Power Consumption in Deep Sleep

I’m trying to cut down power consumption by using deep sleep on the Imp. I’ve shut down the regulator for my sensors, and put the Imp into deep sleep, but still measure anywhere from 0.7mA to 1.0mA as I measure the high-side current. I can see the radio kick in at 40mA+ when it comes out of sleep. I’m using powersave when not in deep sleep. I’m using the SparkFun board which has the encryption chip ATSHA204, and a regulator TPS62172. The current draw on those two parts is very low, and the Imp should be running at about 6uA in deep sleep.

Any tips on how to cut back further? Am I loosing current through some of my input or output pin connections?

Can you disconnect your sensors? Any pull-ups or pull-downs that are against the default state of the pins?

The ATSHA takes nanoamps, and the 62172 takes ~17uA, so you shouldn’t be seeing more than maybe 0.03mA. What voltage is your input?

I’m running with 2 stacked Tadiran 3.6V TL-2150 (non-rechargable Lithium) for about 7.2V. There are 4 of the stacks in parallel for a total of 8 batteries. I find that they run at about 6.8V in the circuit.

The unused pins are open, and the temperature and pressure sensors are analog input on pins 5, 7. The one digital line out, pin 8, controls the 3V sensor regulator.

I’m open to other battery technologies.

BTW, your participation on the forum is impressive. The timely and informative comments are developing a loyal and enthusiastic customer base. Thank you!

Whoah, that’s a significant amount of power :slight_smile:

Just generally, you should be careful about connecting any cells in parallel - especially lithiums. You can easily get into imbalance situations where a weak cell will end up being charged by the other cells and with lithium cells this can be… quite pyrotechnic.

Alternatively, the April board will run with 4 of these in series very efficiently (due to the DCDC) and then you could use a 3.0v LDO to drop from the imp’s 3.3v buck.

Sanity check: if you remove the imp card, how much current do you see? This should be incredibly close to the current in sleep.

As for participation… we learn a lot from the forums. Besides, having been forbidden to talk to customers whilst I was at Apple, it’s nice :slight_smile:

I unplugged the Imp as you suggested, and didn’t see any change! After probing some voltages, I realized that the 3.0V regulator for the sensors wasn’t shutting down. The regulator enable pin is internally pulled high, and I was using the Imp digital out to control it. Guess what happens when the Imp goes into deep sleep? The pins shut down to avoid drawing any current during deep sleep. That allowed the enable to float back high, turning on the 3.0V regulator!

The problem appears to be solved by adding a pull-down resistor on the enable line. The power consumption is way down now. I’ll continue testing.

I’m also reviewing the battery configuration.

Thanks for your help!

Is the pull down really needed on an April board to get below 1ma?
Can someone please verify?

Hi sbright33! The pull-down is for an off-board regulator, not the on-board one. There is no issue with the April regulator.

I am enabling it using one of the Imp i/o lines to turn on power to sensors upon waking from deep sleep.

When in deep sleep, the i/o pins are disabled, so make sure any control signals float to the correct levels.

BTW, I’m using the Adafruit IN219 DC current sensor to monitor power going to my Imp system. The IN219 is hooked up to an Arduino Uno board to monitor and display the readings.

Thanks for the clarification. I know you mentioned the Sparkfun board at the top.

What if I do both of these at the same time?
setpowersave(true) and wake()
How much current does that use on average?

When in setpowersave(true) mode, you’re still totally awake it’s just wifi is in IEEE PS mode. Depending on the peripherals you have in use, you can easily get down to ~7mA, though this will be going lower in best case scenarios in the future.

I presume you mean imp.wakeup()? These just continue to work normally.

Sorry for the typo yes I meant imp.wakeup. So if I use both, the total current will be about 7ma during sleep?

Yes, depending on peripheral usage.

Thanks for answering so quickly and thoroughly!