DCO vs Crystal Power comparison

Wilberto Lopez said:

In terms of power efficiency,  how does it compare to use the internal dco vs using a crystal oscillator for sourcing MCLK?

Driving a crystal generally requires much more power than letting the DCO oscillate. It does, however, provide a way more precise frequency and much lesser temperature drift.

Example: on 1MHz, the crystal takes ~180µA(calculated down), the DCO ~80µA (LPM2-LPM0 current)

Also, it is a good idea to keep the crystal on all the time (even in LPM) because it requires some time to settle. The DCO can be switched off and on with only a few µs settling time.

But if you need precision, then there is no DCO vs. XT question. DCO is disqualified form the beginning :)
(well using FLL and REFO, precision of the DCO can be significantly increased, bu tis still no match to a crystal and also introduces some jitter to the generated clock frequency)