Experimental Medium Band TV Signal Generation

[gary]

In anticipation of building a medium bandwidth mechanical televisor I have been experimenting with generating test signals using just an old pentium 4 computer and a simple 2 resistor DAC attached to the parallel port, although I am not contemplating building a PAL compatible mechanical televisor I thought that if I could generate 625 line monochrome PAL (yes I know that's really a contradiction but you know what I mean) I would be in good stead to produce MBTV (60 lines or greater?) when the time comes.

Here is a quick example of my results so far...
http://www.youtube.com/watch?v=Sa04GhS99pw

Edit: There are a few more there now.

Really this is a 312.5 line 50 fps progressive scan signal rather than 625 25 fps interlaced scan - but in terms of bandwidth and the display it's the same thing, at least in regards to line count and frame rate - you will understand that the horizontal resolution is necessarily limited.

This approach, albeit using PICs, is relatively popular method on the internet these days, but why buy PICs when you have half a dozen old P4s lying about the place

In truth I have discovered I can do the same thing, probably better, using the $4.50 Launchpad modules I referred to elsewhere on this forum, but I have always wanted to try it this way...

[Harry Dalek]

Hi Gary

Interesting work i like the idea of many different tv standards above 60 line and below say 400 ...perhaps we need another forum index for these experiments.

good luck on the project

Yes nothing like the power of old computers .

[kareno]

Hi Gary,

I had an idea a while ago to turn two 20k channels (e.g. stereo audio) into one 40k channel.

The approach that comes to mind involves modulating a 40kHz carrier with the first channel thereby producing a spectrum between 20k and 60k. A low pass filter removes the 40k to 60k part, leaving 20k to 40k for summing with the second channel to generate a full 0-40k.

The carrier phase is the problem. Maybe a 'pilot tone' can be embedded in the 'low band channel'. With a little overlap of the high and low bands, you could arrange that the high band includes an inverted copy of the pilot tone so that it vanishes when the two band halves are stitched together!

TV signal spectra tend to be combs so there's probably a 'space' in which you can hide the pilot tone.

It may be possible to do it even more simply than even this - there may be a way of computing two 20k channels which, when multiplied, produce a desired 40k channel. I'm even now thinking about the maths behind that.

[gary]

Hi Karen,

Yes I think that method could be made to work, however I think it would still require significant outboard electronics to recover the signal (as does all other methods that I am aware of at this time) - the beauty of the approach I describe above is that it requires only 2 resistors, and, of course, if you want more grey scale you could add more resistors, or a cheap DAC. Ultimately, the above approach is intended as a quick and cheerfully cheap way of lashing up a medium bandwidth (which I define for this purpose as exceeding audio bandwidth) signal for testing purposes. With a little tweaking I am pretty sure the above method could give close to full bandwidth 240 line TV - not that I am proposing that but it gives an idea of the range of signal that can be generated.

BTW it was really interesting writing software to generate a PAL (or NTSC as the difference is really only timing) signal, looking at it from a non-hardware angle (as we are used to) makes you realise just what an odd-ball format it really is. Thank goodness we don't have to throw away so much useful bandwidth for NBTV - otherwise the NB might stand for "No Bl**dy"

[kareno]

Hi Gary,

It seems someone should program a USB PIC to perform a high speed DAC function

As it happens, I have submitted an article on PIC-generated video for the next newsletter. I shall say no more here

[Klaas Robers]

The MUTR televisor kit contains a PIC with a simple DAC to have an always working test picture available.

[gary]

Hi Gary,
As it happens, I have submitted an article on PIC-generated video for the next newsletter. I shall say no more here

Beat me to it huh? but if it costs more than US$4.30 it doesn't count!

The MUTR televisor kit contains a PIC with a simple DAC to have an always working test picture available.

Aah! but can it produce monochrome PAL? (actually with reprogramming and an external clock it probably could).

[Klaas Robers]

I don't know Gary. It depends on the clock frequency and the speed of the instructions. I have no knowledge of programming a PIC.

But I think that the more modern 80C51 microcontrollers can do it. They perform 6 instructions per microsecond. Work with tables to output the bytes and you are very fast.

[kareno]

Hi Gary,

An idea recently occurred to me that uses a VGA monitor output.

Rather than scan convert an image, I imagined putting out a blocky black and white picture such that the luminance of any particular line conveys a block of digital data (potentially about twenty bytes per line).

I reckon a PIC could bring this data in, turn it into a smooth continuous stream of samples, and then deliver them to a DAC.

I'm biased towards this kind of solution because Windows isn't involved (much).

My faith in a USB peripheral solution is destroyed by my experience with USB/serial bridges. You'd think they could achieve phenomenal speeds being USB connected, but none I have acquired so far can do much better than 1kByte per second

[gary]

Hi Gary,

An idea recently occurred to me that uses a VGA monitor output.

Rather than scan convert an image, I imagined putting out a blocky black and white picture such that the luminance of any particular line conveys a block of digital data (potentially about twenty bytes per line).

I reckon a PIC could bring this data in, turn it into a smooth continuous stream of samples, and then deliver them to a DAC.

I'm biased towards this kind of solution because Windows isn't involved (much).

My faith in a USB peripheral solution is destroyed by my experience with USB/serial bridges. You'd think they could achieve phenomenal speeds being USB connected, but none I have acquired so far can do much better than 1kByte per second

Hmmm, somehow I missed this post way back when... Interesting idea but I would like to point out my MBTV solution certainly didn't involve windows - in fact it didn't involve any OS at all - unless you regard my simple programme that outputs the MBTV signal an OS - in a way it is at least in the sense that it is bootable.

Nope - no third party OS's for this solution just raw computer power.