorter, which is interpreted as two different shades of gray. Depending on the number of half cycles which go to make up each pixel, this can result in noticeable patterning on the image.

The accepted solution is to average the last two half cycles so that variations due to waveform asymmetry are smoothed out. This effectively eliminates such patterning problems, albeit at the expense of image resolution. As a consequence, the practical limits of resolution are generally closer to the full cycle period rather than the theoretical half cycle limit.  Therefore, pixel times of less than about 400 uS cannot be achieved without degrading picture sharpness, and significantly greater times are required to obtain true pixel-for-pixel reproduction.  

20 January 2008