[sdiy] HF-VCO / counter / memory / DAC ?

[Richie Burnett]

Modern DDS chips are ideally suited to this application. They can generate frequencies from 0Hz up to very high MHz frequencies, with millihertz frequency resolution. They can change frequency at the drop of a hat without any of the issues of locking range and limited slew rate that phase locked loops suffer from. And their frequency drift is as good as the crystal oscillator that you drive them with.

They aren't VCOs though. You typically have to program their frequency digitally as a 32-bit number over SPI.

-Richie,

Sent from my Xperia SP on O2

---- Phillip Harbison wrote ----

>Michael Zacherl wrote:
>> Hi, based on the NCO thread, I have to ask: did a HF-VCO +
>> counter + memory + DAC combination go out of fashion?
>> I understand it’s not really feasible for a commercial
>> product, but for DIY?
>
>This describes my senior project in 1980-81. It was a high
>frequency clock feeding a bank or 4 74LS161 counters which
>I could program to get 256x the frequency I wanted. The
>output fed into 2 more '161s which generated an 8-bit
>address for a (for the time) 1024x8 static RAM which fed a
>fast (for the time) 8-bit DAC. That DAC fed a multiplying
>8-bit DAC that I used as digitally controlled amplifier.
>
>The worked and got me an A, but there was one problem. To
>be able to generate 256x any top octave frequency from a
>single clock I would need a ridiculously fast clock and
>ECL logic both of which were way outside my budget.
>
>I was not familiar with phase-locked loops in those days.
>If I built this again now, I would use the 4x '161s to
>generate the fundamental frequency then use a PLL to
>multiple that by 256. I plan to try it someday. I'm not
>sure a PLL can lock fast enough to not sound like it is
>constantly in "glide mode". I'm not even sure there are
>PLLs with that kind of range.
>
>Another possibility would be to have duplicated the
>clock + 4x 161's a dozen times then have a multiplexer
>that chooses which to use. Each clock/divider block
>would still produce 256x the fundamental, but only for
>one of the twelve notes on the scale. It might even be
>possible to vary the counters to do pitch bending. I
>suppose it would also be possible to have 12 clocks that
>are shared between all voices and only one divider with
>a mux to choose one of 12 clocks. That would be cheaper
>than 12x clocks for every voice.
>
>I will wait for everyone to tell me how crazy this is.
>What can I say. I was 21 years old at the time. I had
>also fallen into the depths of the digital abyss. ;)
>
>I still want to tinker with the PLLs.
>
>-- 
>Phil Harbison
>
>
>_______________________________________________
>Synth-diy mailing list
>Synth-diy at synth-diy.org
>http://synth-diy.org/mailman/listinfo/synth-diy