9.0 Conclusion
9.1 How Successful Was the Project
Overall I feel that this project was a success, I met all of the objectives that I set, these were to
1) Produce a high quality 20Mhz sampler with lots of memory.
2) Interface the sampler with a TMS320 microprocessor and a PC.
3) Process the sampled data with the TMS320.
4) Display the processed data with a C++ program.
Of all the above the embedded software was the most mentally taxing to produce as it required an understanding of PAL, digital filter theory, software weighting techniques specific to writing DSP code, general software weighting ability, a great deal of patience for testing the data produced byte by byte. The phase locking was the most difficult part of the embedded software to tackle as it requires the most software decision making and required knowledge of the colour burst phase shift.
This project has given me an insight into television that I could not have obtained any other way and it had motivated me towards digital television such as mpeg encoding and intelligent surveillance equipment.
9.2 What I learnt about hardware and software development.
Because of the size and complexity of this circuit significant time was spent on each individual module. When connecting the modules together there is a possibility of modules failing, slowing down the development of the final product. For the writing the code it is necessary for all the hardware modules to be working perfectly. The software development is the most complicated part of the circuit and requires the most time. If hardware modules fail then significant time would be lost whilst repairing the hardware. Also self-confidence would be lost. It is essential that the greatest of care is taken at all times in order not to damage work already
Pure software or written assignments are easier in the respect that they can not "catch fire" or "die". I have concluded from this project that the difficulty and the sheer hardness of complex electronic circuit design and implementation is not appreciated by most
10.0 Further Work
10.1 Moving images with less pixels.
The high sampling rate was required to satisfy the nyquist criteria and to allow for the digital demodulation filter to be of a reasonably low Q factor. After sampling, modulating and filtering the data it is clear that there is more data than required for transmission. A a standard PAL video line should comprise of ~740 pixels, but however stored in memory is 1188 Pixels. It would be easy to modify the algorithm to allow faster data transfer. The first method would be to only transmit every other pixel, this would result in the transmission of 600 pixels, but however the modulator and filter would still produce 1188 pixels, consuming valuable processing time. It is also possible to modify the modulating, filtering algorithm so it only modulates and filters every other pixel, thus producing only half the data to be transmitted. The second method is superior and could be modified even further to produce much less pixels, allowing for a fast data transfer time and moving images.
10.2 Motion detection
If the processed data was stored in a buffer, as well as being transmitted it would be possible to do a pixel by pixel comparison of the stored pixels and the recently produced pixels. By doing the comparisons and analysing the difference between pixels it would be possible to see how much each pixel had changed between a new frame and the previously stored frame. By setting a error threshold it is possible to allow an algorithm to decide if a pixel has been changed. On detection of a changed pixel an algorithm could then send data to a PC indication that it has detected a change in a pixel, ie movement.
Furthermore, by dividing an image into a matrix it would then be possible to send the cell number containing the changed pixel. If every pixel in the cell was transmitted with a header indication the location of the cell on a matrix it would be possible to transmit only portions of the image that had changed, thus allowing a much higher effective transfer rate and thus allowing moving pictures.
10.3 Parallel Processing
By linking several microprocessors together in such a format that each instruction executed was executed simultaneously on all microprocessors. This would result in multiple data streams being operated on simultaneously ( Single Instruction Multiple Data, SIMD). The computational load of processing each line would be divided between the microprocessors used, thus speeding up the PAL decoding process.
References.
[1] TMS320C5x users guide Texas Instruments.
[2]Fundamentals of digital electronics Dueck West.
[3]Telecommunications engineering J.Dunlop and D.G.Smith Stanley Thornes.
[4]How to interface PC’s R.A. Penfold Babani Computer Books.
[5]ISSI IS61C1024 128K x 8 High-Speed CMOS Static Ram data sheet July 1997
[6]Elantec EL4581C Sync Separator Data Sheet
[7]Motorola 74F161 Synchronous 4 bit binary counter Data Sheet
[8]Motorola 74F74 Dual D-Type Flip Flop Data Sheet
[9]Motorola 74F00 Quad Nand gate Data Sheet
[10]Analogue Devices AD775 8-bit 20MSPS 60mW Sampling A/D Converter Data Sheet