Cutting out sync pulses.

Forum for discussion of narrow-bandwidth mechanical television

Moderators: Dave Moll, Andrew Davie, Steve Anderson

Cutting out sync pulses.

Postby Viewmaster » Thu Oct 13, 2011 7:14 pm

As per the recent discussion on Steve's idea to remove sync pulses for better locking I have been mulling over this again. I am not up on whether clitches may occur which will upset the apple cart but it's worth trying just for a few quid........

I have ordered the parts for this idea and will let you know if it does remove sync pulses OK.

Using a 4 way AND gate I think ( I hope ) that it should remove 3 pulses together with the existing missing pulse from the club's sync seperator.

With the 4th gate i/p strapped to + (via a 1k R) it should remove just 2 pulses and so on.......... I hope......... which we all live in :-)
Attachments
SIPO.JPG
Sync pulse remover
SIPO.JPG (17.64 KiB) Viewed 14236 times
User avatar
Viewmaster
Frankenstein was my uncle.
 
Posts: 1306
Joined: Fri Apr 06, 2007 4:50 am
Location: UK Midlands

Postby Viewmaster » Fri Oct 14, 2011 6:35 pm

Have just realized that the clock input on the previous little circuit has the missing sync pulse and so that shift register will not work correctly.

The misssing pulse has to be inserted into the clock pulse to give correct constant 400 hz clocking of the shift register.

Fortunately, in the NBTV newsletter Vol33. No1. page5, Peter Smith has a clever missing pulse insertion circuit using a single IC 4538, so this will have to be added in order to obtain correct clocking of the shift register......
.....more work! :-)
User avatar
Viewmaster
Frankenstein was my uncle.
 
Posts: 1306
Joined: Fri Apr 06, 2007 4:50 am
Location: UK Midlands

Postby Steve Anderson » Fri Oct 14, 2011 11:21 pm

May I suggest a somewhat simpler method? A dual monostable and a gate will do this. It could be a 4538, 221 or a dual 555 (the 556). The first of the pair detects the missing pulse as per usual, the second upon missing pulse detection simply gates out as many sync pulses as you desire. There are constraints on timing, but none that critical at all.

If there's interest I'll do a fag-packet sketch tomorrow...it's the end of my day here...

But what I'm still puzzled about is why should this be required? It's confounding me!...and my rubber chicken who's remained unusually silent...

Steve A.

P.S. Cedric tells me that the 221 is a non-starter, it's not retriggerable...the 4528 though is also suitable...
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 Viewmaster » Fri Oct 14, 2011 11:46 pm

Steve Anderson wrote:May I suggest a somewhat simpler method? A dual monostable and a gate will do this. It could be a 4538, 221 or a dual 555 (the 556). The first of the pair detects the missing pulse as per usual, the second upon missing pulse detection simply gates out as many sync pulses as you desire. There are constraints on timing, but none that critical at all.

If there's interest I'll do a fag-packet sketch tomorrow...it's the end of my day here...


Steve, I don't wish to encourage you to smoke just to get more fag packets to draw on. :-)

I'm sure your idea is the best and simplest one but I have only got into this out of academic interest really, not out of necessity yet.

As I have ordered the serial reg etc I shall build it....just to see if I can ever, ever design a circuiit that really works.
You see Steve, I'm a non smoker so cannot do these clever fag packet designs. ;-)

If others would like to see your design though, maybe they could say?

It is 120 line Steve who seems to require the gating out of more pulses and you have already done this for him in your designs shown on the other thread.

Steve Anderson wrote:But what I'm still puzzled about is why should this be required? It's confounding me!...and my rubber chicken who's remained unusually silent...
Steve A.


I don't think it is actually required as I understand it. It's just a more reliable way of framing.
As Steve said on the other thread," but the preference for things to 'come to rest' at the correct spot is clear."

So you too can rest your rubber chicken for now. :-)
User avatar
Viewmaster
Frankenstein was my uncle.
 
Posts: 1306
Joined: Fri Apr 06, 2007 4:50 am
Location: UK Midlands

Postby gary » Sat Oct 15, 2011 12:14 am

To be honest, I seem to have lost track of what the aim of this exercise is as there is mention of "gates out as many sync pulses as you desire", some clarification of the purpose of of doing that would be greatly appreciated.
gary
 

Postby Steve Anderson » Sat Oct 15, 2011 12:46 am

gary wrote:To be honest, I seem to have lost track of what the aim of this exercise...


You and I both...it can be done, for sure, what I don't quite grasp is why, neither does the other Steve who experimentally found that such an arrangement provided for more positive and correct locking of his colour monitor to the camera.

I mentioned elsewhere about time-constants, inertia and a few other matters that are all part of this puzzle...and seem to have been for a long time. Some seem to strike it lucky and all works first time, others go bald as part of the same exercise...

All we are dealing with here is a Phase-Locked Loop which is as reliable as any other electronic component/system, this is in regard to the ever-popular Club motor-control circuit. Where the variables are is in the mechanical components, mass, moment, acceleration, deceleration, friction, air-resistance/drag...these should have electronic analogues that can be designed out...or perhaps I should say, designed in...

The problem appears to be in identifying and quantifying them, then deciding how and what to do with this data...

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 gary » Sat Oct 15, 2011 9:44 am

Well, no doubt, as in most control systems, the exact relationships are quite complicated, however in this case I can't quite understand why circuitry for gating out multiple sync pulses is required.

Firstly, are we talking about improving speed locking (line locking) or angular locking (frame locking)? I can't see how removing some extra pulses could improve the former, and whilst the prevailing explanation of how the standard missing pulse results in frame locking minus one line is plausible, I can't see how that is improved by removing extra pulses as that, assuming the same mechanism as for the single missing pulse, would give frame locking minus - n. I suppose any *deterministic* error could subsequently be adjusted out, but I am not sure that is what in fact is being attempted.

Secondly, why are pulses being extracted at all? Surely that, as in a standard missing synch pulse, is done at the source for a monitor, and compared against those generated by a typical encoder attached to the shaft of the monitor.

I, as previously stated, have obviously missed something along the way to this point.
gary
 

Postby Panrock » Sat Oct 15, 2011 7:09 pm

Hi Gary,
gary wrote:Well, no doubt, as in most control systems, the exact relationships are quite complicated, however in this case I can't quite understand why circuitry for gating out multiple sync pulses is required.


Nor can we! But experience has shown this works better in practice if we expect the club 4046 circuit to not only line lock a Nipkow disk, but provide angular lock (your term) too - what I would call frame phasing..

gary wrote: Firstly, are we talking about improving speed locking (line locking) or angular locking (frame locking)? I can't see how removing some extra pulses could improve the former, and whilst the prevailing explanation of how the standard missing pulse results in frame locking minus one line is plausible, I can't see how that is improved by removing extra pulses as that, assuming the same mechanism as for the single missing pulse, would give frame locking minus - n. I suppose any *deterministic* error could subsequently be adjusted out, but I am not sure that is what in fact is being attempted.


Quite simply, for greater ease of operation. The idea is: now one can turn the thing on and it will run up and lock to line and correct frame phase all at once. There will (or should!) no longer be any need to juggle the speed control manually - that is once it's finally up to speed and line locked - to achieve frame phasing. Even at 30-lines this can be a pain. At 120 lines it would be quite a challenge to get this accurate to the nearest line.

gary wrote:Secondly, why are pulses being extracted at all? Surely that, as in a standard missing synch pulse, is done at the source for a monitor, and compared against those generated by a typical encoder attached to the shaft of the monitor.


If this comparison is to be done automatically using the club circuit, the fact is that having a chunk of pulses missing, rather than just one, helps the process along.

Perhaps this is because, when there is a chunk of the little pulses missing, the comparator sees both the short line pulses and the longer 'frame' pulses that have now in effect been created. It might be that cutting out a full half of the short line pulses (to produce a virtual 'squarewave' of long pulses) would provide the most powerful frame phasing effect, but I found a mere 10% cut was sufficient to do the job.

I hope this makes it clearer, but maybe the more the words - the more obscure... :shock:

Steve O
Panrock
Green padded cells are quite homely.
 
Posts: 870
Joined: Mon Feb 05, 2007 8:25 am
Location: Sedgeberrow, England

Postby kareno » Tue Oct 18, 2011 10:38 pm

I have also struggled to understand how the club motor controller can achieve frame lock using the missing sync pulse method. After all, the phase error caused by one too many sync pulses in one part of the cycle should be balanced and cancelled by one too many opto fork pulses in another part of the cycle. Over a single rotation the number of pulses of both types is still balanced, even without frame lock. My conclusion is that there is an asymmetry at work that does permit a contribution to net phase error when the system is out of frame lock.

For instance, most phase comparators (including the club motor controller) use flip-flops. I can imagine a situation where, say, an extra sync pulse sets a flip-fop resulting in an unusually prolonged 'on' state and a big contribution to phase error. But the circumstances may be such that an extra opto pulse does nothing more than clear an already cleared flip-flop. In those circumstances there IS an imbalance and a net phase error resulting from frame un-lock. If this phase error is sufficient to make the picture roll then frame lock will eventually be achieved.

Maybe Steve's 'four missing pulse system' uses the same principle but amplifies the effect by causing a much larger 'frame phase error' signal thereby making the system more responsive to frame phase error?
kareno
 

Postby Steve Anderson » Wed Oct 19, 2011 12:04 am

For the time being let's ignore the effects of the loop-filter and look at this from a logic point of view. Using (as the Club Motor Control circuit does) the 4046 Type II phase comparator the waveform/wave-shape/duty-cycle should be of no issue as long as it's at logic levels. If there's a missing pulse in the reference and the feedback then all should lock fine.

But herein lies the rub. The loop filter. The mechanical components of a NBTV monitor form part of this PLL loop-filter. They have time-constants which may(?) be asymmetrical, and/or vary widely between different specimens of basically the same species...

I'm tempted to have a go at this tomorrow, just from an electronic point of view and see how reliably a PLL locks up to a 400Hz waveform with one in 32 pulses missing. The reference (our input video) will have one in 32 missing, so therefore so should the feedback, this will require a little bit of logic gating on the output of the VCO to simulate that fed back from a disc/drum/mirror-screw/whatever.

IF, it's a big IF, it locks and seems to lock reliably then we can assume (perhaps) the variations we are seeing are down to mechanical variables.

Your thoughts anyone?

Steve A.

P.S. One thing I might add although probably not pertinent to this low speed application...not all 4046's are born the same... I did a design using Philips/NXP 4046's...all was well...I got paid. .Until the client switched to Fairchild versions of the same device 'cos they were cheaper. But in studying the two datasheets for the Philips and Fairchild both devices met their respective specifications, but the maximum VCO frequency for the Fairchild device was far lower than the Philips device.... buyer beware syndrome...bloody accountants...
Last edited by Steve Anderson on Wed Oct 19, 2011 12:18 am, edited 2 times in total.
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 kareno » Wed Oct 19, 2011 12:10 am

Is Peter Smith on this forum? He's the man to ask!
kareno
 

Postby Steve Anderson » Wed Oct 19, 2011 12:23 am

kareno wrote:Is Peter Smith on this forum? He's the man to ask!


Karen, I don't think so, can't find his name in the memberlist...

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 » Wed Oct 19, 2011 12:47 am

I think the big problem is that the 4046 circuit can't work with its complex inter dependence between phase, frequency etc... and the fact that it appears it does is currently one of those mysteries of the universe :)

One of these days I'll get on with designing a synchronous motor driver... I have to clear other projects out of the way first though :roll:
Find the workbench under the broken UPS, bits of ZX spectrum, network cables... The less said about half restored tractors the better...
M3DVQ
Just nod and pretend you understand me
 
Posts: 338
Joined: Sat Feb 10, 2007 10:52 am
Location: Lincolnshire

Postby Viewmaster » Wed Oct 19, 2011 1:01 am

I don't think this will help much but Peter Smith has a web site........
http://www.radarc.org/From%20Catswhiske ... ision3.PDF

He talks about the 4046 and here is what he says just over half way down the page...........

The 4046 contains' two phase comparitors and the advantage of using No. 2 is that its function is independent of the input waveform's duty cycle. The error voltage on the output (pin 13) of the phase comparitor is a pulse whose width and polarity represents the phase difference between the sync pulse and the pulses from the drum optical speed sensor.
This signal is integrated by R5 and C4 and mixed with a proportion of the signal fed back from the motor via R8, VR1 and R6. This provides the damping of the motor to eliminate "hunting".

.........end quote
User avatar
Viewmaster
Frankenstein was my uncle.
 
Posts: 1306
Joined: Fri Apr 06, 2007 4:50 am
Location: UK Midlands

Postby Viewmaster » Wed Oct 19, 2011 1:20 am

It seems as though it's not only NBTVers that have
trouble with the 4046 . He says the circuit has a mind of its own and not predictable :-)

http://www.youtube.com/watch?v=py8N5RrFO8E

Hope we don't loose our marbles toooo!
User avatar
Viewmaster
Frankenstein was my uncle.
 
Posts: 1306
Joined: Fri Apr 06, 2007 4:50 am
Location: UK Midlands

Next

Return to Mechanical NBTV

Who is online

Users browsing this forum: No registered users and 26 guests

cron