[sdiy] Oscillator sync schemes]

Tue Aug 18 00:36:23 CEST 2009

Cynthia Webster wrote:
> cheater cheater wrote:
>> How does 'burst sync' aka 'kick sync' work? How are the pulses
>> generated? Would that be generated by a UC?
>>
>> How does PLL syncing work?
>>
>> Cynthia: want to shed any light on what your new sync mode do 
>> effectively?
>>   
> It is actually less like sync and more like a woven carpet of 
> oscillators in a matrix of cores, not "clouds".
> 'How to buss it around?'  is the issue at present and we're dreaming 
> up ideas on lovely summer afternoons...
>
>
> Cynthia
>
> thanks for all of that nifty math down below, it's downright 
> terrifying, lol!  (Great Method Though :)
>

I've always thought it would be clever to make Oscillators out of 
Stepper Motor Driver circuits.

With a stepper motor you must feed a four note sequence to four 
different wires essentially to
make the motor turn.  I doesn't run indefinitely simply when you hook up 
the power, instead it
may turn only 1/32 of a revolution and stop, waiting for the next of the 
four wires to be pulsed,
and so on...

Anyone working with stepper driven CNC milling gear has probably heard 
the lovely singing
melodies that the motors make when ramping up and down in speed 
following flowing B-splines
acceleration curves.   There are several types if stepper drivers 
including whole step, half-step,
micro-step, and sine-drive.  Years ago I was a partner in a special 
effects company called, YFS or
Young, Fast, and Scientific. Collectively we design a sine drive motion 
control system and it was
there that I developed a love for full four phase quadrature as 
presently offered in the Zeroscillator.
I also loved my Mutron Bi-Phase ad that's what inspired the switch on 
the ZO between the left and
right morphing sections.

So back to this idea of stepper motor driver chips as oscillators...
perhaps they'd behave more like clocks, (the lovely Modcan Clock module 
for example) which can
only land on certain frequencies depending on whether they are part of 
the present plateau or
loaded de-acceleration curve or not meaning that at audio frequencies 
there could be programmable
silences between certain frequencies perhaps omitting the need for for a 
VCA?

Or if the imaginary stepper was the master and the slave was in a sloppy 
phase locked loop tight, or
like a drunken sailor (Gravity knob) that might be interesting.

In real life you can wire two stepper motors together they will lock in 
sync with one repeating the
actions of the other like servos, except with no power supply.
This is the way synchronous sel/syn motors work, like they way you 
hook-up two movie projectors
to run some 3D movies, with the left projector = left eye, and the right 
projector = right eye in sync.

Or the way setting the time of day on one master clock in a ship's petty 
officer's office, can control
the hands on every other mechanical wall clock throughout an aircraft 
carrier.

This is the kind of sync or CoreLock® that we'er thinking of...

So using this paradigm continues to be interesting if one thinks of 
systems of four and two phase
Motor/VCOs and a bunch of modules that act as gear reducers, slip 
clutches, and planetary gears...

Anyway, hopefully gets you thinking right?

Cynthia

http://www.cyndustries.com/

>
>
>> Personally I was considering this kind of sync scheme:
>>
>> if the slave's phase is near the master's phase (+/- 0.1T for example)
>> then the current going into the integrator is lowered if the slave is
>> ahead, and raised if the slave is behind the master oscillator. How
>> much it is changed by, M, depends on some function, for example it
>> could be:
>>
>> M = a*(1-10*|s-t|)^2
>> for 0 =< |s-t| =< 0.1
>>
>> and M = 0 elsewhere.
>>
>> M is the percentage of mix between the slave's current and the
>> master's current. If M = 100% = 1 then the slave gets the same current
>> as the master.
>>
>> s - phase of master, [0..1]
>> t - phase of slave, [0..1]
>>
>> a - some constant. If it is 1 then the oscillators will end up in
>> perfect lock. I suggest it being a small value.
>>
>> There are probably better formulas for doing this, but this formula is
>> simple for everyone to understand the mathematics of.
>>
>> Option 2:
>>
>> M = a*(1-10*|s-t|)
>>
>> Option 3:
>>
>> M = a*(1-10*|s-t|)^k
>>
>> k > 1
>>
>> Option 4:
>>
>> M = a*(1-10*|s-t|)^k
>>
>> k > 0
>>
>> Option 5:
>>
>> M = a*(1-10*|s-t|)^k
>>
>> k = 0
>>
>> Option 6:
>>
>> M = a*(1-10*|s-t|)^k
>>
>> k - real number
>>
>> So this is a sort of FMish thing. Note that this 'sync' works
>> everywhere, not only where the master's core resets. It has nothing to
>> do with the shunting. However you could make it work only around the
>> master's reset spot by windowing the value of M with another function
>> which assumes 1 at the zero crossing, assumes 0 at -0.1T and at 0.1T
>> and is linear inbetween (so called 'tent function': /\  )
>>
>> Regarding the 'burst sync':
>>
>> maybe the master oscillator could have a sup-octave oscillator running
>> at, say, f_O = 16*f_M (16xfrequency of Master)
>> then it could strobe every time its integrator shunts, and on that
>> strobe you could compare the voltages of the master and slave. Then
>> you could do this:
>>
>> D = |s-t|^2k * (s-t) * a
>>
>> Where a =<1 and k is an integer. (a > 1 could sound interesting, who 
>> knows)
>>
>> You add D to the current of the slave oscillator.
>>
>> So this is again FM, but it's sort of 'stochastic'. But the process is
>> spaced in time in such a way that it is harmonically related to the
>> master oscillator.
>>
>> Now if you replace f_M with f_S (frequency of Slave) then you get an
>> alternative where the 'sync sound' is harmonically related to the
>> frequency of the slave. This would require the 'sup-oscillator' to
>> have its own integrator core, though.
>>
>> D.
>>
>> On Mon, Aug 17, 2009 at 8:38 PM, Cynthia 
>> Webster<cynthia.webster at gte.net> wrote:
>>  
>>> Chris Muir wrote:
>>>
>>>    
>>>> On Aug 17, 2009, at 9:29 AM, Cynthia Webster wrote:
>>>>
>>>>      
>>>>> The article in WikiPedia is Not Correct saying that there are only 
>>>>> two
>>>>> types of sync,
>>>>> as there is also Variable Sync, Reset, and Variable Reset, there 
>>>>> is also
>>>>> Triangle core Dual sync where it syncs twice as often at both the 
>>>>> top and
>>>>> the bottom,
>>>>> there is also Time Reversal, and there is Core Lock.  Who the hell 
>>>>> writes
>>>>> these things?
>>>>>         
>>>> We, the people, write these things. You did correct it while you were
>>>> there, right?
>>>>
>>>>       
>>> Hi Chris!
>>>
>>> I tried but it was above the part that was editable for some reason. 
>>> Maybe
>>> you have to be a member.
>>>
>>> Of course I understand the concept of the Wiki, but I see no name 
>>> associated
>>> with the entry,
>>> so mine was more of a rhetorical question, "who writes these things"
>>> meaning,
>>> (Who thought they were qualified to write that?   I certainly didn't 
>>> think I
>>> was, lol!)
>>>
>>> Aaron is the professor, and Paul has a fine pedigree, there are so many
>>> other brilliant people here
>>> (please don't shoot, I just just design stuff!)
>>>    
>>>> - C
>>>>
>>>> Chris Muir
>>>> cbm at well.com   http://www.xfade.com
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>       
>>> _______________________________________________
>>> Synth-diy mailing list
>>> Synth-diy at dropmix.xs4all.nl
>>> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>>>
>>>     
>>
>> _______________________________________________
>> Synth-diy mailing list
>> Synth-diy at dropmix.xs4all.nl
>> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>>
>>   
>
>