Re: 1-32 Khz in 1khz divisions

Barry S. wrote:
> I need to generate 1khz, 2khz, 3 khz, 4 khz, etc.. up to 32 khz..
> They don't have to be exact.. Each free running, each on their own
> output..

Those are pretty vague requirements. No oscillator is ever exact!
(And no vacuum ever perfect...) How non-exact can they be? If it's
+/-20%, then a bunch of 555 astables with RC's may be perfectly
reasonable.

Output square wave, sine wave, or what? How much jitter is allowed?

> My first inclination was to use counters, but that would be a few
> counters, prescalers, flip flops, etc because the divisions don't come
> out nicely. My second idea was to use a microcontroller with a lot of
> outputs.. E.g. Input A1 gets turned on/off every 1 ms (1 khz)..

I think the microcontroller is perfectly reasonable if you can live
with some jitter. As others have said, making a couple of harmonically
related outputs without jitter isn't a big deal. Doing it for 32
outputs that are not divided down from a master can be tricky (if
anyone knows a good trick I'm all ears! My back-of-the-envelope
calculation indicates the lowest frequency which can be divided down to
all of 1..32 kHz is something like 10^18 Hz...)

31 PLL's (e.g. 4046's) would also work with a reference of 1kHz and
simple dividers.

10-32kHz crystals are available (some standard values, most custom).
I'm thinking there might be some clever way to beat a couple of them
against each other and avoid all digital logic, just use a bunch of
balanced mixers and make the outputs be widely separated so they are
easily filterable. I have done this before (for example beating 32.768
kHz with 32.000 kHz to get 768 Hz... it's easy to filter out the 64.768
kHz nondesired product) with just a few transistors, but extending it
to 32 somewhat arbitrary outputs may be too much.

Below 10kHz I think the crystals are officially known as "tuning
forks". I've seen them show up in old test equipment, but I think you'd
need them custom made.

Tim.

.