Retro SSTV anyone?

Where it all started as far as most are concerned and saw heavy use from the 60s through to the 80s. Colour and Hi-res modes have unfortunately pushed this system into the backwaters of SSTV. Time to resurrect interest in this simple analogue system.

Re: Retro SSTV anyone?

Postby Klaas Robers » Thu Jul 21, 2016 10:06 pm

Steve, would it help if you generate a separate sync pulse, so a real 5 msec. pulse, on the other stereo track of the signal? Then you can see what the video demodulator does.
If you are satisfied with that result, you can do the opposite and generate 1200 Hz sync bursts next to the same sync pulse. Then you can check the sync detection. The advantage is that you can fill in the video part of the signal with white, or black, or something wild. Each time you can see the time shift (delay) of the sync detection.
Wouldn't that help?
User avatar
Klaas Robers
"Gomez!", "Oh Morticia."
 
Posts: 1656
Joined: Wed Jan 24, 2007 8:42 pm
Location: Valkenswaard, the Netherlands

Re: Retro SSTV anyone?

Postby Steve Anderson » Fri Jul 22, 2016 2:39 pm

Klaas Robers wrote:...Wouldn't that help?

Quite possibly. Somehow I need to remove the possibility of the PCs noise affecting anything I'm observing. Which means at this stage removing/disconnecting the PCs altogether.

With the addition of some simple logic to the 'test modulator' idea posted a few days back I should be able to create a modulated sub-carrier that replicates a true SSTV signal quite closely.

However, what I'll probably do is dig out the hardware I used to create all the SSTV files upload some months back and feed that direct into this demodulator bypassing the PCs entirely. Simply take them out of the equation.

The video bandwidth of the modulator is still in question but that won't affect this jitter I'm seeing between the analogue sync and the logic version at the output of the demodulator.

Steve A.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: Retro SSTV anyone?

Postby Steve Anderson » Fri Jul 22, 2016 4:41 pm

PCs removed. Simple modulator as per a few days back done. Result? There's still this jitter but as this is basically driven by the zero-crossing technique I had my suspicions about it. There could be other factors that I need to consider.

But it is better without all the PC rubbish in the signal. The video recovery bandwidth is approx. -3db at 1.2kHz (1500-2300Hz square-wave modulation) - however, recall there is no filtering in the modulation signal path currently. The video rise/fall time is around 300us.

There's no spurious sync detection to full-amplitude 700Hz or any other frequency video modulation (1500-2300Hz).

So a few steps forward, a few back.

The acid test is to see how any of this makes any difference to the viewed picture - of course that's a subjective test, not objective.

Steve A.

Examining the results closer, the jitter is around 400us p-p, less than 10% of a sync pulse, I wonder if it might be visible? I'll find out in time. Actually it probably will be.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: Retro SSTV anyone?

Postby Klaas Robers » Fri Jul 22, 2016 11:37 pm

Yes Steve, that is visible. One pixel = 0.5 msec = 500 usec. So a jitter of 400 usec is clearly visible. You have seen it in the tetsbeeld file. Straight vertical lines are no more straight.

I see this too. To my opinion it is caused by several effects:

1. the phase of the 1200Hz for the sync is randomly at the start of the sync. This phase depends on the video content in the previous line. That might be the reason that in the beginning years of SSTV circuit designers had a separate 1200 Hz generator locked to the line frequency, and this was switched into the FM signal of the video. Then the phase was constant for every sync pulse, but the switching of oscillators introduced false phase jumps that were random from line to line and gave randomly white / grey / black dots on the beginning and the end of each line. So the FM oscillator should remain running and the sync must be co-modulated by that same oscillator.

2. When the 1200 Hz sync burst starts, the resonator of 1080 - 1320 Hz at -3 dB, starts resonating. Then the amplitude raises, but if you rectfiy that, the voltage raises pulse wise. Pulse repetition frequency will be 2400 Hz when you implemented double phase rectification. (I decided in 1973 that this was not needed, I implemented single phase rectification, which should be sufficient to my opinion.) A certain level is always exceeded in that pulse repetition rate, so sync pulses always start and end synchronised with the sync tone (of which the phase is randomly distributed). This gives a sync jitter of 0,4 msec. Unless you implemented a low pass filter in the rectified sync signal, so low, that the tone pulses are no more present and only a rising and falling DC is visible. This can be leveled by a comparator. The low pass filter could have a cut off frequency just above 100 Hz, say 120 Hz. (Currently I have a 3th order passive Butterworth filter. If I could do it again, I would opt for a 5th order filter, two inductors in stead of the current one.)

3. The video level (tone) just before and just after the sync pulse influences the swinging of the 1200 Hz resonator. You have addressed that before. A short black (0.5 msec) at the beginning and the end of each line helps against this, but is not in the CMD-standard. With other words: it should be possible without that.

I have still some jitter, due to cause number 3. I have a test pattern, a black field with vertical white lines. The lines are straight on my screen. But with white and black areas at the borders of a picture (testbeeld.wav) I see jitter, well, more a shift correlated to the whites and the blacks.

I have the feeling that your problem is caused by 2. Your 400 us is too obvious to this cause.
User avatar
Klaas Robers
"Gomez!", "Oh Morticia."
 
Posts: 1656
Joined: Wed Jan 24, 2007 8:42 pm
Location: Valkenswaard, the Netherlands

Re: Retro SSTV anyone?

Postby Steve Anderson » Sat Jul 23, 2016 12:23 pm

The interesting thing in all this is that with the Robot 70 style of demodulator there was no evidence of jitter, or at least I don't recall any. I think I would have noticed it if there were. But I'll fire it up and double-check. However the sync detection was done in a similar manner to most NBTV sync detectors by sync-slicing of the entire demodulated waveform. I didn't use a separate side-chain for the syncs.

There is a related yet different gremlin at work here, what I've called, 'sync-width modulation' in the past. This can be obviated by a front and/or back porch in the signal as per 525/625 TV as you mention. The effect is also visible in NBTV signals too. But at this point it's not the problem I'm having.

I notice that the MUTR NBTV files do incorporate front and back porches, but they're not very well done and vary from file to file.

Steve A.

A bit later - There is a small amount of jitter on the output of the Robot 70 demod, but it's far less, like not even 5% of what I'm getting from the MkII version, I had to really zoom in to notice it. However it's not caused by the same process, it's down to the small amount of 2400 and 3000Hz ripple after rectification and filtering of the sync signal.

So I'm confident all of the phase information is still there after limiting, it's the zero-crossing method that I'm using is missing it.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: Retro SSTV anyone?

Postby Klaas Robers » Mon Jul 25, 2016 11:28 pm

Steve, it should be possible for the detected sync pulse to start and end at EVERY moment, independent from the PHASE of the 1200 Hz signal. Because of the limited bandwidth, say 1080 - 1320 Hz, this 1200 Hz sine wave increases gradually at the beginning of a sync pulse. At a certain moment in time you decide that a certain tripping level is surpassed and the sync pulse starts. This certain moment can also lay at a moment that the value of the 1200 Hz sine wave happens to be zero, at least much lower than the peak value, positive or negative. In that case the previous peak was something too low, and the next the same amount too high, so the tripping moment has been just in between.

So: there should not be any correlation between the starting and stopping of the detected sync pulses and the phase of the 1200 Hz. If the sync pulse is decided on only just after a zero crossing of the 1200 Hz was there, you get a time jitter of 400 us.

This is possible if you double phase rectify the filtered 1200 Hz and send that signal through a low pass filter with a cut off frequency of 120 Hz. This gives a rounded sync pulse without any 1200 or 2400 Hz phase information, which can be level detected.
User avatar
Klaas Robers
"Gomez!", "Oh Morticia."
 
Posts: 1656
Joined: Wed Jan 24, 2007 8:42 pm
Location: Valkenswaard, the Netherlands

Re: Retro SSTV anyone?

Postby Steve Anderson » Tue Jul 26, 2016 11:27 am

Thanks for that Klaas, but I'm not using any rectification or filters anywhere - except one very gentle filter on the final video output.

I must admit I was about to shelve this for a while on Sunday (though not abandon it). But I spent some time yesterday on it and now I feel I have some acceptable results. I've a little more fine-tuning to do today.

Although it's not perfect it's a major improvement over the Robot 70 design. Whether you would notice the improvement on a 5" or smaller screen I'm not sure.

Hopefully I'll be able to post some results soon.

Steve A.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: Retro SSTV anyone?

Postby Harry Dalek » Tue Jul 26, 2016 3:29 pm

Hi Steve looking forward to the circuit or results I am not sure what you have
Come up with need some visual simulation .i am pretty happy with your reworked
Robot demodulator to make a better demodulator Is impressive to me.
The electromagnetic spectrum has no theoretical limit at either end. If all the mass/energy in the Universe is considered a 'limit', then that would be the only real theoretical limit to the maximum frequency attainable.
User avatar
Harry Dalek
"Fester! Don't do that to 'Thing'"
 
Posts: 5364
Joined: Fri Sep 26, 2008 4:58 pm
Location: Australia

Re: Retro SSTV anyone?

Postby Steve Anderson » Wed Jul 27, 2016 2:33 pm

A few measurements and tests done yesterday. Klaas, your multiburst file was -1.6db at 900Hz relative to the lowest frequency which is not too bad and far better than the Robot 70 design. The general video decode frequency response was as follows with square-wave modulation, at the higher frequencies the output was nothing like a square-wave - as you'd expect, so this is quite rough and ready, but it's a guide:-

100Hz 0bd (Reference).
200Hz 0db.
500Hz 0db.
700Hz -0.12db.
800Hx -0.90db.
900Hz -1.30db. A close agreement to above.
1000Hz -1.76db.
1100Hz -2.2db.
1200Hz -5.5db.

I would estimate the -3db point to be 1120Hz.

There is no ripple on the output at sub-carrier input frequencies well below 1200Hz.

Comparing the two demodulators on the TV screen via the up-converter there are very visible improvements. First that damn overshoot on the left-hand edge has gone, that was my number one hate. Second is the improved resolution, it takes an A-B comparison to really notice it but it is sure there.

However, I'm still not happy with the sync handling, it needs further work.

When the sun goes down here I'll do a couple of screen-shots, you'll have to put up with the ragged edges and ragged verticals in the picture though. This is a direct result of the sync jitter - unless I can fix that in the course of the day.

So for the moment it looks like I'm going the have to implement a separate sync-detect side-chain, something I was hoping to avoid. I was keen to get the chip and pin count to much less than that of the Robot 70 design. As it is currently it's just two 8-pin chips and a handful of passives. Oh well.

A couple of waveforms are below, the PC noise has returned as it's the only way I have of playing the wave files, also the 'scopes quantazion and screen resolution play a part - but there's no overshoot/ringing etc..

Steve A.
Attachments
Ascending Bars 1.gif
Ascending Bars 1.gif (20.75 KiB) Viewed 15573 times
Decending Bars 1.gif
Decending Bars 1.gif (20.58 KiB) Viewed 15573 times
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: Retro SSTV anyone?

Postby Steve Anderson » Wed Jul 27, 2016 6:17 pm

I've now built a separate sync side-chain whose output is as stable as the Robot 70 method.

Now my next task is to massage the software in the write processor within the up-converter to accept separate syncs from the video rather than using the composite input. So this isn't going to be today unfortunately.

But the jagged-edge screen-shots I will add later today.

Steve A.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: Retro SSTV anyone?

Postby Steve Anderson » Thu Jul 28, 2016 2:09 am

OK, I've had a bash at getting some screen-shots, but they're not as good as viewing directly. Need to get a decent camera, not this piece of junk I have.

So excuse and forget the ragged edges and ragged verticals within the frame, but things are better than before - once I sort out this sync headache.

In the 'English Cottage' shot the left-hand edge of the roof is much sharper, there's now a telegraph pole where there wasn't before...it's all so much better viewed directly.

'The Albion' has better detail and contrast on the street-lamp and the windows on the left-hand side of the pub. The shop sign beside the pub is now readable - just.

Finally, Klaas's Testbeeld03 file, I can't recall the frequencies of the gratings, but they're all there. This really does show up the sync problem,

The overshoot on the left-hand edge has gone - good riddance.

I'll practice with the camera to see if I can get better screen-shots. The improvement is far better than portrayed here.

...and remember this is still all on a breadboard - all of it. The wave files are the same between shots. Once this demod is sorted I plan to attack the modulator, but that's going to be a subtle change visually.

Steve A.
Attachments
Robot 70 English Cottage 1.jpg
Robot 70 English Cottage 1.jpg (364.95 KiB) Viewed 15552 times
MkII Demod English Cottage 2.jpg
MkII Demod English Cottage 2.jpg (334.7 KiB) Viewed 15552 times
Robot 70 The Albion 1.jpg
Robot 70 The Albion 1.jpg (344.19 KiB) Viewed 15552 times
MkII Demod The Albion 1jpg.jpg
MkII Demod The Albion 1jpg.jpg (331.29 KiB) Viewed 15552 times
Testbeelb03 1.jpg
Testbeelb03 1.jpg (329.24 KiB) Viewed 15552 times
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: Retro SSTV anyone?

Postby Harry Dalek » Thu Jul 28, 2016 3:37 pm

To me i had a look at the pub windows much more detail in the second of each picture ...you notice more towards the side of the picture .

Steve its this off a tv screen or computer ? i am thinking your converter to a tv screen but if i a wrong to a computer screen you can use xshot to screen grab the image to file ...yes camera shots are hard.
The electromagnetic spectrum has no theoretical limit at either end. If all the mass/energy in the Universe is considered a 'limit', then that would be the only real theoretical limit to the maximum frequency attainable.
User avatar
Harry Dalek
"Fester! Don't do that to 'Thing'"
 
Posts: 5364
Joined: Fri Sep 26, 2008 4:58 pm
Location: Australia

Re: Retro SSTV anyone?

Postby Steve Anderson » Thu Jul 28, 2016 4:51 pm

The up-converter has a 625-line output, standard black & white TV so you can't use a PC or even a PC monitor.

Of course you could do all this on a PC but I wanted a stand-alone box that is small. Both my PCs are desktops using large screens for my Autocad work. You can't sensibly do CAD on a piddly little laptop screen...even those of 18", it's just too small. Plus the monitors are 4:3, not widescreen. There are still are few manufacturers making 4:3 monitors, mine are Eizo, pricey but if you want 4:3 you have no choice.

Yes, the screen shots leave something to be desired, what doesn't help is that it's a good old fashioned colour CRT TV with a curved screen both horizontally and vertically, this may look better on the flat-screen TV downstairs. But I'm not even going to dare move the lashed-up breadboard arrangement. When it's a finished item I'll try the other TV.

Steve A.

I could probably do with dusting off the TV screen and properly cleaning the camera lens!
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Re: Retro SSTV anyone?

Postby Klaas Robers » Sun Jul 31, 2016 11:44 pm

Steve, the pictures are not that bad. I can't hardly see the curvature of the screen, but I can see the individual SSTV lines. So don't bother.

From the SSTV-testbeeld it is obvious that there is really a sync detection problem, as the timing errors propagate along the lines. A stable sync could do better, or can reveal other problems that are connected to the audio subcarrier phase. I still don't know what you did in your demodulation channel.

One idea that I havent completed is to create a sync fly wheel in an 8-pins PIC. The PIC then is running on a 20 MHz crystal and assumes a standard line length. If the incoming sync shows a timing deviation the line length is adjusted, but the outgoing sync is not yet affected. Small timing errors are not taken into account. In this way jitter in the sync is eliminated. Penalty is that the picture may be shifted somewhat in horizontal direction.

In the same way after 120 lines under certain conditions, a forced frame sync can be inserted.

In my monitor, if the internal electronic EPROM video generator is selected as a source, the displayed video is running via FM-modulation and FM-demodulation, but the sync is used directly from the source: the electronic video generator. I had to delay the sync to get it more or less synchronous with the 1200 Hz detected sync. For other sources the sync is used from the 1200 Hz detector, and there the PIC fly wheel can be inserted. I used a P12F683 in that place, but until now it only outputs the inputted sync. My PIC programming skills are not yet really too well developed.
User avatar
Klaas Robers
"Gomez!", "Oh Morticia."
 
Posts: 1656
Joined: Wed Jan 24, 2007 8:42 pm
Location: Valkenswaard, the Netherlands

Re: Retro SSTV anyone?

Postby Steve Anderson » Mon Aug 01, 2016 7:54 pm

I'm now holed up in Hong Kong for a few days, but once I get to China I'll be cut off from this forum. Those that my wish to contact me can do so via my e-mail which is on the NBTVA website. My e-mail works OK in China. Don't try and send an e-mail via this forum, it probably won't get though, use your e-mail client or gmail etc..

Yes, as you see Klaas there is a sync problem, I've built up a separate side-chain to handle it which is really quite stable. But until I return to Bangkok and re-work the up-converter I can't say yet if that's that answer.

As far as programming PICs is concerned, this site is a useful tool and source of ideas...

http://www.piclist.com/techref/piclist/index.htm

...it mainly deals with the 8-bit devices, even so there's little if you progress onto the PIC18 family, even though they're also 8-bit. It's mainly aimed at the PIC12 and PIC16 user.

I used to do quite a bit of programming on the Motorola MC6809 and it's variants in the early 80s, then I stopped that sort of stuff until a few years ago. But the concept was still there, it was a bit like riding a bicycle after a 30 year absence, wobbly at first, but I reached my destination unscathed. In Bangkok I'm often overtaking cars these days...it's really not that difficult when they are stationary...

Steve A.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

PreviousNext

Return to Cop McDonald's FM SSTV System

Who is online

Users browsing this forum: Google [Bot] and 3 guests

cron