next-generation PMU discussion

[joerg at openmoko.org]

Am So  20. April 2008 schrieb Werner Almesberger:
> joerg at openmoko.org wrote:
> > with standard battery we get a standby of 1200 / 0.6 = 2000h max. Sounds 
quite 
> > neglectable, but like Andy said, every bit of current counts. It's all 
> > summing up.
> 
> Whoa, that's the MPU going full blast all the time. SETI at MPU, here
> we come ;-)
> 
> If the MPU can sleep every once in a while, we get a low-power mode
> that's about three orders of magnitude more battery-friendly.

So maybe we really just don't care about the full power up consumption of MPU?
Probably it's just not worth the additional effort.


> 
> > We can handle this. Have the MPU at some booster that's to be leveled down 
> > from e.g. 3.0V default to sth like 1.8V, what is task of the MPU to set up 
> > PMU.
> 
> Yeah, with an LDO, we'd just burn the power elsewhere, so it should
> be a switching supply if we really care about power consumption there.
> Not sure if there's anything really nice and small out there, though.
> 
> Maxim have a bunch of boost converters that don't look too bad, e.g.,
> the MAX867EUA+T [1] (3.3V, 3x5mm footprint). Needs external caps,
> inductor, and diode. 
> 
> The tiniest I could find is the TI TPS62700YZFR [2] (3.09V or
> adjustable 1.6x1.5mm), which doesn't even need an external diode. But
> it's not cheap, and it burns 300uA on its own, basically negating any
> gains we could get.
> 
> [1] http://datasheets.maxim-ic.com/en/ds/MAX866-MAX867.pdf
> [2] http://focus.ti.com/lit/ds/symlink/tps62700.pdf
> 
> > All the other chips will have additional FET-low-R switches (one for 
> > each powerline that need it) to protect them from OV

Missunderstanding: I'm talking about OverVoltage protection, until PCF50633 is 
settled for correct booster output, not to burn chips. 
No 0Volt

> 
> I don't think the PCF50633 ever pulls strongly to GND. It normally
> uses a weak pull down if the regulator is off, though.
> 
> > Some VoltageDetection for PowerGood if we can spend.
> 
> So far, we got away with what's inside the PMU :-)

Yea but hat was original concern about bad defaults of PCF50633, so it's good 
to have a MPU...


/j