What means 'phaselocked'? (was: Re[3]: Fourier-Analysis

[Don Tillman]

   Date: Thu, 30 May 1996 19:48:05 +0200
   From: mz at bacher.co.at (Michael Zacherl - Bacher Systems EDV GmbH)

   As far as I can recall all tonewheels sit on a single axle.

Ummm, no.

The tonewheels sit on little axles, spring-coupled to black metal
gears, driven by white nylon gears, which sit on other little axles
that are linked together in line and driven by the motor.

   So my question: Is this phaselocked or not? 

"Phase locked" implies that there is an active mechanism to detect the
phase of the signal and adjust it if it's not right.  Using that
phrase would be slightly deceptive here.

   Date: Thu, 30 May 96 13:04:06 PDT
   From: gstopp at fibermux.com
   To: synth-diy at horus.sara.nl
   Subject: Re: What means 'phaselocked'?  (was: Re[3]: Fourier-Analysis

   As for digital dividers - a top octave generator such as the MK50240 
   generates 12 master semitones using ratios of large whole numbers, 
   like 355/113 and such. These divisors are neither Fourier 
   phase-locked nor irrationally related. Frequency division with 
   digital logic cannot produce true irrational relationships (due to 
   the fundamental definition of the word "irrational") and so the 
   output frequency intervals of top octave generators are merely close 
   approximations of equally-tempered scales. Again no phase lock.

If I remember correctly, and I don't think I have a sheet on a
top-octave-generator chip here to check, but I think the TOGs were all
just a collection of 12 divide-by-N circuits, with a single oscillator
running at, say, frequency M, so your outputs were M/N[1] for C,
M/N[2] for C#, etc.

The Hammond "oscillators" are driven by gear ratios, which are
required to be the ratio of two integral number of gear teeth.  So
that M[1]/N[1] for C, M[2]/N[2] for C#, etc.

Note the difference!  In a TOG chip there's a common numerator, in a
Hammond tone generator there are individual numerators.  Sort of like
a bicycle with a single front gear vs. a bicycle with a front
deraileuer (I have no idea how to spell that), the latter has much
finer control.

So it's likely that the Hammond tempering is a lot better than the IC
top-octave-generator.  It would be interesting to find out for sure.

   Date: Thu, 30 May 1996 17:59:05 -0400
   From: Robert May <wfrb at miworld1.miworld.net>

   I play a Korg CX-3 organ, which uses the MK50240 top octave chip, and the
   equal tempered approximations do not seem quite the same or as accurate as
   those in the Hammond organ. I have a data sheet with divisors used by the
   50240, but would like to know the various gear ratios used in the Hammond
   tonewheel organs. I posted this question to the Hammond list, but got no
   response.

Good question.  It's not in my Hammond manual and it's a real project
to count them all by hand.

      On a related note to this thread, would be worth the trouble to take
   several stages of all-pass (phase shifter) filtering and modulate each stage
   with filtered and uncorrelated noise to simulate the slight unsteadiness in
   the tonewheel organs?

I mentioned in the Mellotronists list that the gear used for tach
control of the Mellotron servo motor has enough variation in its
individual teeth to be readily visible on a scope, especially with my
period-measuring tachometer circuit (measures the period of each
individual cycle).  Not surprisingly, the pattern of unevenness
repeats every gear rotation.  I wouldn't be surprised if this was an
important part of the Hammond sound, as well as the phase variations
from the spring couplers.

So, perhaps a more accurate emulation of the Hammond tone generator
would have individual Phase Locked Loop M[i]/N[i] Frequency multiplier
circuits off (the equivalent of) the AC line.

Add a hashing function off the N[i] counter to tweak the PLL VCO on a
cycle-by-cycle basis to generate the wobble from individual teeth.
And add something else to simulate the spring coupling.

  -- Don