[sdiy] Saw DCO

[Richard Wentk]

At 11:30 07/10/2004 +0200, Fredrik Carlqvist wrote:

>Thanks for the suggestions. I think I will try the reference Vbe method
>first though, as it requires virtually no change in the hardware.
>
>Richard, are you suggesting leading the Vbe of another OTA into the
>trimmer's position on the integrating OTA? Could that work? Or perhaps even
>the OTA's own Vbe into the '-' input!?! At least, the temperature dependency
>could be reduced.

I haven't doodled out a circuit but I'd imagine you should be able to 
derive an anti-phase (as it were) temperature compensation reference from a 
spare OTA fairly easily. If you use a dual OTA there's not a huge 
difference in board real estate and the temp matching and compensation 
should be excellent.

But the real question is how much amp variation the temp sensitivity is 
causing. More than a dB or so would could be problematic, but then again 
long term temp variations in most working environments are unlikely. So 
unless you're planning to carry your synth from indoors to outdoors in 
Eastern Europe in winter while continuing to play it, it may not even be an 
issue. It might just mean you need to turn up/down the output level during 
the summer. :-)

>I think a control loop will be too slow even though I agree it would solve
>many problems (and freeing some precious code memory). When the pitch
>changes drastically, it will take at least a couple time constants to
>adjust, which would be 100ms with the lower freq limit 20Hz.

That was my problem with Magnus' suggestion.

>John, the hard synch comes in by multiplying the saw slope by a scale factor
>K larger than 1. This makes the saw hit the roof and reset. It is still
>reset every DCO cycle. The resulting frequency is a combination, just like
>in normal hard sync, where a faster oscillator is reset by a slower one. The
>fundamental is at the DCO frequency f, and there is an additional
>"fundamental" at f*K.

Neat. I'd be interested in your DCO circuit, if you want to make it public.

Richard