Re: [avr-chat] Error in Timing.

[Galen Seitz]

Erik Christiansen <address@hidden> wrote:

> On Wed, Mar 01, 2006 at 11:31:14AM +0530, Royce Pereira wrote:
> > I used 2 33pf caps in series at each pin to GND. This gives 16pF each.
> > I set the low fuse bits to 0xFF, ckopt=1 (unprogrammed). This means 
> > external xtal, 3-8 MHz, xtal osc-slowest rise time (right?).
> 
> Low fuse == 0xFF makes SUT == 0b11. That looks good for CKSEL == 0b1111.
> It's hard to fault.
> 
> Stray capacitance may be increased unless the wiring to the capacitors,
> crystal, and earth are kept very short. However, I wouldn't expect that
> to pull the frequency very far.
> 
> Is the crystal just designed for a different oscillator circuit? e.g.
> need 33 pf or more, or doesn't like the drive level? I'd try CKOPT = 0,
> just for fun, before trying a more friendly crystal. It's unlikely that
> you can pull the current one far enough by making one of the caps a
> trimmer, as we did in the first digital clock for the Ford Falcon, a few
> decades ago. (I remember asking the manufacturer to tweak the crystal,
> to improve the adjustment range in our circuit, because of a smaller
> frequency offset than you have.)
> 

One other remote possibility is that your crystal is spec'd for series
resonance.  However, the difference between series and parallel resonance
is typically much smaller than the error you are seeing.

Galen