radio

[Lowtone]

http://www.mediafire.com/view/9wizqhdr3 ... vradio.PNG

http://www.mediafire.com/view/fckvar8tp ... testok.PNG

Ah thanks, now my next test will be to try with the radio
(when i tryed sstv last time i got the sync lost i don't know why)

edit : well i got no luck with radio, the picture cannot be recognized
I tried with NBSCPlayer, the picture was less unstable, but not good either

[gary]

Because the format doesn't use syncs there can be no "instability".

You may get a poor picture for many reasons but it should be a rock solid poor picture if you know what I mean.

If the picture is jumping about all over the place you must be doing something else wrong wrong like NOT turning off sync processing, give us more information (in fact ANY information) we may be able to assist.

One immediate suggestion is have you determined that you PC is not a) swapping left and right channels, b) inverting the audio - both of these are common issues and can be compensated for in TBP.

I can only assume about what you are doing since you have provided little in the way of detail:

1) Playing a Chris Long Shortwave (CLS) either from an audio player or directly from Video2NBTV itself

2) Using a radio transmitter to send the CLS signal (what kind of transmitter AM? FM? SSB? etc)

3) Using a radio receiver to receive the CLS signal

3) Plugging the output of the radio into the "line in" of your PC's audio card

4) Using TBP or NBSC (and here is a thing - I am pretty sure that NBSC doesn't support CLS as he has probably never heard of it before this - what say you Dom? You may get some sort of picture as it is 32 line, but at half frame rate, and is syncless, but it will definitely NOT be able to lock) to view the CLS signal through the soundcard.

In any case I can assure you that the above process works and works well and I am doing it and enjoying a rock solid clear picture as we speak.

I don't have a Amateur Radio License so my transmission is limited to a few metres and if you were doing this for real you would have issues like rain fade, multiple propagation paths, phase variations, interference, etc etc. BUT the picture will remain locked and will not be jumping around and at least some of the time should be at least as good as the Chris Long video on Youtube.

Give us more information and I am sure I/we can help but you will have to go through it step by step and detail each one for us.

BTW for any forum new comers, young scallywags etc. reading this, Steve Anderson has explained this a number of times: There is a wealth of knowledge and experience on this forum and if you are looking for help you will get it much more often than not, BUT, it is NOT helpful at all to just come on and say "it doesn't work". You MUST explain exactly what you are doing and in what way "it doesn't work". In regards to Dom and I's software in particular, (and I am sure I can speak for him here), we are not in the business (or rather hobby) of producing software that doesn't work - sure any software will have the odd bug (which we would be delighted to hear about BTW), but in general we have tested it ad nauseam ourselves, and many other people have used it successfully around the world. The moral of the story is, expect things to work, and if it doesn't, assume that you have done something wrong and methodically check your work at every stage. If you have checked and tried everything then come on the forum, explain what you have done and tried, and what your results are, and I am sure we can solve the problem quickly.

[Lowtone]

Ah yes the "it doesn't work" is a way to speak, i don't mean the softwares are themselves faulty, this can come from anywhere else, especially i'm getting back to NBTV but it's been a few years that i was away, so i forgot how to use the softwares.
I can have ideas, but i'm not an electronician, mathematician or programmer. I happened to be here because i like the History of télévision ( and related techniques ) and the technical things.
My message was not very detailed, i'm sorry

There was two files, one shortwave, and one standard format.
http://www.mediafire.com/listen/ai92pbx ... ong-sw.wav
http://www.mediafire.com/listen/4vr9pyd ... -nbtva.wav
Both sent on PMR446 ( NFM ) a few metres aways. ( this radio is only able to NFM and FM formats, no WFM/AM/SSB )
There was only two concrete walls until the receptor
There was a few ( quite few ) disturbances during the transmition ( to see what can happen )
I put a portable digital recorder near the receptor
i took the file from this, ( direct transfert, no analog line-in step )
some parts are raw, and i filtered some others ( to remove noise that is out of the bandwith )
I splitted this in two separate files for both formats
http://www.mediafire.com/listen/rzh49aw ... islong.wav
http://www.mediafire.com/listen/t4reac1 ... onbtva.wav
( so that got some stability in NBSC was the NBTVA one )
There cannot be left and right inversion, because the same signal is on both tracks
I haven't checked the phase inversion

Why do this format don't use sync pulses ?

[gary]

Ah yes the "it doesn't work" is a way to speak, i don't mean the softwares are themselves faulty, this can come from anywhere else, especially i'm getting back to NBTV but it's been a few years that i was away, so i forgot how to use the softwares.
I can have ideas, but i'm not an electronician, mathematician or programmer. I happened to be here because i like the History of télévision ( and related techniques ) and the technical things.
My message was not very detailed, i'm sorry

No need to apologise, I just wanted to make it clear that the more information you can give us the better chance to address your problems (also I am very interesting in what you are doing AND by documenting what you are doing it is likely to be of use to others down the track).

I also realise that English is not your first language and translating all the time must be a pain and make you want to be as brief as possible, I do appreciate you going to that trouble.

There was two files, one shortwave, and one standard format.
Both sent on PMR446 ( NFM ) a few metres aways. ( this radio is only able to NFM and FM formats, no WFM/AM/SSB )

Aaah, now I am not all that familiar with narrow band fm although it seems it is just a narrow band version of FM.

Using FM to transmit FM has been a problem generally: viewtopic.php?t=405

However I was under the impression that the only major problem was due to this: http://claessonedwards.com/index.php?op ... &Itemid=89

This attempt was moderately successful: https://www.youtube.com/watch?v=TGgmCiJ9iEM

Certainly the channel bandwidth of the PMR446 is wide enough (especially FM) - but I wonder about it's phase linearity? That is relatively unimportant for audio but not for video.

There was only two concrete walls until the receptor
There was a few ( quite few ) disturbances during the transmition ( to see what can happen )
I put a portable digital recorder near the receptor
i took the file from this, ( direct transfert, no analog line-in step )
some parts are raw, and i filtered some others ( to remove noise that is out of the bandwith )
I splitted this in two separate files for both formats ( so that got some stability in NBSC was the NBTVA one )
There cannot be left and right inversion, because the same signal is on both tracks

Fair enough

I haven't checked the phase inversion

If you are getting any image at all then it can't be inverted.

Why do this format don't use sync pulses ?

Because the (important) frequency components of a sync pulse are very low and mostly outside of the frequency range of typical audio and hence is high pass filtered out anyway.

In any case they are mostly unneeded anyway - the only thing you get that is useful is automatic framing - if you have frame and line adjustments you can quickly centre the picture and leave it.

If the picture was coming from, say, a mechanical camera the situation is different as the scanning rate is never perfect and the line syncs help to minimise the effects.

When the picture is computer generated the scan rate is (nearly) perfect - this is why the sample rate of the CLS is 48 kHz - that is the native sample rate of most codecs in a sound card and is an integer multiple of the xtal oscillator on the card and therefore quite (but not perfectly) accurate and drifts only very slowly over time (which can easily be adjusted for). Some sound cards are worse than others in this respect. In any case, for Amateur Radio channels syncs have proved to be more trouble than they are worth.

I would appreciate it very much if you could post the unfiltered CLS wave file you recorded - if nothing else I would be very interested in having a look at it.

[Lowtone]

Hello Gary,
thanks for your reply

I also realise that English is not your first language and translating all the time must be a pain and make you want to be as brief as possible, I do appreciate you going to that trouble.

Oh it was mostly because i was also doing something else. Language was not the problem, i was just giving up and switching to another activity.
I don't transalte, i just try to think in English. But in France language lessons are not very good, so expect from me only limited vocabulary and simple sentences.

Aaah, now I am not all that familiar with narrow band fm although it seems it is just a narrow band version of FM.

There 3 FM formats.
the WFM, is what is used by commercial FM radio on VHF II, about 15kHz wide
FM is narrower and used on some special radios
NFM is even narrower, and is the standard used in PMR446, it ± starts at 250Hz and ends around 4kHz only.
Those radios are made for voice communication, so the format suits that, and can compare to old analog telephone lines, with limited bandwith. ( But the law allows any kind of transmission, SSTV, NBTV, RTTY, AX25, CW etc… )

I got also a WFM transmitter, that can output 1 or 7W, but 7W is illegal without licence. I'll may try NBTV too with this, but it's all electronic with pre-emphasis etc…

This attempt was moderately successful: https://www.youtube.com/watch?v=TGgmCiJ9iEM

nice one !

Certainly the channel bandwidth of the PMR446 is wide enough (especially FM) - but I wonder about it's phase linearity? That is relatively unimportant for audio but not for video.

I don't know either, plus there is the whole line of it.
One computer was reading my file, then the signal gone towards radio with a cable, but for the receptor we need to take account of the linearity of the speaker, and the one for the recorder's microphones. Errors can cumulate.

Because the (important) frequency components of a sync pulse are very low and mostly outside of the frequency range of typical audio and hence is high pass filtered out anyway.

Could it be possible to make a kind of "inverted" synch pulse, or put it in a place where it is not erased ?
the SSTV sync system is interesting, it's an 1200 Hz beep wich is steady, and remains way under the signals

for Amateur Radio channels syncs have proved to be more trouble than they are worth.

I imagine it's due to interferences, or reflections of signal that can double the sync and mess the picture up

this is why the sample rate of the CLS is 48 kHz - that is the native sample rate of most codecs in a sound card and is an integer multiple of the xtal oscillator on the card and therefore quite (but not perfectly) accurate

I like this. And i rage everytime i see people using 44.1 in their projects. It's only good for CD.

I would appreciate it very much if you could post the unfiltered CLS wave file you recorded - if nothing else I would be very interested in having a look at it.

Unfortunatly i ditched the original file, but in both files there are filtered and non filtered parts, spetrograph can reveals this clearly

[gary]

Oh it was mostly because i was also doing something else. Language was not the problem, i was just giving up and switching to another activity.
I don't transalte, i just try to think in English. But in France language lessons are not very good, so expect from me only limited vocabulary and simple sentences.

Well I wish my French was a fraction as good as your English.

I don't know either, plus there is the whole line of it.
One computer was reading my file, then the signal gone towards radio with a cable, but for the receptor we need to take account of the linearity of the speaker, and the one for the recorder's microphones. Errors can cumulate.

Aaaah I didn't realise you were interfacing to the receiver via loudspeaker and mic, that will introduce a lot of variability - although it doesn't explain the results you were getting (at least as far as your .png file indicated). That was indicative of loss of sync as some of the lines appear to be offset against the main picture and as there ARE no syncs that is very strange unless that picture was from your NBTVA format experiment, in which case it isn't surprising at all.

I suppose the transmission frequency you are using doesn't give many options in terms of what receivers you can use (i.e. one with an earphone socket).

Could it be possible to make a kind of "inverted" synch pulse, or put it in a place where it is not erased ?
the SSTV sync system is interesting, it's an 1200 Hz beep wich is steady, and remains way under the signals

Well inverting the pulse would not help things as the frequencies involved are the same.

To answer this question properly you need some understanding of Fourier analysis, but to simplify things a finite width pulse (such as a sync pulse) is actually the infinite summation of a DC component and a series of sine wave harmonics e.g. 0f + f + 2f + 3f... so obviously over a finite bandwidth some of those components are lost (filtered out) which, especially if the low frequency response is poor, substantially changes the shape of the pulse, add that to phase distortion, noise, multiple propagation paths and a sync detector has problems detecting syncs.

Besides all that, syncs are just not needed. Baird didn't use them.

The problem can indeed be mitigated by changing from a rectangular pulse to a short segment of a high frequency sine wave - however detection of this is MUCH more difficult than pulse type syncs - difficult enough in software - very difficult in hardware (and don't forget this stuff is intended to be used as a source for a mechanical monitor).

It's only good for CD.

Not even that - the worst technical decision ever made IMHO - all (or mostly) to be compatible with the line and frame rates of PAL and NTSC television - WTF?

I would appreciate it very much if you could post the unfiltered CLS wave file you recorded - if nothing else I would be very interested in having a look at it.

Unfortunatly i ditched the original file, but in both files there are filtered and non filtered parts, spetrograph can reveals this clearly

Aaaah that's a pity I would have been very interested in analysing it. Oh well if you ever decide to experiment again we can pick up the issue then. Cheers.

[Lowtone]

Well I wish my French was a fraction as good as your English.

merci

Aaaah I didn't realise you were interfacing to the receiver via loudspeaker and mic, that will introduce a lot of variability - although it doesn't explain the results you were getting (at least as far as your .png file indicated). That was indicative of loss of sync as some of the lines appear to be offset against the main picture and as there ARE no syncs that is very strange unless that picture was from your NBTVA format experiment, in which case it isn't surprising at all.

The first links on the top of the page with the testcard ?
This one didn't use the radio, this was only a test with the computer. The file was encoded to sw format ( so with no sync ) and played back.
But what appears to look like a sync is from the original picture ( look at the colour version ). I reserved a guard at the bottom of the picture ( in dark red ) in order to avoid to put some picture here ( according to the NBTVA standard ). The white pixels on the begining of the first line was an idea of mine, in order to strongly keep sync in the MUTR toy ( it is quite sensitive and mistakes black with sync pulse, picture often got cut in half or jump ). But usualy if I remind well, your software put a black signal over this, so my idea can't work.
Well, on every 32 line video, i used this red and white pattern at the bottom of the screen ( except on this radio experiment because i took the Baird Phonovision archives from a montage i did prior )
So, it is not technically sync pulses, but just a dark area

I suppose the transmission frequency you are using doesn't give many options in terms of what receivers you can use (i.e. one with an earphone socket).

I can do, but when i record radio i usually use open mic, so i did the same. I was not very scientific, but next time i can plug the recorder. And even add a better antenna.

Besides all that, syncs are just not needed. Baird didn't use them.

He invented so many things, but not the TBC
I don't know if it's possible to make a program that can "add" or restore the pulses, and then output
( and even if possible, would it be that useful ?)

(and don't forget this stuff is intended to be used as a source for a mechanical monitor).

Oh yes, i must find where it is, but definitly should test with the real televisor

Not even that - the worst technical decision ever made IMHO - all (or mostly) to be compatible with the line and frame rates of PAL and NTSC television - WTF?

Prior computers, they used to store digital music on video tape. The digital signal was sent to a converter that changed it to white and black bars, on the video portion of a video tape. Hence the Nyquist theorem, they needed twice the hearing range for the sample frequency ( about 40 ) but the U-matic tape allowed 44.1…
They waited a until the advent of DVD to switch to 48, and blu ray for 96
( but there are also retrocompatible hi-res audio formats, at… 88.2… )

I found also this ham experiment on youtube
https://www.youtube.com/watch?v=Sc759yJrNIU
The quality is not that bad, i would like to acheive this

[gary]

My mistake I was referring to your testnbtv.png but I realise that was before you realised you had to turn syncs off.

I assume that you had no problems playing the test video nbtvbaird-chrislong-sw.wav directly on TBP and that you only had a problem when it came from the radio?

NBTVer is ME! That was a test performed between Vic Brown and I think Peter Smith using Vic Brown's VSB system - I had been helping out with some software based filters and this was a results sample he sent me. I think I did some time base correction on it prior to posting, I can't remember for sure. You can clearly see there that trying to pick sync pulses out of all that noise would somewhat problematical - that is an example of syncless transmission over quite a distance.

[Lowtone]

I assume that you had no problems playing the test video nbtvbaird-chrislong-sw.wav directly on TBP and that you only had a problem when it came from the radio?

Yes this mode works, but not by radio ( at least for me )

NBTVer is ME! That was a test performed between Vic Brown and I think Peter Smith using Vic Brown's VSB system - I had been helping out with some software based filters and this was a results sample he sent me. I think I did some time base correction on it prior to posting, I can't remember for sure. You can clearly see there that trying to pick sync pulses out of all that noise would somewhat problematical - that is an example of syncless transmission over quite a distance.

Ah nice job

I had an idea, i don't know if someone ever thought of it or even try, or if it is possible but :

http://www.mediafire.com/view/hj9v1cgkyt1x0ld/log.PNG
What if we use a kind of compander, that compress the signal prior the transmission, and then it will be expanded at the reception ( with loss of details )
Then there may be a pulse restauration, and the signal outputs from computer in NBTVA format.


( also what do you think of the idea i exposed in the previous message about the first pixels of the first line being white and acting like a frame sync ?

[gary]

Frequency companding is an interesting concept and may be worth investigating as a means of increasing the frequency response of a narrow bandwidth channel. Unlike audio I don't think there are any "unimportant" frequencies in the video spectrum however. None the less, if an NBTV video was analysed on a frame by frame basis it may be possible to determine low amplitude frequency regions that could be discarded without substantial loss - trading some contrast for better frequency response.

If I had time I would do some experimentation, however it would be very computationally intensive and I am not sure it could be done in realtime.

I think your idea of the first pixels being white is an excellent idea, alas for you, it is not original - Baird himself used the technique (albeit black) - and has been replicated many times in both black and white - it doesn't really matter if it is black or white the main problem is that if the scene becomes black or white then you lose sync - thus for early television experimental transmissions black was not allowed!

But the problem of this technique for radio transmission is that it doesn't overcome the problem of severe noise which, unfortunately, is always present.

You still seem to be determined to find a way to get synchronisation pulses into the video signal and I am wondering why?

[Lowtone]

Unlike audio I don't think there are any "unimportant" frequencies in the video spectrum however.

We may disagree on this, since i'm a musician and i also like to mix multitrack recordings. I can be picky on this
But i got your point

low amplitude frequency regions that could be discarded without substantial loss - trading some contrast for better frequency response.

Since it's radio -and noisy- here the quality won't be the main concen. The picture should be recognizable at least

I think your idea of the first pixels being white is an excellent idea, alas for you, it is not original - Baird himself used the technique (albeit black) -

ah ^^ black on the first line, intersting, so it was only a frame pulse, no pulses on the other lines ?

and has been replicated many times in both black and white - it doesn't really matter if it is black or white the main problem is that if the scene becomes black or white then you lose sync -

So the sync is always lost ?
I thought that since the other pulses are blacker-than-black, a white one would made a very high contrast and cannot be misinterpreted as a line pulse.

thus for early television experimental transmissions black was not allowed!

Hence the chess board

You still seem to be determined to find a way to get synchronisation pulses into the video signal and I am wondering why?

Because i remind having troubles with the MUTR to keep sync. And also, if the software can add/restaure pulses it will be compatible with the NBTVA standard.
I like to keep doing all the production line in one format. But well if it's not necessary…

[gary]

We may disagree on this, since i'm a musician and i also like to mix multitrack recordings. I can be picky on this
But i got your point

LOL - yes, I meant in terms of voice intelligibility - music is a completely different kettle of fish.

ah ^^ black on the first line, intersting, so it was only a frame pulse, no pulses on the other lines ?

No they were indeed line syncs, but if you have line syncs you can adjust speed to frame the picture. Have just a few white pixels would make it very susceptible to noise and level, and don't forget this is all AC coupled so the signal amplitude is varying with picture content so how would you know if it is white or grey floating high? This is where blacker than black (or indeed whiter than white) syncs are helpful - DC restoration.

So the sync is always lost ?
I thought that since the other pulses are blacker-than-black, a white one would made a very high contrast and cannot be misinterpreted as a line pulse.

Ok, I had trouble following your argument, so you have white pixels instead of a "missing pulse" in conjunction with blacker-than-black sync pulses? In that case I would suggest that you would almost always miss the next line sync or 2 because the white has driven the average signal level above the sync slice - if I get time I will create a bit of software to demonstrate what I mean.

Sync will only be lost if some part of the signal "looks" like a frame pulse but isn't - with a very noisy signal that can happen any time.

Because i remind having troubles with the MUTR to keep sync. And also, if the software can add/restaure pulses it will be compatible with the NBTVA standard.
I like to keep doing all the production line in one format. But well if it's not necessary…

I see. But, of course, the signal does not need to have HAD syncs in it to have syncs added - they can be added any time either in software or hardware.

Obviously if you could get the syncs through the sytem intact that would be ideal, but experimentation has shown that is very difficult - it makes more sense to modify the MUTR unit to allow it to accept a series of artificially created sync pulses, or to mix that pulse train with the sync-less video.

[Klaas Robers]

Interesting: NBTV video without sync pulses. When assembling the NBTV CD No 3 we had that problem. In the NBTV Convention 2001 (whow, is it already that long ago) I recorded the output signals of the camera part of Eddie Greenhough's first disc camera. He built a combined camera monitor, using a disc with 1½ spiral, which gave good picture of his slowly rotating "Toby Jug". For him sync pulses were not needed as the single disc was always synchronised with itself.

So, back at home, I found myself with digitally recorded NBTV video without sync pulses. The speed of the disc was also "only about" so I needed to program something to synchronise a software flywheel in the PC program with the video. With a free running disc I could replay the video and I could analyse what it should be. After several attempts I managed to get the software synchronised with the video and I could place some form of sync, do DC resroration and gamma correction before the signal was inserted in the CD.

I know that in the repeated transatlantic experiments they synchronised the receiving disc onto a crystal source and received without any sync. If there are no hic-ups in the video this is the best way to go. Then when you see something, you know at least that the video is in sync.

[Lowtone]

Ah very interesting

Ok, I had trouble following your argument, so you have white pixels instead of a "missing pulse" in conjunction with blacker-than-black sync pulses?

That was it

In that case I would suggest that you would almost always miss the next line sync or 2 because the white has driven the average signal level above the sync slice - if I get time I will create a bit of software to demonstrate what I mean.

I would like to

Sync will only be lost if some part of the signal "looks" like a frame pulse but isn't - with a very noisy signal that can happen any time.

So it must he hard contrast picture. Maybe putting a white horizontal line above the sync can create a good contrast

I see. But, of course, the signal does not need to have HAD syncs in it to have syncs added - they can be added any time either in software or hardware.

I had though of that, having another device with only pulses playing and eveyrhing being mixed with the picture.

I have good news, i found the MUTR

[gary]

So it must he hard contrast picture. Maybe putting a white horizontal line above the sync can create a good contrast

It is certainly not immediately obvious to me how a "hard contrast" helps in sync detection as a "hard contrast" can occur within the video signal itself either intentionally (e.g. alternating black/white bars), or otherwise (noise).

Synchronisation is normally achieved by making the synchronisation signal completely unambiguous with respect to the video signal. Hence "blacker-than-black". Unfortunately, even in a "noise free" signal the sync signal may not be unambiguous due to AC coupling. This is why we incorporate DC restoration into our video decoding circuitry and software. There is, however, a limit to how effective this technique can be especially in a very noisy signal especially in an NBTV signal where the number of pixels available for the sync signal is necessarily low.

(Note that a blacker than black pulse not only gives sync but also gives a reference point by which picture contrast can be adjusted).

In a very noisy signal the concept of "flywheel synchronisation" becomes useful - this technique uses an average of syncs rather than each individual sync - this "filters out" false triggers. Again, however, there is a limit to how effective this can be.

Compare this to my preferred method, manual adjustment, either by software, or by a thumb on the side of a disk. Once set it will remain locked for a very long time, and is totally immune to noise, phase distortion, reflections, etc.

This works very well unless your monitor is built such as to insist on syncs being present in which case you need to add syncs or, send syncs in the signal itself and hope they work!

Elsewhere you imply that the standard NBTVA format includes sync. I am not sure if I am repeating myself here, so forgive me if I am, but, this is not quite true, the use of syncs has always been an "optional extra" to the best of my knowledge.