Community
cancel
Showing results for 
Search instead for 
Did you mean: 
MPERE18
Novice
1,202 Views

Power consumption in Sleep and DeepSleep

Hi,

I am using POWER example provided in ISSM on Intel Mint Field Fab C Development kit Quark D2000. I succesfully put the micro on sleep and wake up from RTC interrupt, but I cannot wakeup micro from deepsleep. Analog input A5 pin is 3.3V by default, but if I connect A5 to GND before running the example UART does not work.

On the other hand I am checking current consumption in sleep and deep_sleep mode on Intel Mint Field Fab C Development kit and the less current consumption I can get is 11mA, how can I achieve the 3.4uA that datasheet says for sleep with RTC available?

Thanks in advance,

Regards.

12 Replies
Sergio_A_Intel
Employee
112 Views

Hi,

We'll investigate how to wake up the Quark from deep sleep and how to achieve the 3.4uA listed on the datasheet.

Sergio

idata
Community Manager
112 Views

Hello,

I've made some measurements and it seems that to wakeup from deep sleep the A5 pin needs to be pulled down.

I don't have mentioned above UART issue.

The problem is that this pin has ~6kOhm pullup and that alone causes ~500uA current draw.

The lowest current consumption I've achieved after removing this resistor is ~200uA in deep sleep.

Vin 3.3V

~5mA halt

~876uA sleep

~200uA deep sleep

How can I reduce the current further? I'll need at least some hints on what is causing 200uA current in deep sleep.

Help is greatly appreciated.

Best regards,

Mariusz

AdrianBurns

bartosz.boguszewicz

bartosz.firmwave

avijit.M

BZiol
Novice
112 Views

Hi Intel,

Please give us a sequence/solution how to achieve 3.4 uA power consumption with Quark D2000.

My results are below:

Setup:

  • DC power supply DPS-305BM
  • Agilent 34450A
  • Vin = 3.0 V

I've used board only with D2000 and necessary components populated to run Quark. I've removed all peripherals, pull-up and pull-down resistors etc.

I uploaded power mode example program that you delivered also tried to turn-off all unnecessary comparators and stayed with internal crystal and oscillator.

Regards,

Bartosz

AdrianBurns

bartosz.boguszewicz

mariuszr

avijit.M

idata
Community Manager
112 Views

Hi,

I've made some further trial-and-error attempts.

It seems that the example provided in ISSM is configured for the lowest current consumption.

First I thought that the hybrid oscillator (for some reason) does not enter power down in halt mode.

Not enabling the QM_OSC0_PD bit results in increase of deep sleep current from 200uA to 366uA.

Also not activating RAR retention mode results in 200uA -> 277uA increase.

I'm slowly running out of ideas...

Help will be greatly appreciated.

Best regards,

Mariusz

AdrianBurns

bartosz.boguszewicz

bartosz.firmwave

avijit.M

Sergio_A_Intel
Employee
112 Views

We're still working on your case. We'll post back when we have more information. Thank you for your patience.

-Sergio

BZiol
Novice
112 Views

Hi Sergio,

So everything looks good, there is some current leakage on our board and eval board on reset button and user button and there is impact from pull-up and pull-down resistors.

We found solution, Quark looks nice with 2 uA current consumption in deep sleep mode. You need to measure power consumption on J24 header on evaluation board (Fab. D).

Regards,

Bartosz

AdrianBurns

bartosz.boguszewicz

bartosz.firmwave

avijit.M

MPERE18
Novice
112 Views

Hi,

I am also checking current on J24 jumper but I cannot get less than 500uA. Does anynone gets less current on J24 jumper with soc_sleep function?

I have notice that soc_sleep() bsp function sets:

/* Set system clock source to hyb osc, 4 MHz, scaled to 512 kHz. */

clk_sys_set_mode(CLK_SYS_RTC_OSC, CLK_SYS_DIV_8);

I thought that scaling it to 32kHz (clk_sys_set_mode(CLK_SYS_RTC_OSC, CLK_SYS_DIV_128) will reduce power consumption but it did not.

How can I achieve the 3.4uA on sleep mode with RTC and GPIO interrupt available?

 

Thanks in advance,
MPERE18
Novice
112 Views

Sorry, I mean clk_sys_set_mode(CLK_SYS_HYB_OSC, CLK_SYS_DIV_8) and clk_sys_set_mode(CLK_SYS_HYB_OSC, CLK_SYS_DIV_128).

Also I have tried to set system clock to RTC clk_sys_set_mode(CLK_SYS_RTC_OSC, CLK_SYS_DIV_128) and to disable hybrid oscillator before cpu halt (QM_SCSS_CCU->osc0_cfg1 |= QM_OSC0_PD), but I does not reduce consumption.

I will appreciate if someone can clarify how to achieve 3.4uA because many applications won't work with the acutal 500uA that I am getting on sleep mode.

Regards,

BZiol
Novice
112 Views

Hi marperez,

We used power example from Intel QMSI (1.0) package without any changes and it works (I reached 2 uA of current consumption).

How can I achieve the 3.4uA on sleep mode with RTC and GPIO interrupt available?

You can achieve this level of power consumption only in DeepSleep mode.

Regards,

Bartosz

MPERE18
Novice
112 Views

Thanks Bartosz. If I try power example from Intel QMSI 1.0.0 (https://github.com/01org/qmsi/blob/v1.0.0/examples/power/main.c qmsi/main.c at v1.0.0 · 01org/qmsi · GitHub ) the less current I can get is 200uA on my D2000 development kit Mint Field FabC. I notice that on github there are also 1.0.1 and 1.1.0 versions for QMSI and that version 1.1.0 alpha has a big change on several libraries.

I understand you are also using v1.0.0. I do not know why I get this current consumption with the same code.. According to quark D2000 datasheet is possible to get 3.4uA consume on sleep mode with RTC available:

But I cannot achieve this with soc_sleep neither with soc_deep_sleep functions.

Thanks in advanced.

112 Views

Hi,

The power figures in the datasheet

1) refers to the SoC and, meaning that we are not considering any leakage nor any other platform contribution. In order to achieve these numbers you would need to re-work the CRB removing those unnecessary components and trying to reduce the leakage to the minimum

2) it can only be achieved by using Comparator or RTC (the comparator consumes even less power than the RTC and both are in the nA range). While all 25 GPIOS are Always-on interrupt and wake capable, I wouldn't recommend it for very low power application as GPIOs would require a lot more power than Comparator and RTC

Regards,

Alex

Sergio_A_Intel
Employee
112 Views

Hi marperez,

Do you still need help with this thread?

-Sergio