Question about scanning and "sync pulses"

[ct92404]

I just started reading a vintage book about televisions ("Basic Television - Second Edition" by Bernard Grob). It was suggested by Argus25 in another topic. I'm trying to do some research and learn more about how antique TVs work, so I can hopefully finish fixing my Olympic tv.

I just finished chapter 4, about scanning and synchronization. As I understand it now, a high frequency sawtooth signal is used to deflect the electron beam horizontally. The signal is fed to either deflection coils or to electrostatic deflectors. As the voltage rises in the sawtooth signal, the beam is deflected progressively farther across the screen. Another sawtooth signal, with a lower frequency, is used to deflect the beam vertically. Do I have that right so far? What I'm confused about though is when they start talking about a "synchronization pulse." What exactly does the synch pulse do? I'm reading over the chapter again to try to understand it, but I'm still not quite getting it. The books says that synchronization pulses aren't what does the scanning - the sawtooth signal does the scanning. Does the pulse start the sawtooth signal?

Another thing I was wondering about is that it looks like the horizontal scan lines slope downward as they move towards the right on the screen, because of the vertical deflection signal. But if that's the case, wouldn't the image on the screen look tilted? How do they make the image straight?

I'm sorry if these are really basic questions, but this is all still pretty new to me!

- Chris

[TonyDuell]

As I understand it the sync pulse does re-start the sawtooth to begin a new line or frame.

The scan lines do slope down slightly due to the ongoing vertical deflection signal). But its not noticeable for 2 reasons, firstly the slope is small (you have 525 or 625 lines, so <1mm 'error' between the ends) and secondly (at least in a traditional TV system) the camera tube was scanned in the same way so the scan lines on the photosensitive target slope in the same way.

[Radio Wrangler]

Let's build a TV.

We have a CRT and they're usually magnetically-deflected. There is a dual electromagnet assembly around the tube neck just where it starts to expand.

The amount of beam current needed to make a watchable picture would burn a hole in the phosphor coating of the screen very quickly if the beam wasn't being scanned across the whole screen. If one deflection system failed, the beam would be left painting a single line repetitively and this would also be damaging.

So we need the deflection systems to run continuously, even when there is no signal. Later TVs had circuits to detect if vertical scanning had stopped, and if so, to shut off the beam current. The horizontal scan driver was also used to generate the extra high voltage used to power the beam. If horizontal scanning failed, the beam would fade quickly without any extra circuitry.

So the idea of a sawtooth generator waiting to be triggered (like an oscilloscope timebase) would be damaging when there is no signal or tuning between stations. Think instead of the sawtooth oscillators being designed to naturally repetitively scan the spot at a slightly low frequency and to move it slightly too far. The speed of the scan across the face of the tube would be right, but the picture would be slightly too big. The sync pulse covers the time the deflection system has to move the spot back to the start for a new ramp part of the sawtooth. The sync pulse triggers an early end to the old line, allows time for the retrace, and then releases the sawtooth generator for the start of the next line If there is no sync for any reason, the old line ends a little later and a new one is started so continued scanning is the default.

Without sync pulses, the CRT may be being scanned, but the scanning will have the spot in a place not related to where the camera is currently scanning. The deliberate mis-adjustment of the scanning oscillator's speeds will result in a rolling picture. Add horizontal syncs and the picture stops rolling horizontally. Add vertical syncs and it stops rolling vertically.

Without syncs, the set still gets a stream of picture brightness information, but it doesn't know where to put it on the screen. Sync pulses are VERY important. recognition of this caused a changeover from positive to negative modulation in the UK when we changed from 405 to 625 lines. this made the syncs the biggest and therefore best protected part of the signal.

There are short sync pulses to set the timing of the horizontal scan. There are long sync pulses to set the timing of the vertical scan. A 'sync separator' circuit sends the appropriate ones to the right oscillators.

BUT, there is a subtle complication. In USA numbers, you are supposed to have 30 pictures drawn per second, but the vertical scan runs 60 times per second.

This is an anti-flicker measure. One vertical scan draws all the odd numbered lines, then the next one draws all the even numbered lines between them. This is achieved by circuitry timing the relationship between horizontal and vertical scans in the camera and it even does the half-line shift that's needed. This is called interlacing.

There is a lot of subtlety in the timing and choice of frequencies in TV systems. The tricks Dr Bruch invented to fix the phasing problems in NTSC colour are certainly the most complex.

David

[Argus25]

Basically, the horizontal scan oscillator and the vertical scan oscillator (in the absence of separated sync pulses from the received signal) are "free running"

These scan oscillators generate the basic waveforms (similar but not exactly sawtooth in shape) that drive the horizontal power output stage and H yoke coils, and the vertical power output stage and V yoke coils. The result is, as you now know, being a H and V sawtooth current in the yoke coils to deflect the electron beam. (In electrostatic sets it is a sawtooth voltage driving deflecting plates not sawtooth current). You will learn later why the actual drive voltages from these oscillators are not exactly sawtooth in shape, they are a shape known as trapezoidal, the effect of this shape results in actual sawtooth current occurring by the power output stages that drive the deflection yoke, which is what is needed for a perfect scan.

So when the picture signal (often called the composite video signal as it is sync & picture information) comes in and appears at the output of the video detector stage, the syncs are separated off, as H and V syncs by sync separator circuits.

These sync pulses are used to "Lock" the free running H & V oscillators so that the picture information you see on the screen starts in the correct place on each scanning line and the start of each field (two fields make a whole frame). In each case there is a rapid fly back.

(You are probably now aware that the image you see on the screen is the result of the picture information part of the composite video signal modulating the intensity of the picture tube's scanning beam. This video signal voltage is applied to either the grid or the cathode of the picture tube depending on the set design)

In the case of the vertical sync pulse, it is really a small voltage blip that is injected into the V oscillator circuit. Normally, when the V oscillator is set to run a tad slower than the sync rate, the blip triggers the oscillator just before it is about to fly back anyway. At that point the oscillator is more sensitive to the sync pulse. This is how the oscillator can be "locked" to the sync pulse.

If the received signal is weak and the V sync too low to lock the V scan oscillator, you can set the V hold control (Vertical oscillator frequency control) so that you see the wide horizontal bar between the frames rolling upward a little in this unsyncronized state. That means the oscillator is running, in the absence of sync, just a tad slower than the vertical sync frequency, which is about where the control should be set.

In very old sets, the H sync was done in the same way, but in this case line by line and the H sync pulse triggered fly back of the line (H) oscillator, just before it would have flown back anyway. This system though was pretty sensitive to noise pulses causing a tearing look of groups of lines.

However as time went by the H sync and oscillator circuits became more complicated and instead the H scan oscillator became an oscillator controlled by a DC voltage rather than synchronised by a discrete sync pulse. Then a circuit using a pair of diodes compared the phase of the oscillator output (often sampled from the H output circuit) and the incoming H sync, to create a DC error voltage to control the H oscillator frequency.

On these TV sets sometimes said to have AFC (automatic frequency control) when you turn the H hold control, once the picture has locked up, it then stays in lock over a large range of H hold control motion, and the picture appears to move left & right in the scanning raster. So the H Hold control then acts like a defacto H picture centering control, but you can only push that so far before it "drops out of lock" and the picture tilts over on its side to become an array of diagonal bars.
This type of circuit generally is known as a "Phased Lock Loop"

It will become much clearer, keep working through the book, and once you have gone through it once, go back to the beginning and do it again and it makes more sense the second time etc. But you are obviously picking it up very quickly.

[1100 man]

Hi Chris,
In the UK, horizontal oscillators controlled by a DC voltage are known as 'Flywheel sync'. They were developed to cope with noisy or missing H sync pulses due to interference or poor reception. Manufactures sometimes offered a standard model for good signal areas, or a 'fringe' model for weak signal areas. These usually had flywheel sync and cost more to purchase.
The idea is that when running 'locked', the 'flywheel' action keeps the thing locked even if H sync pulses are not present for a few lines. Just like a physical flywheel.
The big, big downside is that they don't respond to rapid changes in timing of the sync. This was never a problem with broadcast TV, but when VCR's came along, some TV's took violent exception.
Due to the mechanical nature of VCR's, the timing of the H sync is all over the place. Because the flywheel can't respond rapidly enough, the result is anything from bent verticals at the top to the whole picture shaking like a jelly!!
There were many mods to try to get round this to enable a TV to be used with a VCR with varying levels of success.
I've never liked the DVD format and have a very soft spot for VCR and open reel video recorders. So on vintage Tv's, I like to use material on VHS played on early machines.
Trying to modify flywheel circuits to cope is usually fraught with problems!! A nice simple blocking oscillator is so much better!!
All the best
Nick

[red16v]

Originally Posted by TonyDuell

As I understand it the sync pulse does re-start the sawtooth to begin a new line or frame.

The scan lines do slope down slightly due to the ongoing vertical deflection signal). But its not noticeable for 2 reasons, firstly the slope is small (you have 525 or 625 lines, so <1mm 'error' between the ends) and secondly (at least in a traditional TV system) the camera tube was scanned in the same way so the scan lines on the photosensitive target slope in the same way.

When we introduced CCD cameras into the studios in the latest 80's when CRT tellies were still very much in the majority, we did ponder on the fact that the cameras were now scanning 'perfectly' horizontally whilst the displays were still scanning with a small 'slope'. But as you rightly point out, the effect was so small as to be a non-problem.

[Argus25]

I could have mentioned that even if the very slight down tilting of the Horizontal lines was bigger, it wouldn't matter. The H scan lines can be made perfectly horizontal by rotating the yoke on the neck of the tube.

When you get used to repairing these sets you will release the clamp on the yoke and rotate the yoke to get rid of any picture rotation error.

Also many TV sets have a pair of magnets near the rear or attached to the rear of the yoke assembly to centre the scanning raster on the tube face. Some sets can run a DC current via the yoke coils to centre the picture that way, but that is less common and done in some computer monitors. In vintage sets with focus coils or focus magnets, the focus coil or magnet can often be moved to help centre the picture.

If you do release and reclamp a deflection yoke, never do the clamp up too tight , just enough to stop the yoke rotating easily.

So in magnetically deflected TV sets to rotate the image you rotate the yoke which rotates the scanning raster and therefore rotates the picture that is modulated onto that raster.

With Electrostatically deflected TV sets, many vintage American ones use the 7JP4 CRT, the deflection plates are part of the CRT, so the only way to rotate the picture is to rotate the whole CRT. This is also one of the reasons why "square faced" electrostatic CRT's for TV work were not made, it would have required super precision alignment of the gun/deflection plate structure with the tube face, and then rotation coils for fine adjustment.

Although square face CRT's are common for oscilloscopes and the image (scan line) is rotated by "rotation coils" that are added around the neck of the tube, because a square face tube can't be rotated.

[TonyDuell]

Originally Posted by Argus25

If you do release and reclamp a deflection yoke, never do the clamp up too tight , just enough to stop the yoke rotating easily.

The old joke (and anyone who doesn't realise this is a joke probably shouldn't be inside a TV set) is : Tighten the yoke clamp until the CRT implodes and then back off quarter of a turn.

More seriously, many manufacturers (at least for monitors) put tape round the CRT neck under the clamp and the yoke tends to stick to it. I have never had the courage to force the yoke round after slackening the clamp. I remove the clamp totally (it's a bit like one of those 'jubilee' hose clips) then use a screwdrver in the slots in the yoke that the clamp goes over to free things. Put the clamp back very loose, turn the yoke as required and tightn to little more than finger tight.

Years ago I made an inserting cross hatch generator to line up tube based broadcast cameras, this means the lines where horizontal not sloping.

[Argus25]

Originally Posted by TonyDuell

More seriously, many manufacturers (at least for monitors) put tape round the CRT neck under the clamp and the yoke tends to stick to it.

Yes, An excellent type of tape for this application is the very sticky and a little soft Scotch number 27 "Glass Cloth Electrical Tape" from 3M, Austin TX, USA, most electrical stores have this in AU, its called class B insulation.

It sticks to the neck well but its fabric like surface is rough enough that only a little yoke clamp tightening locks the yoke well. Conrac appear to have used this on the necks of their crts quite a bit, not just where the yoke clamps, but also on the neck and first part of the bulb where the yoke would otherwise make contact with the glass.

[Argus25]

Originally Posted by merlinmaxwell

Years ago I made an inserting cross hatch generator to line up tube based broadcast cameras, this means the lines where horizontal not sloping.

Yes, I think the intention for the alignment of cameras and monitors/TV's was that the horizontal scan line itself should be exactly horizontal with respect to a camera tube or TV tube's mask. Practically all pattern generators generate the horizontal lines of the cross hatch and segments of those for the vertical lines during a discrete number of H lines. So what is assumed to be the horizontal axis of the image, is the horizontal scan line itself and that resolves any microscopic tilt issues, but as noted they are only 64uS/20mS or about 0.32% of the picture height down-tilt across a scan line from left to right or just under 1mm tilt for a 12" high picture. And for NTSC frequencies just a tad over 1mm.

[cmjones01]

I have a feeling that more recent frame scanning chips don't actually produce a smooth sawtooth output, but a step per line, so each line really is horizontal. However, I can't find a reference at the moment.

Chris

[ct92404]

Ok, I've been reading the book some more and I think it's starting to click.
Let me see if I have this right so far:

The oscillators in the tv are normally scanning the screen by themselves, but the sync pulse from the video signal temporarily "overrides" them.

The composite video signal uses a negative voltage, and the amplitude determines the brightness. Lower amplitude means a brighter image, and a higher amplitude means a darker image. The amplitude varies the voltage on the control grid on the CRT, which is why a higher voltage reduces the intensity of the electron beam and darkens the picture. A higher voltage on the control grid repels electrons from the cathode. The picture signal by itself (or what the book is calling the "camera signal") is continuously varying the voltage on the control grid to create the light and dark parts of the picture.

The blanking pulse comes at the end of a scanned line, right before the sync pulse. It's a high amplitude, so it reduces the intensity of the beam all way down to make the horizontal line blacked out so that retrace lines don't overlap a line that was just scanned. Then the sync pulse makes the electron beam get deflected or "retraced" back to the left. Then the next line is scanned. I guess it's timed perfectly so that right when the beam is retraced back to the left, then the vertical deflection is about to move the electron beam down to scan another line.

Am I understanding it right?

[Argus25]

In the American and British early TV systems the modulation of the video signal on the transmitted carrier was different. In your system the sync pulses made the carrier go to a higher level and the picture signal reduced it.

In any case, once the composite video comes out of the video IF stages "detector" , maybe a 3 to 5 volts in amplitude, you have a video signal that needs to be amplified up and fed to the CRT to modulate the CRT's beam current.

Depending on the polarity of the detector diode, you can get either polarity video signal out of the video detector. Also each time it is amplified by a tube stage the polarity fips.

Typically in an American TV set the detector diode is placed with the anode as the output and the negative half of the video carrier is rectified or detected, that way you have about a 4V amplitude signal with negative going sync pulses.

Then that passes to a single video amplifier valve, which amplifies it up to about 40 or 50V with the syncs then going positive, then that is fed to the CRT cathode. But in other sets, I think like yours, last time I looked, there is a video driver stage, that flips the signal polarity again, but it doesn't matter because the designer can choose to use either polarity video detector or drive the CRT grid or cathode to make sure it works out the right way around.

With a CRT, if you take the grid negative with respect to the cathode it DARKENS the beam, this is the same as taking the cathode positive with respect to the grid.

It takes roughly around 30 to 40 volts peak to peak to modulate the CRT for full contrast, depending on the CRT.

[julie_m]

The best description I ever heard of the TV transmission process was that it's as though you take a white knitted jumper; screen-print a design right onto the knitting with deep-pigment inks that will soak right into the wool, colouring it all the way through; and then undo the stitches. Now you have a single, continuous length of wool that changes colour all along its length according to the printed pattern. This is your video signal. You can then pass the end of this wool through a keyhole; and another person, on the other side of the locked door, picks it up and begins knitting with it.

As long as your friend casts on the correct number of stitches in the first row and uses the same size needles so each stitch will use the same amount of wool as before, then the end result will be identical to the original, printed sweater.

Now, you already agreed in advance on the number of stitches per row, the diameter of the knitting needles and the amount of tension in the wool, which between them determine the amount of wool that should be used in each stitch. Synchronising pulses in a video signal are like the slight kinks in the wool where the original stitches used to have been, that the person behind the door can feel and keep fine-tuning the tension as they go, thereby ensuring the re-knitted sweater matches the original perfectly. Even if you cannot feel each individual stitch, as long as your assistant uses the right number of centimetres of wool in each row and finds some way of dealing with extra or missing stitches, you will get a reasonable enough approximation to the original picture. And in real life, the error is only goimg to be visible for 1/50 or 1/60 of a second anyway .....

The slight downward slope of the scanning lines is self-compensating if the camera is an analogue device, since the lines it sees will have the same slope as the lines drawn on the screen. You shoild be able to show this with just GCSE-level maths. But even if the source is digital and there is effectively no slope to the source lines, the difference is so tiny in practice (just 1 part in 287.5, in 625-line countries with 575 visible lines in the picture -- it takes all of 25 lines' worth of time to scan the beam back up again, at the bottom of each field -- or 1 part in 241.5 in 525-line countries with 483 visible lines) that it will never be noticeable in practice.

[Radio Wrangler]

When I was a kid I had a picture/pattern pullover that mum knitted with specially dyed wool that produced the pattern this way. The knitter had instructions on where certain colour changes should fit in - to stop cumulative error. "Mary Maxim" I think was the brand of the kit.

David

[julie_m]

Wow -- I never expected anybody actually to do that for real!

Even if it works perfectly, and that's a pretty big if, it's still a very slow conjuring trick .....

[Radio Wrangler]

It's amazing just what has been done at one time or another. It must have been in the early/middle sixties and the memory has lain dormant until I read your post. Even the maker's name came back! I have no idea how they coloured the wool in bands I remember being mystified by the balls of wool. The finished item lacked all the colour change knot tails on the inside and was a christmassy thing with reindeer. Rather blocky pixellated reindeer we'd say today.

For a job, mum used to sew in new threads to repair weaving errors and damage in newly woven fine worsted cloth. Skilled is a bit of an understatement. Even so, the knitting of that pullover entered the family language, a 'Mary Maxim job' being synonymous with 'York Minster job'.

David

[1955APREN]

I have realy enjoyed this thread , very interesting about the fly wheel sync and its workings. This thread must have put a spell on my recently obtained grundig
mtv. When I first collected it . It had a frame fault this was due to a bad cap in the scan coil circuit. Set then worked ok but when switched on yesterday after a few min's use loss of sync. Pic 1 showes the set working pic 2 showes fault. I have not had time to locate fault as yet but sent this post to show what lack of sync looks like.
regards Derrick

[ct92404]

Hi Derrick,
Yeah, I've enjoyed this discussion too and I've definitely learned a lot recently. I hope you can fix your tv soon. It seems like it should be a minor problem - the picture looks strong and clear otherwise.

I'm just still mind blown at how complex cathode ray TVs are. The people who figured how to do all this were geniuses!

- Chris