[sdiy] Timing : How close is close enough?

[Tom Arnold]

On Wed, Feb 04, 2009 at 11:11:09AM +1030, Matthew Smith wrote:
> Me too, but not that often ;-)  I'd certainly use a microcontroller and 
> would probably be setting the frequency/BPM using a set of BCD switches 
> (means you need no display.)

If I use an LCD or LED readout ( both cheap, LCD easier to implement ) it
means the hardware is far more generic.  Just a CPU with a couple buttons
and a bunch of jacks. 

> If you wanted "fairly good" you could use a decent quality crystal for 
> the uC clock.  If you wanted "really good", you could have a 
> higher-accuracy timing source firing an interrupt pin, which would then 
> drive the counter that does the business.  There are plenty of ovenised 
> quartz oscillators on eBay, or you could find a GPS module that puts out 
> a GPS-disciplined 1Mhz signal.  (I think that the Trimble Thunderbolt 
> has one of these.)

Yeah, standard crystal I think is accurate enough in this regard.  :-)

> Slightly more down to earth, if a 32,768Hz clock would give you the 
> required resolution, you could drive your interrupt pin with something 
> like a Dallas DS32KHz, which is a temperature-compensated quartz oscillator.

Now we have the accuracy vs error tradeoff.

>From emails I've gotten, sounds like as usual I spend time thinking about
that which isn't important.  Probably why I never get anything done...

Tonight I think I'll code this up and see what happens.

Thanks!

-- 
 -----------------------------------------------------------------
 - Tom Arnold       -  Free Synth DIY webspace http://www.sdiy.org 
 - SynthGeek        -
 - BBD Fanatic      -  "...is it a virus, a drug, or a religion?" 
 - echo evho wjxo   -    Juanita Shrugs. "What's the difference?" 
 --------------------