On Apr 8, 2013, at 05:40 , Graham Davies <[email protected]> wrote:

> Rick Mann wrote:
> 
>> ... I'm not sure I can calibrate the internal oscillators via DFLL using an 
>> external crystal [on the XTAL pins] ...
> 
> You can't.  The only calibration sources for the DFLLs are the 32.768 kHz 
> internal oscillator, the 32.768 kHz crystal oscillator using the TOSC pins 
> and, for the 32 MHz DFLL only, the USB start-of-frame signal.
> 
>> There doesn't seem to be a way to drive the RTC from the XTAL pins ...
> 
> No.
> 
>> ... nor can you calibrate the internal 32 kHz clock from an external crystal 
>> (on XTAL1 & 2).
> 
> Yes you can, but you have to do this with a firmware DFLL using the RC32KCAL 
> register.
> 
>> So, I guess the best thing is to just stick a 16 MHz crystal on the XTAL 
>> pins and be done with it? I need USB. If I understand things correctly, I 
>> can PLL that up to 96 MHz, then divide by 2 to get 48 MHz.
> 
> This is correct.  You could use any crystal or resonator that will multiply 
> up to 48 or 96 MHz and meets your frequency accuracy requirements.  However, 
> the maximum CPU clock frequency will be 24 MHz.  You could also generate the 
> USB clock independently of the CPU clock using the DFLL to synchronize the 32 
> MHz clock to the USB start-of-frame signal, which will cause it to run at 48 
> MHz.  Then you have more options available for the CPU clock.  You do not say 
> how accurate you need the CPU clock to be.

Thanks, Graham, it's very helpful to get confirmation.

I only need it to be "accurate enough" to do USB and serial reliably. A future 
project that requires RTC will get a proper crystal for that, but this one 
doesn't. So I'm putting in a MHz crystal and it should be fine.

-- 
Rick




_______________________________________________
AVR-chat mailing list
[email protected]
https://lists.nongnu.org/mailman/listinfo/avr-chat

Reply via email to