Re: weak crystal driving mode (was: Re: [avr-chat] (no subject))

[David VanHorn]

> > Check your CKOPT fuse. You should not run with that fuse in the
> > default state without a lot of research as to wether your crystal will
> > work on that drive.
>
> As I said, I verified that it works reliable, so I can recommend
> using that feature _if_done_right_ to save power, since the
> internal oscillator isn't really precise, even if calibrated
> (and it still depends on temperature).

Yeah, that's what we thought too. Unfortunately, with standard
crystals, it's a major disaster.
IMHO, making the Vittoz mode the default was NUTS, but that's the way it is.
Atmel did a very poor job of making people aware of this problem, and
when I went through it, seemed mostly unaware of the problem.

> > In the default state, the Vittoz oscillator is enabled, which is weak
> > and can cause BIG problems.
>
> Which problems? I don't see any problems mentioned in the Errata Sheet(s).
> Of course, you mustn't drive any other or any big load with the oscillator
> and have a low power (uWatt) capable crystal.
> (BTW: I'm using the ATMega168 right now).

IF you have a crystal designed for working in this mode, then it
works. No argument there.
Atmel USA was not able to refer me to any part number that they
thought would work, and were unable to tell me how to qualify a
crystal for that mode. In fact, they asked me to point them to
anything that I found on those lines.   "Normal" crystals frequently
"work", but on closer examination may sometimes fail to start (if
you're lucky) and may skip cycles causing baud rates to be a bit wrong
(slow), cause emulation problems, or non-deterministic execution.

Driving any external load is not specified regardless of mode.

Been there, done that, had the enormously expensive recall.