NBTV Camera Question

[Harry Dalek]

erm, I'm not sure if that was what I am saying or not I think so.

With the encoder you have at present 32 slots but only 31 are filled, so you only have 31 pulses, but your PLL is comparing them against 32 pulses coming from you clock circuit - so yes you would expect some problems completely "locking" - I am saying to correct for that simply fill in the missing "slot" that should then allow "locking" but now you won't have the missing frame pulse. To correct that you can add another single slot either inside or outside the 32 slots and have another sensor pick that up - you can then mix the two pulse streams with an exclusive or to add in the missing frame pulse which is then sent to the PC (or whatever) - obviously this stream of pulses is not the one sent to the PLL it's the output of sensor 1 that is sent to PLL for locking purposes.

Of course, as a first iteration, you needn't add the second sensor etc - just get the locking working first is my suggestion.

No worries gary i was thinking it out from your idea and how i would try and do it...i will stick with this for now see how it gos but my gut feeling is you are right ...

I didn't play around with it last night but will today ,i need the neaten it up again adjusting the thing its all over the place..

So lets see if i can do better today .

[gary]

Well to test the theory it is really only a matter of removing that pasted on bit of paper so....

[Harry Dalek]

Well to test the theory it is really only a matter of removing that pasted on bit of paper so....

what the hell i will try and see ..if it does work better it would be worth looking into part 2 of the idea...

perhaps the experiment gos get it best i can as is then stop it remove the missing sync paper from the encoder try again and see or record both and see..mmmm

[gary]

Well, experimentation is fine if you feel you want to try it, but when the implementation goes against theory, and there is documented evidence from others who have tried it (usually accidentally) and it has caused problems then one wonders if you could put your time to better use.

Just a little "layman" theory regarding PLLs Harry, basically it it is comparing the two pulse streams - when one is high and the other is low there is an error signal output which either speeds up or slows down (depending on which is high and which is low) the motor (in this case). If the two pulse streams are in exact synchronisation then the error signal is zero meaning there will be no speed change.

Obviously the latter scenario can never happen in your case because there will always be one spot in the pulse stream when one is high (32 pulses) and one is low (31) pulses. Sure the system will get to the syncronisation speed but the missing pulse will cause a slight speed change each time it comes around causing a little "hunting" around the sync speed, which in turn causes a little waving of the sync line.

How much it contributes in your case remains to be seen, but I feel it is very likely it contributes some, if not all, of the variation.

[Harry Dalek]

Well, experimentation is fine if you feel you want to try it, but when the implementation goes against theory, and there is documented evidence from others who have tried it (usually accidentally) and it has caused problems then one wonders if you could put your time to better use.

Just a little "layman" theory regarding PLLs Harry, basically it it is comparing the two pulse streams - when one is high and the other is low there is an error signal output which either speeds up or slows down (depending on which is high and which is low) the motor (in this case). If the two pulse streams are in exact synchronisation then the error signal is zero meaning there will be no speed change.

Obviously the latter scenario can never happen in your case because there will always be one spot in the pulse stream when one is high (32 pulses) and one is low (31) pulses. Sure the system will get to the syncronisation speed but the missing pulse will cause a slight speed change each time it comes around causing a little "hunting" around the sync speed, which in turn causes a little waving of the sync line.

How much it contributes in your case remains to be seen, but I feel it is very likely it contributes some, if not all, of the variation.

Thats pretty informative gary ..i was thinking it would have to always cause a bit of motor swing ,theres no delay in the clock pulse but there is in the mechanical clock the encoder system so they can never be the same .

I should but haven't really looked yet at the out put pulse of the PLL i suppose its pulse width change that adjusts the motor speed along with the voltage to the transistor . .


So it sounds like i am stuck with a bit of a motor speed swing .the bistable switch seemed to work just as good perhaps less fussy than the PLL on the last monitor motor ...

I tried removing the encoder missing pulse i didn't notice a change .

[gary]

i was thinking it would have to always cause a bit of motor swing ,theres no delay in the clock pulse but there is in the mechanical clock the encoder system so they can never be the same .

Well only if you have a 32/31 arrangement if it is 32/32 or 31/31 (as in a monitor) there is no reason you should have a substantial swing.

i suppose its pulse width change that adjusts the motor speed along with the voltage to the transistor . .

I never think of it that way but yes I suppose the pulse width will represent the phase difference between the 2 inputs - really the important thing is the charging of the capacitor in the low pass filter - it is adjusted up and down until the phase AND frequency are identical at which point the output goes open circuit allowing the capacitor to remain at just the level to keep the motor rotating at the right speed.

So it sounds like i am stuck with a bit of a motor speed swing .

Well as I say, only if you persist with the 32/31 arrangement - or go to a much larger disk with high inertia - that is claimed to overcome the "missing sync" variation.

the bistable switch seemed to work just as good perhaps less fussy than the PLL on the last monitor motor ...

Well, they work in almost the exact same way - in fact I use that arrangement because I feel it is easier to detect problems and adjust.

I tried removing the encoder missing pulse i didn't notice a change .

Well that surprises me. Can you post an example?

BTW what power transistor are you using?

[Harry Dalek]

Hi Gary

Well only if you have a 32/31 arrangement if it is 32/32 or 31/31 (as in a monitor) there is no reason you should have a substantial swing.

I will go with that reasoning .
I have to point out one thing on my circuit i used the idea that drove the last monitor apart from this one is a mosfet ... talking about the motor driving part ...i do have the club one on the circuit board its just by passed at the moment .
Perhaps i should just use the club circuit in full ,was not happy last nights tests ..i could see it swinging on the strobe disk so i gave up after a bit of playing around...found 2 bad grounds and a trimmer pot on the motor power supply was a bit ruff cheapo one from china .

I never think of it that way but yes I suppose the pulse width will represent the phase difference between the 2 inputs

Does mean the encoder motor side is most of time the problem...changing this width but would also change the frequency slow down or speeding up of that disk?

Just an idea

I wonder if the IR led were connected to the crystal clocks output and flashing at 400 hz in stead of always being on so if the encode circuit works it would mean the disk was in sync with the crystal clocks pulses...i think ?

- really the important thing is the charging of the capacitor in the low pass filter - it is adjusted up and down until the phase AND frequency are identical at which point the output goes open circuit allowing the capacitor to remain at just the level to keep the motor rotating at the right speed.

Is this some thing you could adjust if need be in the circuit .

Well as I say, only if you persist with the 32/31 arrangement - or go to a much larger disk with high inertia - that is claimed to overcome the "missing sync" variation.

Oh thats strange i would of thought it would have more to do with the motor to size of disk ..i am using a tiny motor that was meant for this type of disk to run at correct speeds....but if you think its better to try no missing sync 32 32 sounds better to me anyway...

Well, they work in almost the exact same way - in fact I use that arrangement because I feel it is easier to detect problems and adjust.

Yes not much to them worked really good on the again last monitor iwas glad i used your suggestion when working this out at the time ..perhaps i should of went this root again darn .

Well that surprises me. Can you post an example?

Its was really off last night i was getting no where same results no where as good as night before so gave up last night some times best to rest and rethink it ....reason i didn't post i could see the swinging so either way it was not a good result ...

BTW what power transistor are you using?[/quote]

Mosfet BUZ71A ...gate is connected to the cub circuit same place the the clubs fets gate is ...i have the Drain going to the motor and the motor other connection to a lm317 var pot 12v and S is to ground ...so its a little different ....i just used that for a start as i could just switch the gate from my PWM motor driving circuit to the PLL one..cheating a bit ..i do have the rest of the clubs motor drive circuit built on the board just not used at the moment .

[gary]

Does mean the encoder motor side is most of time the problem...changing this width but would also change the frequency slow down or speeding up of that disk?

No, the PLL works only on leading edges so the duty cycle of the pulses doesn't come into it. No the most common problem is just the components marked "adjust on test" - as they have to be tweaked for almost every new implementation (different motor, disk size, etc) - and there is no real guide as to how to tweak them.

OTOH if you meant changing the width of the output of the PLL, very definitely NO - this is determined by the PLL and represents the phase error it detects. It wouldn't work at all if you arbitrarily changed it.

I wonder if the IR led were connected to the crystal clocks output and flashing at 400 hz in stead of always being on so if the encode circuit works it would mean the disk was in sync with the crystal clocks pulses...i think ?

Thinking about that concept makes my brain hurt I think it would guarantee instability. No the idea is to pass a stream of pluses from the encoder to the PLL that is at the exact same frequency and the exact same phase as that of the clock circuit - that only happens when the disk is running at 750 RPM and the encoder pulses are aligned in time with the clock pulses (at least the leading edges of those pulses).

Remember, the idea is to have NO output from the PLL - this only happens when the disk is running in sync with the reference (clock) pulse stream.

Is this some thing you could adjust if need be in the circuit .

yes the capacitor value is the most common thing that I "tweak" in this type of circuit.

Oh thats strange i would of thought it would have more to do with the motor to size of disk

Well both have an effect but the disk (being a flywheel) is almost always the greater source of inertia.

i am using a tiny motor that was meant for this type of disk to run at correct speeds

Generally, this kind of circuit should work with any motor.

....but if you think its better to try no missing sync 32 32 sounds better to me anyway...

I do, it is the only way to remove that particular variable - this is really the only way to debug these sort of things - eliminate variables one by one.

Mosfet BUZ71A ...gate is connected to the cub circuit same place the the clubs fets gate is ...i have the Drain going to the motor and the motor other connection to a lm317 var pot 12v and S is to ground ...so its a little different ....i just used that for a start as i could just switch the gate from my PWM motor driving circuit to the PLL one..cheating a bit ..i do have the rest of the clubs motor drive circuit built on the board just not used at the moment .

I think the BUZ71A should be fine. But what is in between the PLL and the Gate?

[Harry Dalek]

No, the PLL works only on leading edges so the duty cycle of the pulses doesn't come into it. No the most common problem is just the components marked "adjust on test" - as they have to be tweaked for almost every new implementation (different motor, disk size, etc) - and there is no real guide as to how to tweak them.

Does sound like a bit of trial and error...i'd rather how it is now than not work at all..have another look tonight .

OTOH if you meant changing the width of the output of the PLL, very definitely NO - this is determined by the PLL and represents the phase error it detects. It wouldn't work at all if you arbitrarily changed it.

Ok that sounds like a no goer but this beast needs taming .

I do, it is the only way to remove that particular variable - this is really the only way to debug these sort of things - eliminate variables one by one.

I see your thinking here ...the only other thing i can think of that is different is the change in the motor circuit i did .


I think the BUZ71A should be fine. But what is in between the PLL and the Gate?[/quote]

A switch 100k pot and the cap ...you see that in this upload..where the 200k pot is i have 2 100k using that first one as an adjustment..just as this 200k does ...i can draw the circuit if it doesn't make sense... i am doing the fast slow control via the other end of the motor with the lm317.

[gary]

Ok, that's fine Harry, that's pretty much the standard circuit, at least to the right of the 4046, something you wrote caused me to assume you had made major changes to that part.

Presumably your clock circuit is coming in on pin 14?

With what you have shown there I can see no reason why it shouldn't work.

I suppose the next step, if you haven't already done so, is to use your scope to have a look at all the signals in and out.

Of course, in your case, the switch is unnecessary but it can't hurt being there (unless faulty).

[Harry Dalek]

Ok, that's fine Harry, that's pretty much the standard circuit, at least to the right of the 4046, something you wrote caused me to assume you had made major changes to that part.

Ok gary no its pretty basic perhaps to basic reading some of these improvements the odd added part here and there ..i looked into whos got it going what they did ,i think there are a few more but i like these .

Presumably your clock circuit is coming in on pin 14?

Oh yes sir ...what i will do is draw the schematic today ...i know it sort of works and worked better at the start than at the moment ...it will be good for me reviewing .

With what you have shown there I can see no reason why it shouldn't work.

Well it does but a slight oscillation back and forth...which i know is big for the end result ..much better the other day .

I suppose the next step, if you haven't already done so, is to use your scope to have a look at all the signals in and out.

Yes i need to look but reading what others have down with same sort of problem perhaps looking into these ad ons and such what they did might be worth a step back before i go forward. i was very interested in Doms first go Andrew as theres sounded worse than mine even Doms so so at the start and both ended up working fine so must be a solution ...mine is working but oscillating and i think mine has to be fussy as i have a PC this time telling me whats right what wrong not much % movement wanted for these beasts of the modern age .

Of course, in your case, the switch is unnecessary but it can't hurt being there (unless faulty).

Its a tiny one for circuit boards i just added it just in case as i didn't know why some of the 4046pll circuits had it easier to ad now just in case .

Looking at the dom screen grab i think i am using the same type of 4046 so this is interesting .

Bigscreens circuit add ons or corrections i know i don't have 10k on pin 3 from a quick look ..

So i see as you say there have been some custom fixes to the PLL problem.

I will Draw my circuit tonight .

[Harry Dalek]

Heres my circuit so far for the motor control ...some parts left out on the club circuit as its not in use apart from what you see

The Gate to the fet has a 2 way switch to swap to PWM motor control as i started with that ..left it out here as this will do for now interested in th e PLL problem.

The PLL is a`CD4046BE.

[Harry Dalek]

Hi Gary

I am going to do the sync tests today again i found a problem with a ground to the sync boards so it seems to work better again.
these two files i again adjusted the circuit so i could not see the strobe move .
These 2 are both with the missing sync i will remove the paper again and do the test again and post more than likly a 3rd file in this post so check back later if its not on yet.

Ok back again i hooked up my scope know the meter is slow but what i am seeing is noticed it took longer to get to 400hz and for me didn't go over that just under a bit and back up to 400..

I didn't adjust the thing between the tests just removed the missing sync so i had 32 slots.

Another update

just did a video file i had noise problem had to knock back the light sensor perhaps the batties low ? any case did a test card video of my cat face nbtv drawing it must be arounf the right speed now as i see its almost full screen plays ok .

will have to track down noise problem !

[Harry Dalek]

I didn't have a chance to have a good look at last nights test card i filmed till now ,i can see removing the missing sync the start and end line looks to me to be in the middle of the picture ..could be a random start ..not sure .

On the schematic i did do a change i am using the output of the 555 timer right to the pin 3 of the PLL ...

Thinking about it perhaps its not going to matter as the encoder speed and motor are controlled by the PLL but wonder if i out put the PLL to the video mixer instead of the 555 timers .

[gary]

I didn't have a chance to have a good look at last nights test card i filmed till now ,i can see removing the missing sync the start and end line looks to me to be in the middle of the picture ..could be a random start ..not sure .

That's an arbitrary thing Harry, there is no frame sync without the missing pulse - totally normal.

On the schematic i did do a change i am using the output of the 555 timer right to the pin 3 of the PLL ...

Harry, i have only just seen your schematic - it wasn't there when I originally read your post - but I see there is a considerable change on the right hand side in that you are not using the feedback portion of the handbooks schematic - that may be worth adding. In fact, if my understanding of the circuit is correct (strangely there has never been, to my knowledge, a real explanation of how the motor current feedback part works) then that 22 uF capacitor and the second trimmer shouldn't be there if you are not feeding back some of the motor current.

With regards to the output of the 555 going to the 4046 - that is what I had assumed you were doing and is perfectly correct.

if i out put the PLL to the video mixer instead of the 555 timers .

What would you be trying to achieve by doing that Harry? it would mean you would have a varying sync whilst it is coming up to speed, and no sync pulse at all when it was at speed. *However, you COULD use the clock output to feed the mixer - it would only be correct when the disk speed is exactly 750 RPM however.

Harry I am trying to analyse your wave files and finding some strange things that I am having difficulty explaining. Can you answer a couple of questions please?

1) with all of the files except the cat card there is not meant to be any picture right? Can you tell me was the FSS sensor picking up any light at all during the test or was it disconnected?

2) are you adjusting the speed at all during the actual recording or are you setting it first and then recording?

3) is the encoder sensor vibrating at all when running the disk?

BTW the cat example shows that the circuit IS trying to lock but is hunting around the lock point - this is a common syndrome most of us using this circuit has experienced and usually is overcome by tweaking the component values.

OTOH the others seem to be locked but at just slightly the wrong speed.

Actually it would be very useful if you could record the clock pulse stream directly (as indicated at *) so I could check it's speed.

Oh and BTW, why have you reverted to a 44.1 Khz sampling rate? It's a cow of a sample rate to use and serves no purpose unless you are going to burn to CD. 48kHz would be MUCH better. In fact ANY sampling rate at and above 20kHz and an integer multiple of 400 is MUCH better.