Missing sync pulse and the club standard --- Why?

Forum for discussion of narrow-bandwidth mechanical television

Moderators: Dave Moll, Andrew Davie, Steve Anderson

Missing sync pulse and the club standard --- Why?

Postby ac7zl » Tue Jun 19, 2007 7:03 am

I have been giving thought to various ideas of implementing adaptive, rock-solid sync control of the Nipkow wheel. I keep finding myself wondering how the club standard evolved and why the frame indicator is expressed in the manner in which it is.

My understanding is that the start of the frame (and of line 1) is indicated by a missing pulse. In my view, this makes frame sync more difficult and less deterministic than it could be. The sync schemes I've seen thus far involve using one-shots to project where the missing pulse would have been. Line one is thus sync'd to where we *think* it should be, not necessarily where it really is.

I think a better standard would have been to arrange for a missing pulse at the *last* line of the frame, not the first. The double-wide gap in pulses could be detected using one-shots, and used to "arm" circuitry intended to detect the real start of the frame. In this case, the start of the frame is indicated with a real event (the hard, rising edge of the sync pulse created by the wheels optical encoder), not the guessed-at timing of an event that never occurs.

I suppose the existing standard could be executed using the latter approach by simply moving the disk's optical detector to an angular position representing one line earlier. This would cause the image to be displayed as if we were sync'ing on the missing pulse, when in fact we were sync'ing on the first real pulse after the gap.

Any thoughts on this?

Pete
AC7ZL
User avatar
ac7zl
Mad Scientist
 
Posts: 62
Joined: Thu Feb 01, 2007 5:04 am
Location: Tucson, Arizona, U.S.A.

Re: Missing sync pulse and the club standard --- Why?

Postby Steve Anderson » Tue Jun 19, 2007 1:26 pm

ac7zl wrote:Any thoughts on this?

Pete
AC7ZL


...yes many!! I have been itching to start a thread on this for ages, but have been concentrating on getting my facts right in terms of my thoughts being plausable. You've just beaten me to it!

Why do we have sync pulses at all? We know that a line is 2.5mS in duration, all we need to know is the start of picture. My Panasonic TV doesn't use the line pulses at all, and I've proved this by feeding it a signal the the line syncs stripped off. It still needs the field sync though, i.e. start of field/frame and also identify if it's a 50Hz signal or 60Hz.

In SSTV there is a system called AVT, (still analogue) this has no syncs, just a 'header' to notify the start of frame and the standards used, no. of lines, colour encoding method etc.

Now, this all fine and dandy for a nice stable electronically generated signal, but a mechanical camera? It was one of the reasons I started to work on a TBC (TimeBase Corrector) for mechanically derived signals.

You know what? Without any syncs the whole subject of sync stripping goes away. Great!

Another discussion I have seen is does the 'missing sync' denote the start of line 1, or the end of line 32, the start of frame or the end? There have almost been knock-down fights over this one, and it's a subject I'm trying to avoid.

...more to come I'm sure!

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: Missing sync pulse and the club standard --- Why?

Postby gary » Tue Jun 19, 2007 2:49 pm

Steve Anderson wrote:Why do we have sync pulses at all? Steve A.


I'll have more to say on this thread later but my initial question would be (assuming sync is NOT an issue - although I think it is) how would one implement DC restoration?

EDIT: Of course I am referring to using the known (relative) level of a sync tip as a reference, and I am assuming AC coupling of an analogue signal would still be required in some instances.
gary
 

Re: Missing sync pulse and the club standard --- Why?

Postby Steve Anderson » Tue Jun 19, 2007 3:53 pm

gary wrote:How would one implement DC restoration?


This is another 'hot potato', and one I have devoted quite a few sleepless nights to. In another thread I posted an item I did for the newsletter which completely negated the need for DC restoration, it was flat from DC up to about 15kHz, however its implementation wasn't that simple.

A half-way house might be as shown in the sketch below. The top trace shows the current arrangement. The second trace shows the sync pulses omitted and the header information inserted into line 1 (I'll come back to that later).

The next trace shows the locally generated clamp pulses, in concept the same as 625/525 TV. These would be used to clamp the black level at some defined voltage. However, this does tend to make the signal uni-polar and line tilt would still be an issue.

The last trace shows the clamping 'porch' shifted up to half the amplitude of the total video signal (in red), this is certainly not an absolute solution, it does reduce the line tilt factor, but it's still there.

As for using line 1 for header data, we lose some 3% of our resolution in doing so, so I'm not that happy in doing that.

What all this hand-waving and neuron activity has done is drawn to me the conclusion that an audio circuit is not ideal, yes, it can work, but it's not ideal. Which is why myself and others have headed off in the direction of digital storage with dedicated A-D and D-A arrangements. Klaas's and Vics' colour encoding standard being an example.

...as I said...more to come!!

Steve A.

For those using RF communication for NBTV, omitting the syncs means you gain about 3db in S/N ratio.
Attachments
Clamping 1.gif
Clamping 1.gif (22.89 KiB) Viewed 14470 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: Missing sync pulse and the club standard --- Why?

Postby gary » Tue Jun 19, 2007 4:39 pm

ac7zl wrote:I have been giving thought to various ideas of implementing adaptive, rock-solid sync control of the Nipkow wheel. I keep finding myself wondering how the club standard evolved and why the frame indicator is expressed in the manner in which it is.

My understanding is that the start of the frame (and of line 1) is indicated by a missing pulse. In my view, this makes frame sync more difficult and less deterministic than it could be. The sync schemes I've seen thus far involve using one-shots to project where the missing pulse would have been. Line one is thus sync'd to where we *think* it should be, not necessarily where it really is.

I think a better standard would have been to arrange for a missing pulse at the *last* line of the frame, not the first. The double-wide gap in pulses could be detected using one-shots, and used to "arm" circuitry intended to detect the real start of the frame. In this case, the start of the frame is indicated with a real event (the hard, rising edge of the sync pulse created by the wheels optical encoder), not the guessed-at timing of an event that never occurs.

I suppose the existing standard could be executed using the latter approach by simply moving the disk's optical detector to an angular position representing one line earlier. This would cause the image to be displayed as if we were sync'ing on the missing pulse, when in fact we were sync'ing on the first real pulse after the gap.

Any thoughts on this?

Pete
AC7ZL


In actual fact, by club convention, a sync pulse is added at the *end* of each line except for the last one where it is omitted. So the frame sync (missing line sync) *is* actually at the end of the frame. Of course you only have to advance processing the signal by one sync pulse width and you have syncs at the beginning of the line and the frame pulse at the beginning of the frame.

The reason for using a missing pulse for frame sync is explained in:

"Start or Finish" by Doug Pitt: NBTV Newsletter Vol. 29 No. 4
gary
 

Re: Missing sync pulse and the club standard --- Why?

Postby Steve Anderson » Tue Jun 19, 2007 5:12 pm

gary wrote:Of course you only have to advance processing the signal by one sync pulse width and you have syncs at the beginning of the line and the frame pulse at the beginning of the frame.

The reason for using a missing pulse for frame sync is explained in:

"Start or Finish" by Doug Pitt: NBTV Newsletter Vol. 29 No. 4


Here we're getting into the realm of symantics, the mechanics or electronics doesn't care what we call it, it just gets on and does its job. So should we.

Steve A.

Heck, sorry about the above, it's just the project manager coming out in me.

Did I hear someone mutter "Barstard"?
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 AncientBrit » Tue Jun 19, 2007 5:29 pm

If we adopted crystal lock some form of "sync" detection is still necessary on the received signal stream to determine flyback.

If the stream is a digital one then a unique sync word may be embedded to perform that purpose.

If the signal is analogue then we need to flag the presence of flyback somehow. Amplitude discrimination seems the obvious way and so a super white or super black sync pulse is one solution.

As regards DCR, as long as there is a known reference level in the signal and timed "clamping" pulses available the signal may either be at black level or at the bottom of syncs at that point.

In fact for component colour difference signals which are non-composite, clamping does occur at black level on the signals.

GL
AncientBrit
Green padded cells are quite homely.
 
Posts: 858
Joined: Mon Mar 26, 2007 10:15 pm
Location: Billericay, UK

Postby AncientBrit » Tue Jun 19, 2007 5:34 pm

"Pedant watch"

Strictly speaking I should have said "zero value colour difference", not "black level", for the colour signals since they are bipolar.

You know what I mean......

GL
AncientBrit
Green padded cells are quite homely.
 
Posts: 858
Joined: Mon Mar 26, 2007 10:15 pm
Location: Billericay, UK

Re: Missing sync pulse and the club standard --- Why?

Postby Steve Anderson » Tue Jun 19, 2007 5:51 pm

gary wrote:In actual fact, by club convention, a sync pulse is added at the *end* of each line except for the last one where it is omitted. So the frame sync (missing line sync) *is* actually at the end of the frame.


So is the following correct?

Steve A.
Attachments
Clamping 2.gif
Clamping 2.gif (13.44 KiB) Viewed 14449 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: Missing sync pulse and the club standard --- Why?

Postby gary » Tue Jun 19, 2007 6:10 pm

Steve Anderson wrote:
gary wrote:Of course you only have to advance processing the signal by one sync pulse width and you have syncs at the beginning of the line and the frame pulse at the beginning of the frame.

The reason for using a missing pulse for frame sync is explained in:

"Start or Finish" by Doug Pitt: NBTV Newsletter Vol. 29 No. 4


Here we're getting into the realm of symantics, the mechanics or electronics doesn't care what we call it, it just gets on and does its job. So should we.

Steve A.

Heck, sorry about the above, it's just the project manager coming out in me.

Did I hear someone mutter "Barstard"?


That's OK Steve I used to be a project manager myself and I understand the psychological damage it can do to you if you're not careful...

As a result, later, as a senior executive, I knew I had to keep an close eye on them...

An engineer and his project manager left their cabin in the jungle to go out on safari, hunting lion. A while later both came screaming back with a lion nipping at their heels. The project manager ran up to the cabin door, opened it, let the engineer and the lion run inside, slammed the door and yelled out to the engineer: "There! I've caught the lion... now YOU skin him!"

(of course that joke works for marketing types as well)
gary
 

Re: Missing sync pulse and the club standard --- Why?

Postby gary » Tue Jun 19, 2007 6:47 pm

Steve Anderson wrote:
gary wrote:In actual fact, by club convention, a sync pulse is added at the *end* of each line except for the last one where it is omitted. So the frame sync (missing line sync) *is* actually at the end of the frame.


So is the following correct?

Steve A.


Looks good to me. Note that if we were displaying that video "verbatim" including the syncs that the black bar caused by the syncs would be at the top of the picture and the missing sync at the far left of that (image 1).

Just to be "semantic" ;-) if we were to advance the processing by a sync width the bar would be at the bottom and the missing sync at the right of it (image 2).

EDIT: Sorry it took me quite a few go's to get the order of these images correct.
Attachments
sync2.jpg
Image 2
sync2.jpg (9.88 KiB) Viewed 14434 times
sync1.jpg
Image 1
sync1.jpg (9.49 KiB) Viewed 14434 times
Last edited by gary on Tue Jun 19, 2007 7:12 pm, edited 1 time in total.
gary
 

Postby Steve Anderson » Tue Jun 19, 2007 7:00 pm

AncientBrit wrote:If we adopted crystal lock some form of "sync" detection is still necessary on the received signal stream to determine flyback.

GL


Well this was my whole point. For the moment let's assume that both the sender and receiver have access to a crystal contolled reference, better still is a GPS disicplined (GPSDO) oscillator (better than 2 x 10<sup>-11</sup> accuracy within 1 second).

The worst crystals or crystal-controlled oscillators that I can find are 100 parts per million (100ppm), or 0.01% error in frequency, within a NBTV frame that amounts to 250nS (nano-seconds), double that if each is at opposite extremes of its tolerence. Or around 0.3 of a pixel per frame at 48 pixels per line and 32 lines.

So, once we know when a frame starts, we know with a high degree of accuracy when the next and subsequent lines start. This is how the SSTV AVT system works. There is no need for sync pulses, just a reference point, a 'header', that says "This is where you start, off you go, but keep your tempo correct". Nor do we need a sync pulse.

A digital stream might consist of the arrangement shown in Clamping 3.gif, flyback with mechanical TV displays is not an issue, so we don't need a front or back porch as used in conventional 625/525 TV.

Where we do need a reference is in an analogue transmission system that doesn't have a response down to DC, so perhaps the last waveform above could be used.

These are all just musings, conceptions, random thoughts, not proposals, and I'm sure that others will come up with better ideas than these. I hope it starts some discussion on the matter.

Steve A.
Attachments
Clamping 3.gif
Clamping 3.gif (7.91 KiB) Viewed 14438 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

Postby AncientBrit » Tue Jun 19, 2007 10:39 pm

I record NBTV digital files in 64x32 format to disc and ensure that each frame starts at 0 pixel 0 line and is exactly of 2048 bytes duration.

So on playback no form of sync detection is necessary, the frames are taken as 2k chunks and painted straight to screen.

Neither is there any need to write header info.
The assumption is that the file rec/playback process from disc is error free.
(Indeed if it were not I'd have file problems on the PC)

My hardware copes with some mechanical variation in speed by using a line gated pixel clock so that the start of each NBTV line always aligns with a 64byte boundary.

Any variation in speed manifests itself as a "breathing" in displayed line content. The pixel clock is set such that there are never more than 64 bytes written for each line.

This process could to be adapted to 44.1kHz sampling so that each frame always occupied a known number of bytes and aligned exactly with a pre-defined boundary.

GL
AncientBrit
Green padded cells are quite homely.
 
Posts: 858
Joined: Mon Mar 26, 2007 10:15 pm
Location: Billericay, UK

Postby gary » Tue Jun 19, 2007 11:05 pm

Indeed 44.1 kHz gives exactly 3528 samples per frame so frame alignment is trivial, unfortunately each line has 110.25 samples so either a saw tooth top and bottom border is exhibited (with line sync) or else the picture is skewed upward from right to left (without line sync).

32 kHz or 48kHz are perfect for NBTV - alas not CD compatible...
gary
 

Postby Steve Anderson » Tue Jun 19, 2007 11:29 pm

AncientBrit wrote:I record NBTV digital files in 64x32 format to disc and ensure that each frame starts at 0 pixel 0 line and is exactly of 2048 bytes duration.

So on playback no form of sync detection is necessary, the frames are taken as 2k chunks and painted straight to screen.

Neither is there any need to write header info.GL


Yep. we're on the same wavelength here. You've done it at a 64 byte per line system (nothing wrong with that), my earlier suggestion was based on 48 pixels per line, yet the .wav files I'm using are 120 samples/line (48kHz). This avoids the rather awkward 110.25 samples/line of the audio CD format.

What this means is that at the convention you'll need to turn up with a laptop rather than a small audio CD player!

At the moment I'm still incorporating the sync pulse, but within house I'll soon be dumping it, there is simply no need for it. However, one still needs some form of identification of when and where the frame starts. This has been the cause of my sleepless nights, I don't want to give away 3% of the resolution to framing by using a complete line (#1). So I'm very open to suggestions.

Sure, the NBTV header info could be considered superfluous, and incorporated as part of the header information of the data file (just as avi/wav/riff files do in the 'fact' or 'junk' sub-chunk sectors).

Steve A.

I get the feeling that this thread might go on for some time....
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Next

Return to Mechanical NBTV

Who is online

Users browsing this forum: No registered users and 27 guests

cron