Hello all you crazy people!

Anything not specifically related to NBTV, but at least of some technical nature that might be of interest to NBTV members. Items for sale and links to retailers do not belong here.

Moderators: Dave Moll, Steve Anderson

split image

Postby holtzman » Wed Dec 14, 2011 7:39 am

Welcome to the club :)
As I understand from your post, the picture gets split suddenly, after you have it already synchronized for a while. This may be caused not by your TV, but by the signal you are feeding into it. If you use a CD player, or MP3 player, there may be skips in playback. Even if you don't feel them while playing music, they may be present and can cause these sudden out-of-sync issues. Another reason may be that you are playing a short audio file in a loop. Video2NBTV converts videos in such a way that the beginning of audio file is the beginning of the first NBTV frame, and the end of audio is NOT the end of a frame. So probably your file ends somewhere in the middle of a frame, then playback jumps to the start, and thus the sync is gone.
holtzman
Just nod and pretend you understand me
 
Posts: 215
Joined: Wed Apr 22, 2009 7:54 pm
Location: Jerusalem, Israel

Re: split image

Postby gary » Wed Dec 14, 2011 8:12 am

holtzman wrote:Welcome to the club :)
As I understand from your post, the picture gets split suddenly, after you have it already synchronized for a while. This may be caused not by your TV, but by the signal you are feeding into it. If you use a CD player, or MP3 player, there may be skips in playback. Even if you don't feel them while playing music, they may be present and can cause these sudden out-of-sync issues. Another reason may be that you are playing a short audio file in a loop. Video2NBTV converts videos in such a way that the beginning of audio file is the beginning of the first NBTV frame, and the end of audio is NOT the end of a frame. So probably your file ends somewhere in the middle of a frame, then playback jumps to the start, and thus the sync is gone.


No, that's not right, Video2NBTV *NEVER* produces a partial frame, if the audio length and the number of frames do not match the audio is either truncated or extended with silence.

It's impossible to create a partial frame as the conversion is done on a frame by frame basis, a frames worth of the *converted* audio is then muxed in to suit. This also solves the problem whereby the use of VBR compression in the original causes lip sync problems in other conversion schemes.

A partial frame CAN be caused by later externally editing the wave file in some way.

If Metallica Man X can post the wave file this can be easily checked.

Edit: Of course, if you can demonstrate that this does, in fact, happen then you should give me full details as it represents a bug that needs to be fixed.
Last edited by gary on Wed Dec 14, 2011 8:42 am, edited 1 time in total.
gary
 

Postby gary » Wed Dec 14, 2011 8:40 am

Steve Anderson wrote:From my perspective I've always viewed the 'missing-sync' arrangement somewhat lacking
Steve A.


But Steve, surely the problem is not detecting the frame sync but what to do with it when you have?

With, say, a CRT system it's easy to simply set the trace back to the beginning of frame. With a mechanical device it's not so easy as you must retard or advance the disk rotation to bring it into frame lock whilst still retaining line lock (or vice versa).

It's my understanding that automatic frame lock is still pretty much the holy grail of mechanical monitors, the 4046 PLL cct does *seem* to achieve this (sometimes) but, to my knowledge, it's still a bit of a mystery as to why it does. There was a somewhat plausible write up in the Newsletter a few years back which proffered the theory that it achieved this due to the fact that it could not achieve line lock, *because* of the missing pulse, until achieving (a form of) frame lock. The fact that the missing pulse is detected *after* the event appears to account for the "out by one line" phenomenon.

This kind of falls in to line with Panrock's advice that an extended missing pulse (or, I suppose, any form of frame sync identifier) results in quicker locking, as, until frame lock is achieved, there is only one, short, variation in the sync stream being acted upon. EDIT: without extending it that is.

If anyone has a better understanding of this principle it would make great, and enlightening, reading.
gary
 

Postby Metallica Man X » Wed Dec 14, 2011 8:56 am

Interesting stuff.

I can't really post a wav file since I usually just run one of my video files through the program for my amusement lol...I could post an entire episode of Dexter's Laboratory via the 2nbtv program if you really want to check it lol.


I just had an idea! Dunno if anyone else has thought of this but...

What if you made a disk, have the regular spiral holes, and the set of sync holes for the regular disk speed control, but then also have another hole in another spot (maybe closer in towards the center?) that's aligned with what would be the start of each frame. From there you could add another pulse specific to what would be the start of each frame. Combined with an additional sensor, perhaps this could help to make sure the picture always stays centered, even when there's a skip in the audio?


Just a random thought lol.
User avatar
Metallica Man X
Master Craftsman
 
Posts: 54
Joined: Sun Dec 04, 2011 10:40 am
Location: Middle of nowhere Wisconsin, USA

Postby gary » Wed Dec 14, 2011 9:13 am

Metallica Man X wrote:Interesting stuff.

I can't really post a wav file since I usually just run one of my video files through the program for my amusement lol.



Aaaah! That could explain it then. Depending on the speed of your computer, there is NO guarantee that the conversion process can be achieved in real time, in other words, you could be having occasional skips in the audio. Try recording the same material as a wave file and playing that out. If that always plays fine then that would be the problem.


Metallica Man X wrote:I just had an idea! Dunno if anyone else has thought of this but...

What if you made a disk, have the regular spiral holes, and the set of sync holes for the regular disk speed control, but then also have another hole in another spot (maybe closer in towards the center?) that's aligned with what would be the start of each frame. From there you could add another pulse specific to what would be the start of each frame. Combined with an additional sensor, perhaps this could help to make sure the picture always stays centered, even when there's a skip in the audio?


Just a random thought lol.


Well that is already done really with the missing pulse, it just needs a bit more electronics to extract it (if you wish to do it). The real problem is aligning it with the missing pulse in the audio stream. Once a "skip" has occurred the location of the missing pulse in the audio stream has changed in time compared to that of the rotating disk. Of course if your sync processing electronics provides automatic frame lock it will come into sync by itself anyway.
gary
 

Postby Lowtone » Thu Dec 15, 2011 1:00 am

Maybe there is 2 seconds silence beetween the tracks of your CD and that messed up the synch thing :o

DrZarkov wrote:For NBTV "Ceefax" (actually it would be a kind of "Videodat" which was used by some television stations during the late 1980s/early 1990s for transmitting computer software in "Basicode 2" we could use a second line, which gives us 32 Bits (or 4 Byte) per frame for information. Per second we could transmit about 54 letters, which is not even enough for subtitles, but would be enough for a callsign and some information about the transmitter. It could indeed be an interesting item for shortwave broadcasts.

Ah yes callsign lie RDS for radio ? that could be nice :P
r a d i o P T T v i s i o n
User avatar
Lowtone
Just nod and pretend you understand me
 
Posts: 320
Joined: Sat Jan 16, 2010 9:45 am
Location: France

Postby Benton82 » Thu Dec 15, 2011 1:36 am

Hello there guys! I just logged in here so it's awesome to meet you! My name is Benton and I'm a nice guys! Cheers!
Benton82
 

Postby Steve Anderson » Thu Dec 15, 2011 2:54 pm

gary wrote:...But Steve, surely the problem is not detecting the frame sync but what to do with it when you have? With, say, a CRT system it's easy to simply set the trace back to the beginning of frame.


The most frequent manner of detecting the absence of a sync pulse for frame reference I have seen is using a re-triggerable monostable set to run for longer than one line, but shorter than two, say 3.5ms. (Club Handbook and many Newsletter items). In the absence of s sync pulse its output(s) toggle indicating a new frame. However we're already part-way through the first line.

With a CRT whose frame flyback is triggered by this part of line 1 would be in the bottom-left corner (a phantom line 33) and reappear at the upper right of the picture.

To address this I knocked-up a frame-sync detector/estimator/guesser, ("Eureka! I've found it!" Newsletter Vol. 35 No. 4 Page 4). And if I may say so, it works amazingly well. It also includes DC restoration and a few other things, but the impetus behind it was this damn frame-sync.

For a CRT display as I use it is the Holy Grail.

gary wrote:...With a mechanical device it's not so easy as you must retard or advance the disk rotation to bring it into frame lock whilst still retaining line lock (or vice versa).


Here I lean back in my chair, raise my hands into the air and say, "Hands up, I have no idea."

But with the Eureka MkI surely having an accurately placed frame-sync would assist?

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

Postby M3DVQ » Fri Dec 16, 2011 1:42 am

If you take your "has there been a pulse within one frame period" signal, and the incoming frame sync and gate them together appropriately you can obtain a "line 2 sync". This gets you an accurate place to re-sync your frame clock. You just have to accept that line 1 follows on from line 32 with no sync but your sync should be happy to free-wheel for a single line.

Or is that what you were getting at already?
M3DVQ
Just nod and pretend you understand me
 
Posts: 338
Joined: Sat Feb 10, 2007 10:52 am
Location: Lincolnshire

Postby M3DVQ » Fri Dec 16, 2011 1:53 am

It's a long time since I had a mechanical monitor built up, and when I did I just had the club motor circuit and a big speed control pot so what follows is speculation but...

Presumably once the line lock circuit is stable a frame lock will be fighting it to try to advance or retard the disk. You have two feedback loops trying to compensate for each other unlike in an electronic monitor where the two syncs can be separate clocks that can be synchronised independently.

Would it not be a sensible idea if the absence of a frame lock disabled the line sync in such a way that until the disk got to the point of drifting back and forth between the start of 32 and end of line 1 the frame sync (i.e. course adjustment) was in control. Then once we are within 1 line period of the target the line sync feedback is introduced to fine tune the speed (fine adjustment)

I've spent five minutes on this theory so it may have a gaping flaw, and the design of the circuit is an exercise for the reader :wink:

(one obvious problem is that there is no line 32/1 pulse to sync against, so we should just sync on line 2 or whatever)
M3DVQ
Just nod and pretend you understand me
 
Posts: 338
Joined: Sat Feb 10, 2007 10:52 am
Location: Lincolnshire

Postby Steve Anderson » Fri Dec 16, 2011 1:20 pm

M3DVQ wrote:...Or is that what you were getting at already?


Not quite, here's the pertinent paragraphs from the article, there's more in the full Newsletter version...

Within a standard NBTV frame there are 31 line sync pulses, the missing one defining the end of one frame and the start of the next. Between these 31 pulses there are 30 lines of nominally 2.5ms each = 75ms total. A counter inside the PIC is clocked at 500kHz from the start of the first pulse after the frame period, where it is set to zero, then stopped at the start of the 31st pulse. Nominally it should read 500,000 x 0.075 = 37,500.

This 500kHz comes from the internal RC clock oscillator of the PIC which runs at 4MHz; this is divided by four for the processor proper then further divided by two for this 16-bit counter. It's not as accurate as a crystal but here it doesn't need to be. It is specified at ±1% after adjustment during programming. It wouldn't matter if it were 10% off, all internal timing is referenced in some manner to this, as long as it is reasonably stable, which it is, all is fine.

This 37,500 (nominal) is divided in software by 30 to yield 1,250, this is the average duration of those 30 lines in 2µs steps (500kHz). 1250 x 2µs = 2,500µs, i.e. 2.5ms. The same counter is re-loaded with that figure and after this nominal 2.5ms produces a 55µs pulse right where the missing pulse would have been.

It should be intuitive that other line-rates simply alter the figures, they're all proportional. Because it averages across 30 lines a large percentage of any erratic syncs should be ironed out. But it does fall flat on its face if pulses are missing, though this is also true of most other frame detection systems. So a good clean signal is needed, even if off-frequency and perhaps a little erratic.


But to continue with what I think was your train of thought, it is possible to use the sync between line 1 and line 2. This is the first pulse of the frame. So detect the missing pulse in the conventional 'missing-the-bus' monostable manner, set a bistable to enable sync pulses (gating), once that first pulse comes along reset the bistable. No more sync pulses will pass until the start of the next frame.

This will generate a pulse of 12.5Hz related to the frame rate. But it's a line late. Now, my thinking is that on a mechanical monitor you shift the position of the frame detect hole in the disc/drum/etc. to between line one and two.

Mechanical guys...make sense? Feasible?

Unfortunately this can't directly be used with a CRT without further processing.

Or have I gone off the rails too?

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

Postby M3DVQ » Sat Dec 17, 2011 4:12 am

Steve Anderson wrote:This will generate a pulse of 12.5Hz related to the frame rate. But it's a line late. Now, my thinking is that on a mechanical monitor you shift the position of the frame detect hole in the disc/drum/etc. to between line one and two.

Mechanical guys...make sense? Feasible?

Yep this was what I was suggesting :)
Once you have a disc locked to the frame rate you can rotate the position of the holes to the appropriate phase.

Steve Anderson wrote:Unfortunately this can't directly be used with a CRT without further processing.

You can if you generate a staircase using a 5 bit binary counter instead of an analogue sawtooth. Makes it very easy to rotate the whole sawtooth, I can't think how you'd do that in the analogue world :)
M3DVQ
Just nod and pretend you understand me
 
Posts: 338
Joined: Sat Feb 10, 2007 10:52 am
Location: Lincolnshire

Previous

Return to Off Topic

Who is online

Users browsing this forum: No registered users and 14 guests

cron