Construction Diary -- Part 2, Spinning the Nipkow Disc

[Andrew Davie]

OK, big news. It works!

The attached image is the first I took with the new synchronisation in action. I have to say, though, it's incredibly touchly... and woeful in performance. The picture isn't that stable -- perhaps shifting up and down in a sort of swing cycle... sometimes it's good, sometimes bad.

I've yet to find the magic combination where things don't tend to go in a runaway loop -- overadjusting once, then getting out of phase, and accelerating in the wrong direction... and then never recovering. I can get it stable, but as I said... very touchy.

I'm sure I'll be happier with it once I learn how to adjust it properly. I was very disappointed with my picture when I first got an image... but warmed to it after a few tweaks. Probably the same with the automatic motor control.

The motor is quite noisy when it's turned on and off by the motor circuit. An almost 'clunk' sound... quite disconcerting when it's synching and turning on/off all the time (every few seconds). Perhaps a different motor will cure this.

But, it has to be said... it is actually functional. So, what changed? Well, truth be told, I had a misplaced connection of my 100K resistor to the 100k trimpot. I 'missed' on the breadboard by one column, so in reality the 4046 wasn't connected. Whoops! It was quite a subtle error, as the trimpot hung over the column in question, it just wasn't the right one.

Ah well. Many thanks to everyone who has helped me to this point. I'm not quite sure what I'll do from here... probably play with how this circuit works, for a while. I have this gut feeling, though, that it *has* to be better than this. For example, it can take up to 20 seconds to obtain that first 'frame lock', and that just doesn't seem right to me.

[Klaas Robers]

I made a new diagram. The 4011 is used only half. It contains 4 NAND gates and I used only two. Don't forget to connect pin 7 to ground and pin 14 to +12.

The circuit forms a Schmidt trigger. At a certain voltage level which is about 6 volts the output changes very fast from 0 to +12. The 1M resistor sees that this happens. If you connect one channel of your oscilloscope to the input, pins 5 and 6, and the other to the output, pin 3, then you can see it happen. Adjust the 100k pot (I changed the value because you didn't reach the correct voltage with 47k) that the output gives nice pulses. This should be very uncritical!

I changed also the motor supply to 20V as I saw that 12V is too low.

[Viewmaster]

What puzzles me is that the original motor control circuit is in the handbook and a similar one on Peter's site too.
At the recent NBTV convention about 12 machines were running ..all syncing well!
So, surely many are using the original circuit very successfully without a Schmidt trigger to sharpen up the pulses or did they all suffer trouble and used an entirely different method to sync?
Albert.

[Klaas Robers]

Congratulations!!!

However, please look back to the 100k pot. Connect it as if it was a 220k pot. The real value is rather uncritical. Put it in the middle position and refind lock. Now if the disc is over reacting (this is what I read from you now) slide it towards the 10k potmeter, i.e. away from the 4046. The system will get more quiet.

Then gently put your finger shortly against the disc and watch how the disc comes back to the correct framing. This should happen almost without the picture swinging up and down. You can adjust that with that 220k pot (or now 100k). The further you slide away from the 4046 the slower the disc reacts. It should also not be too slow of coarse.

[Andrew Davie]

Congratulations!!!

However, please look back to the 100k pot. Connect it as if it was a 220k pot. The real value is rather uncritical. Put it in the middle position and refind lock. Now if the disc is over reacting (this is what I read from you now) slide it towards the 10k potmeter, i.e. away from the 4046. The system will get more quiet.

Then gently put your finger shortly against the disc and watch how the disc comes back to the correct framing. This should happen almost without the picture swinging up and down. You can adjust that with that 220k pot (or now 100k). The further you slide away from the 4046 the slower the disc reacts. It should also not be too slow of coarse.

Will wonders never cease?

I followed the above instructions, but things were no better than before. Basically I had a very poor synchronisation/speed control, the thing often went haywire, and it was rolling across probably 2/3 of a frame in position even when synched.

I replaced the rubber band with a spare "real" rubber drive belt. And while I was doing that I also figure-eighted the band (so my picture was the right way up). And voila! Not only much improved, but nearly perfect! The synching is strong and reliable. When it's running, the picture is almost -- not quite, but almost -- rock solid in position. Only every 10 seconds or so is a minor waver of a fraction of a millimeter. When I slow the disc with my finger and let go, it snaps back to lock in a half-second or so. Better yet, it synchs automatically from startup. That is, if I remove power, then start with the disc not spinning, it will spin up to synch automatically.

The only thing it doesn't do is slow down if it is already going too fast. Since I'm going to have a button to press to allow the frame timing/line synch (centering the frame) anyway, this will disconnect the motor and down speed. So that will solve the 'too fast' problem anyway.

I've gone from very disappointed to fabulously happy with this in the space of a half an hour. So many thanks once again, and here's my advice to everyone following this thread and building their own...

DO NOT USE RUBBER BANDS AS DRIVE BELTS!

[Viewmaster]

It is also interesting that I couldn't find a similar restriction in the datasheet for the MC14046B, the Motorola equivalent to the CD4046B. I don't know if that means the 14046 doesn't have the restriction or if they just left it off the datasheet.
Steve Treadwell

I note that Peter Yanczer uses this MC14046 for his motor control circuit
http://www.televisionexperimenters.com/tv_build6.html

So this maybe the solution as Tex indicates.
As Peter joined this forum last month and if he is reading this thread I wonder if he has any comments to make on the use of MC14046 as against the 4046 ?
Albert.

[Tex]

Andrew - congratulations! You've finally seen some rewards for all your hard work.
Here are some suggestions, just off the top of my head, for solving the remaining problems:
For the belt needing to be in a "figure 8" - you may need to insert a small idler pulley mounted halfway between the ends of the loop, and at right angles to the motor and hub in order to prevent the belt rubbing against itself.
To maximize the synchronization range, I would temporarily disconnect the 4046 and adjust the free-running motor speed as nearly as possible to the desired value, then reconnect the 4046.
Inability to slow down - you probably have quite a bit of inertia in your disk, and as there is no dynamic braking on your motor, the 4046 can increase speed much more rapidly than it can decrease. One thing you might try so as to provide some braking capability is either a diode and resistor in series around the motor or several diodes in series and no resistor. The diodes should have the cathode ends toward the 20 volt supply and should be rated to carry at least the motor current, maybe something like 1N4001? If you use a resistor, it should not be too large a value, maybe just a few ohms, and capable of dissipating several watts. I can't give specific values, since I don't know the details of your motor - perhaps others on this forum could give more help here, and advise if they think this is worth a try.

Steve Treadwell

[Tex]

Addendum to my last post - I'm thinking maybe the diode circuit around the motor might not work after all. I was trying to get some sort of a "dynamic braking" which I have seen before where the motor leads are shorted together after power is removed to allow the motor "back-EMF" to create a reverse motor current. It looks like the diode scheme I proposed above won't allow the current to reverse. Anyone have any ideas on a way to implement this?

Steve Treadwell

[Tex]

Here's a possible reason that the motor won't slow down if it's going too fast:
First, consider the situation where the motor is synchronized and then begins to run just a little bit slow - the 4046 sees that the pulses from the phototransistor are delayed a little from the incoming sync pulses, so speeds up the motor, which it can do quickly and remain "in sync".
Now consider the situation where the motor is synchronized and then begins to run just a little bit fast - the 4046 sees that the pulses from the phototransistor are coming in just a little earlier than the incoming sync pulses, so tries to slow down the motor by decreasing the current. But due to the inertia of the disk, the motor speed doesn't slow appreciably. Meanwhile, the phototransistor pulses are coming in at a faster rate than the sync pulses, so they are slipping backward in phase with respect to the sync pulses. After they have slipped back by half a line, they now seem to be coming in too late with respect to the sync pulses, so the 4046 thinks the motor is going too slow and tries to speed up again. This alternating speedup and slowdown process continues with the result that the motor speed just increases without ever again achieveing synchronization. It seems that some way is needed for the 4046 to slow down the motor as rapidly as it speeds it up.

I assume others have built this circuit - I wonder why they haven't reported the same problem.

Steve Treadwell

[Andrew Davie]

I assume others have built this circuit - I wonder why they haven't reported the same problem.

Possibly because the circuit appears to be very capable of 'catching' the overshoot when the speed is almost right? I only see the runaway speedup when there's no video signal (eg: going between tracks on the CD and there's an appreciable pause)... and here, I think, lies the problem.

Occasionally when I don't have a signal, my motor slows down and stops. Occasionally, it goes full hell for leather. I believe it's to do with the trimpot controlling the IR signal strength. I suspect I just need to adjust that back down so that there's no motor when there's no signal, and then the runaway speedup won't be a problem. The disc would never be able to get into that state. I suspect this is why this problem is not on others' systems, because their motor is not speeding up when there's no signal present.

Many thanks for your assistance with this circuit, Steve. It's been a great learning experience, and the results are fantastic!

[Roland]

Roland,
As I recall (and this is going back a long time) these dynamos generated about six Volts AC at a 'normal' bicycle speed. To use them 'in reverse' you'll need to supply it with about the same, perhaps a little less.

Steve, thanks for the info. I'm currently busy with other projects so I can't get down to experimenting with it just yet - but at least I have a better idea of what to do now. My dynamo is the old 'bottle' sort - probably at least 20 years old - so hopefully won't have any nasty surprises.



Roland.

[Steve Anderson]

As Peter joined this forum last month and if he is reading this thread I wonder if he has any comments to make on the use of MC14046 as against the 4046 ?
Albert.

There's no difference between the parts, it's just different manufactures way of numbering the devices. MC14046 (Motorola), HEF4046 (Philips), and 4046 is just the generic name. There is an exception, e.g. A 74HC4046 is NOT the same.

The 74HC4046 (and 74HCT4046) are functionally the same, and the pin-out is the same. But the HC(T) devices are for use with 3V to 6V supplies. (Mostly 5V), the others are for 3V to 15V.

Why the difference? The HC(T) parts are much faster, the others move at a snails pace, but for NBTV it's usually of no issue.

I attach two datasheets for this device for those that don't have them.

Steve A.

Postscript....Be aware that many manufactures of the MC/CD/HEF series of logic chips are discontinuing production as demand has dropped off. You'll probably be able to get them for the next few years, but it's best not to use them in new designs.

[Andrew Davie]

Today was a bit of a consolidation day, now that the motor control circuit basically works. I hooked up the oscilloscope and had a poke around looking at the signals. The surprising bit was that the signal from the IR sensors was very ugly. The top half of every second pulse was missing, with a sort of wavy flat top to it, and superimposed over the entire was a visible copy of the line pulse. Looked to me like there was an AC signal coupled to it AND the line pulse. I've seen this before, don't know what caused it.

So, I pulled out wires here and there, and had a look-see to see what was happening. One wire that seemed to be at fault was the one going to pin 3 of the 4046. I don't really know how it was wrong, but this wire 'fed' the line pulse noise and the IR signal cut-off 'into' the signal. I carefully re-seated all components and wires, double-checking against the circuit once again -- and it just sort of 'came right'. Now I have a very clean IR signal, once again - sitting around 10V with no noise. It's reasonably possible that I'd missed connecting pin 16 to ground. Not sure, just admitting possible liability

Then I replaced the cassette motor, firstly with the store-bought modern motor. I bought it, so I wanted to see how it worked. This was, I might say, a total waste of time. It's like taking away someone's rolls royce and asking them to drive a moped. The new motor clearly didn't belong. It's hard to describe, but it just felt less substantial... I junked it after less than a minute. Unfortunately, I'd re-used the small wheel/pulley that sits over the motor around which the belt wraps... and I broke it when I forced it on the new motor, so it was not useable for the original anymore. I'll have to make another, or possibly buy one -- but Jaycar doesn't carry them. So now I'm without my prime motor

There is another, sitting on the second motor that Peter Yanczer sent me, so I thought I'd try out that motor. My first impressions were not that great -- a lot of stopping and starting -- but after playing with various things like the belt tension (by repositioning the motor), and the wiring (as noted above)... it started to work pretty well.

It's clear that there are a large number of variables each of which needs to be 'pretty right' for the whole thing to work nicely. Belt tension, belt material, IR signal, motor torque, motor speed setting, 'lag', etc., etc. I've seen the thing work brilliantly, and I've seen it work very badly. Right now it's working 'well', though it takes forever to synch. I do know it's just a matter of finding the right adjustments for all the variables.

One thing I seem to have 'lost' is for the motor to turn completely off when there is no signal. It seems to want to run (slowly), but occasionally kicking into some sort of feedback loop and speeding up to warp factor 9. Again, I suspect that there are some adjustment ranges I can choose where this sort of thing doesn't happen.

I think the next thing I'll do is add a momentary switch so that I can 'frame' the scanlines correctly -- this would, I think, be best placed across the motor +12 line and would just remove power from the motor while pressed. Actually, the motor has a +20V, doesn't it. And so, I've just checked out how it all works by running the motor off the separate 12V supply, but also the LED array off the same. And it works beautifully. Lucky I accidentally designed the LED array to work off 12V.

I might even have a play driving everything off the same 12V supply, as suppled by the NBTV boards. I know, bad idea. I'll have a play, anyway, and report the results. My gut feeling is that it's going to work A-OK. I kind of like the 'elegance' of only having one rectifier/supply.

After that, I think it's about time to transfer the whole circuit from my prototype board onto the single-board with my other circuitry.

One other thing I did today was to connect up the sound circuit I built now oh-so-long ago. This is the first time that it's all been together and fully functioning as a working TV! I got to hear the people on the later tracks of the CD talking about this and that... this brought the whole thing together. Now I can turn up the volume, brightness, contrast... and the picture auto-synchs. Just need the framing momentary switch and all the mechanics will be complete!

Attached are various images taken from the latest incarnation -- using the 2nd Peter Yanczer motor -- playing with various settings on the camera. Typically, though, I'm using 1/8sec exposure at f2.5 -- whatever that means Images are not enhanced -- they're exactly as they came out of the camera.

[Klaas Robers]

I should place a photograph too, but I am not impressed of what my camera makes of it. Too automatic.

In the white (yellow) circle you can see the uneven ilumination from the different LEDs. I have just a single plastic diffuser sheet, which still shows circles for each LED.

[Andrew Davie]

Bit of a flash of inspiration -- I wondered if putting an old webcam in front of the mechanical TV would produce a usable image. It does! And now I'm able to include a short movie showing the synching in action. This is on track 44 of club CD #1 -- Karen. The picture quality isn't so hot, and the diagonal line wandering over the picture is an artefact -- but the video does give a good idea of how the syncronisation works at the moment.

The disc will sync in 15 seconds from a cold startup. I'm happy enough with that, but the excessive wander before it synchs is something I'd like to fix.

Creating this AVI required me to install an Indeo Video codec. I don't know if it will play unless you have that codec -- if not, I will make an effort to convert it to a different format. Edit: Attached is a WMV version of same.