Power consumption tuning on Cortex-M0.

[lpcware]

Content originally posted in LPCWare by rkiryanov on Mon Feb 01 11:25:07 MST 2010
Hello.

I think, "battery-powered" microcontrollers should have some options to control power consumptions. For example, it's usefull to add to LPC111x controlled IRC (400kHz-4MHz), "slow" (~8kHz) IRC, "32kHz" crystal oscillator and "fast" (1-12 MHz) crystal oscillator. The main point - "slower oscillator - lower power consumption", divider is not really good thing.

[lpcware]

Content originally posted in LPCWare by rkiryanov on Thu Feb 11 07:28:17 MST 2010
http://focus.ti.com/lit/an/slaa322b/slaa322b.pdf

1.1 The Crystal

For an ultralow-power design, only low-frequency crystals are usable, because with higher-frequency oscillators, the current consumption increases significantly. Tuning-fork crystals typically have a frequency range of 10 kHz to 200 kHz in fundamental mode and a maximum drive level of 1 mW.

[lpcware]

Content originally posted in LPCWare by Gerrit on Thu Feb 04 01:06:20 MST 2010
Gents,
Please check your facts!
The EM part you mentioned looks nice, but you need to look at the complete package. Indeed their power-down current seems low, but at what cost for power-up time? While running what is the power consumption and for how long? The active current of the M0 is (way) lower..

And a last think, try to get hold of the EM parts!!

Cheers!

[lpcware]

Content originally posted in LPCWare by rkiryanov on Wed Feb 03 00:18:01 MST 2010

---

Quote: igorsk
EnergyMicro

---

Thanks :) Now we are using MSP430X, but they are only 20 bits, slow and requires high voltage supply to run at maximum frequency.

[lpcware]

Content originally posted in LPCWare by igorsk on Tue Feb 02 16:25:11 MST 2010
(I hope I won't get banned but) you might want to check out EnergyMicro chips which claim some ridiculous power figures.

[lpcware]

Content originally posted in LPCWare by rkiryanov on Tue Feb 02 10:12:13 MST 2010

---

Quote: NXP_USA
but in general it is better to run somewhat faster and sleep than it is to run the CPU at extreme low speeds like 8k or 32k.

---

We need a 10+ years of work under battery, power consumption should be as low as 2uA (software RTC on 32KHz crystal and rare GPIO interrupts). I think, fast oscillator will consume more than slow oscillator even with sleeping CPU. Of course, I know about sleep mode.

[lpcware]

Content originally posted in LPCWare by NXP_USA on Tue Feb 02 10:00:07 MST 2010

---

Quote: rkiryanov
Hello.

I think, "battery-powered" microcontrollers should have some options to control power consumptions. For example, it's usefull to add to LPC111x controlled IRC (400kHz-4MHz), "slow" (~8kHz) IRC, "32kHz" crystal oscillator and "fast" (1-12 MHz) crystal oscillator. The main point - "slower oscillator - lower power consumption", divider is not really good thing.

---

Hello Rkiryanov,

Besides the main 12 MHz osc (with 1% over temp & voltage) there is a secondary programmable IRC which can run from 500 kHz to 3.4 MHz. Other powersaving features include each digital peripheral has an independent clock gate bit and each analog feature has a powerdown bit. A 32k crystal osc would be nice for timekeeping with an RTC (soon!) but in general it is better to run somewhat faster and sleep than it is to run the CPU at extreme low speeds like 8k or 32k. This is because there are both fixed and variable currents. The fixed currents do not change at low frequencies, but they are removed during powerdown modes resulting in a lower average power. Finally, running the CPU at a higher clock only when it is needed results in faster response then running it more slowly. Of course... you may never need to run it as fast as 48 MHz.

-NXP