AVR-Chat

At 10:30 AM 7/4/2005, Paul Maddox wrote:
>Dave,
>
> > I don't, because of space,
>
>doesn't take up much more than a crystal and two caps.

The smallest I've seen are square, about 400 mils

>Huh?
>i would've thought the output of tehse would've cleaner and better defined
>that a crystal with caps.

They are capable or relatively high output current, and that alone 
makes for more EMI.
The more the output departs from a sine wave, the more odd harmonic 
content you have to contend with.
Plus you have their power feed to consider.
To limit EMI, a resistor in the feed to each device is a good idea, 
and now you have to think about multiple feeds with resistors, or a 
single feed with taps, and what do you do about the reflection off 
the end of that feed? Even at 10-20 MHz, you're doing RF design now, 
and you have to think in those terms.
Finally, you need a low Z path between the ground of the module, and 
whatever it feeds, so that the return current has a definite path.

Individual crystals, for six clients, is certainly more board area 
than an oscillator, but you don't have all the distribution issues to 
contend with, nor the track space.

>Unlike having to ensure a good screen around the crystal when it 
>comes to laying out a PCB?

I don't actually do that much in that regard.
I take the caps back to the uP ground by a discrete track, and I take 
the crystal leads back to the uP in parallel.
I have done small discrete ground zones, but I haven't seen much 
advantage from it.
I don't run other signals through that area though.


> > Driving multiple chips is one advantage though.
>
>yep, I used it on one board to drive 6 AVRs, very handy..
>
>In fairness, both have advatnages/disadvantages..

Engineering is the art of compromise :)