Doepfer

Here are my thoughts, but I stand to be corrected:

> look much like sine waves at all. Is there a user advantage or is 
this 
> just a trade-off design for cutting costs (i.e. lame diode tri-to-
sine 
> converters)?

It really amounts to a 'practical necessity' to what is a hard 
problem...

> Sadly from the scope shots, I noticed even the 111-2 still 
maintains 
> these low-quality sine waves. What am I missing?

...which is that to design analogue circuitry to produce a high-
purity sine wave voltage-controlled oscillator which will track over 
many octaves is a very demanding requirement (and which is perhaps 
what you are 'missing'?). Producing a fixed-frequency sine wave 
oscillator is hard enough, as the gain needs to be 'just right' to 
limit the amplitude, and normally some non-linearity (i.e. make the 
gain 'drop off' at some point) is needed to stop the wave getting too 
big, and this will inevitably introduce some distortion. Then making 
the frequency voltage-variable just adds more difficulty, especially 
if it is required to have the frequencies track in musically 
acceptable way.

Thus most designs opt for simplicity, using a triangle wave passed 
through some sine-shaping mechanism, and just accept that the 
output 'sine wave' may not be very 'true'. In the case of the A-110, 
the output is further 'muddied' in that the tri wave is generated 
from the sawtooth oscillator core, by 'flipping' one half of the saw 
(hence the little glitch you may have noticed in the tri and sine 
waveshapes, especially at higher frequencies).

The CEM3340 chip at the heart of the A-111 *is* a triangle-core type 
oscillator, but the sine-shaping is done by circuitry external to the 
chip. I don't know whether the A-111-2 is a discrete design, or based 
on same chip, but the problems will be the same.

If you want high-purity sine waves for additive or FM synthesis, then 
doing it analogue is probably not a good way to go - this is 
something that is far better suited to being done digitally, where 
controlling purity and things like the exact ratios of frequencies is 
more easily done!

(Just my thoughts, and like I said at the start, I stand to be 
corrected by more knowledgeable sources!)

Tim