## First time working with battery power, am I going down the right path?

21 posts / 0 new
Author
Message

I have to design an unusual device, with a small AVR (not yet selected) at its heart, that runs off 3 D or C cells.  When running the device will draw maybe 200 mA, but it only runs for an hour or so once a week, or as little as once a month.  It has to be battery powered, no chance of being powered off the AC line.  Further, this device needs to be turned on and off with a passing magnet (so, magnetic reed switch).  That means it will be in a sleep mode for most of its life, and I just want to be sure I'm not missing anything obvious.

First a simple calculation.  If I budget for a 10% energy loss in sleep mode vs the run energy, that's 20 mAH over 1 month, or 27 uA allowed in sleep mode.  That looks to be one or two orders of magnitude over the chip current in Power down mode.  I know I have to factor in any other currents as well.

I'm guessing I can use the reed switch to wake up the device via an INT pin, and of course shutting it down via the same pin shouldn't be a problem.

So, what if anything am I missing at this initial stage?  Thanks.

Time keeping?  Still, that should easily fall within your 27 uA budget.  Most AVR can do that in less than 1 uA.

Note that typical alkaline C cells have about 8000 mAh, and D cells have about 12000 mAh.  Most good commercial cells claim a self-discharge rate of between 2 and 3 percent per year, so the self-discharge current of a C cell would be about 18-27uA, and that of a D cell would be 27-41uA.

Note also that, ignoring other currents, a set of C cells could run for at best 40 hours, D cells for 60.

 "Experience is what enables you to recognise a mistake the second time you make it." "Good judgement comes from experience.  Experience comes from bad judgement." "Wisdom is always wont to arrive late, and to be a little approximate on first possession." "When you hear hoofbeats, think horses, not unicorns." "Fast.  Cheap.  Good.  Pick two." "We see a lot of arses on handlebars around here." - [J Ekdahl]

kk6gm wrote:
First a simple calculation.  If I budget for a 10% energy loss in sleep mode vs the run energy, that's ...

Digi-Key

Battery Life Calculator

https://www.digikey.com/en/resources/conversion-calculators/conversion-calculator-battery-life

IIRC, ka7ehk has experience with field use of C primary cells to power a mega328P.

"Dare to be naïve." - Buckminster Fuller

kk6gm wrote:
I have to design an unusual device, with a small AVR (not yet selected) at its heart, that runs off 3 D or C cells.  When running the device will draw maybe 200 mA, but it only runs for an hour or so once a week, or as little as once a month.
What's the device's ambient temperature range?

Reason : capacity is dependent on temperature

An assumption is the cells are OTC alkaline instead of a primary lithium electro-chemistry.

http://data.energizer.com/pdfs/l91.pdf (Li/FeS2, AA, 3.5Ah, 1.5V, 120mR to 240mR, -40C to 60C, 15g, 20y shelf)

http://www.batteryspace.com/canddSizePrimaryLithiumBattery.aspx (lithium thionyl chloride, 3.6V, D, 19Ah, 200mA max, -55C to 85C, 110g)

Horizontal drilling is popular with one of the operators using a drill locator :

https://www.powerstream.com/double-C.htm (replaces 2 series alkaline C cells, lithium thionyl chloride, 3.6V, 17Ah, 200mA, 108g)

"Dare to be naïve." - Buckminster Fuller

Disclaimer:

I've done extremely little with really low power designs...

That said, if the device will be powered down for perhaps a month at a time, why not consider leaving the device off all month, and having the reed switch turn on power to the device.

When it powers up the device then turns on its own power to continue being powered up when the reed switch goes open.

The micro can, also, turn itself off when it is done with its task.

This would save the power consumption from sitting in one of the low power states for a month at a time.

JC

gchapman wrote:

kk6gm wrote:
I have to design an unusual device, with a small AVR (not yet selected) at its heart, that runs off 3 D or C cells.  When running the device will draw maybe 200 mA, but it only runs for an hour or so once a week, or as little as once a month.
What's the device's ambient temperature range?

Reason : capacity is dependent on temperature

An assumption is the cells are OTC alkaline instead of a primary lithium electro-chemistry.

Essentially room temperature

DocJC wrote:

Disclaimer:

I've done extremely little with really low power designs...

That said, if the device will be powered down for perhaps a month at a time, why not consider leaving the device off all month, and having the reed switch turn on power to the device.

When it powers up the device then turns on its own power to continue being powered up when the reed switch goes open.

The micro can, also, turn itself off when it is done with its task.

This would save the power consumption from sitting in one of the low power states for a month at a time.

JC

Interesting.  I've seen this mentioned now and then, but was never in the market so never paid any attention.  So the reed switch would apply power, and then the AVR would turn on a transistor across the reed switch?

Offhand I don't see how another closing of the switch could signal the AVR to turn off (this is the requirement), but it could be that using the ADC with a small resistor to generate two slightly different voltages when the reed switch is open vs when closed would work.  Have to think about this.

An AVR can draw as little as 100 nA in power down, and still be woken up by an external interrupt (pin change, or level).  C cells have a self-discharge rate at least 180 times greater.  I don't think you need to sweat sleep current, unless there are other components which will need to stay in their own 'sleep' state.  Better would be to control power to those parts via the AVR and one or more FETs.

 "Experience is what enables you to recognise a mistake the second time you make it." "Good judgement comes from experience.  Experience comes from bad judgement." "Wisdom is always wont to arrive late, and to be a little approximate on first possession." "When you hear hoofbeats, think horses, not unicorns." "Fast.  Cheap.  Good.  Pick two." "We see a lot of arses on handlebars around here." - [J Ekdahl]

My name was referenced, above, so I guess I had better chime in.

Rather than guesses about sleep power (maybe 10% of the run power), my strategy is to work on the basis of stored ENERGY. You will want to use one of the very low power switchers, as that will let it run down to a MUCH deeper discharge. My current favorite is TI TPS-63051, It is a very low quiescent current buck-boost switcher that outputs 3.3V (on the fixed output version) or 2.5V to 5.5V on the adjustable output version. Input voltage range is 2.5V to 5.5V and it will output up to 500mA. Quiescent input is 10uA. Takes very few parts and it "just works". 3X C or D batteries stacked starts out around 4.7V and you can discharge down to 2.5V (That is 0.83V per cell - I challenge you to do that any other way).

With a varying input voltage, you really need to work on the basis of stored (and consumed) energy. You don't say what the circuit operating voltage is, and that is critical in estimating life. IF it is 5V, then your 200ma is 1W and the energy use is 3600 Joules per hour. 25uA of sleep current at 5V makes 125uW or 450mJ per hour. A single C cell contains 1800mA-Hr * 1.2V (nominal) or 7700 Joules (1.8A-Hr * 3600 sec/hr * 1.2V). 3, combined, gives you 23000 Joules stored energy. THAT is what you have to work with. IF (big IF) you work with a switcher. If you go "bare" or use a linear regulator, then the figuring is a bit different.

Jim

Until Black Lives Matter, we do not have "All Lives Matter"!

Last Edited: Thu. Sep 21, 2017 - 05:15 AM

I should have added the acceptable Vcc range: from 3 cells fully charged (maybe 4.8V) down to 3.5V.  No regulation required.

If your acceptable range is only down to 3.5V for a stack of 3, you will probably get only 80% of the rated energy, maybe less, out of the cells.

Jim

Until Black Lives Matter, we do not have "All Lives Matter"!

If your acceptable range is only down to 3.5V for a stack of 3, you will probably get only 80% of the rated energy, maybe less, out of the cells.

Possibly, but then there are losses associated with a switcher as well.  The TPS-63051 you quote has a spec'd 95% efficiency at VIN=3.6V, VOUT=3.3V, IOUT=200mA, but that drops to 80% at 0.5 mA.  The OP mentioned 200 mA, but unless that is sustained it may not make sense to go with a switcher.

 "Experience is what enables you to recognise a mistake the second time you make it." "Good judgement comes from experience.  Experience comes from bad judgement." "Wisdom is always wont to arrive late, and to be a little approximate on first possession." "When you hear hoofbeats, think horses, not unicorns." "Fast.  Cheap.  Good.  Pick two." "We see a lot of arses on handlebars around here." - [J Ekdahl]

joeymorin wrote:
The OP mentioned 200 mA, but unless that is sustained it may not make sense to go with a switcher.
Indeed, an "interesting" combination of criteria.

IME we strive to run battery apps without a regulator.  But certainly there may be apps where not suitable (as perhaps some circuitry needs a known voltage).

200mA is getting up into "flashlight" territory. ;)  If driving some sort of actuator -- for the sake of discussion, an H-bridge -- then run that off the "raw" and the regulated load would be much smaller and much more predictable.

kk6gm wrote:
If I budget for a 10% energy loss in sleep mode vs the run energy, that's 20 mAH over 1 month

I'm confused about that.  An AVR in deep sleep mode will draw a few uA.  Not mA.  In fact, IME when the numbers are run the "shelf life" loss of a battery pack is usually more than AVR deep sleep current.  That's when I quit trying to find anymore uA.

Let's noodle with a typical alkaline D cell.  It has the ooomph to do 200mA.  At the expense of [roughly] 1/2 the total capacity versus modest draw...

A lot depends on whether the device really draws the e.g. 200mA continuously doing the "active" period, or only sporadically.

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

theusch wrote:

kk6gm wrote:
If I budget for a 10% energy loss in sleep mode vs the run energy, that's 20 mAH over 1 month

I'm confused about that.  An AVR in deep sleep mode will draw a few uA.  Not mA.  In fact, IME when the numbers are run the "shelf life" loss of a battery pack is usually more than AVR deep sleep current.  That's when I quit trying to find anymore uA.

What I meant was that if I assume 1 hour running 200mA per month, that's 200 mAH/month.  So I was throwing out 10% of that monthly figure as a reasonable 'sleep' loss, which, divided by 744 hours in a month gives a continuous 'sleep' budget of 27uA.

But yes, if these figures are in the low range of the self-discharge battery figures, then, as you say, it becomes meaningless to find any more uA.

Offhand I don't see how another closing of the switch could signal the AVR to turn off (this is the requirement), but it could be that using the ADC with a small resistor to generate two slightly different voltages when the reed switch is open vs when closed would work.  Have to think about this.

When closed, the switch applies power to the micro through a schottky diode & also directly to an input (perhaps adc input, to handle decaying batt voltages)...The micro app output is set as a flip-flop to turn on/off a transistor to keep the power pin energized (such as a pnp  or pchan fet).  Now the ADC  can be used to monitor the opening/closing of the switch for desired actions (with proper debouncing).  Of course the transistor has its own leakage, which must be examined for suitability.       Watch out for resistor dividers that stay active & waste power.

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

If you can live with no power when at "sleep" I would go for that. (like #5).

The main reason is the design of the rest, (when you use 200mA I assume that other things are running).

And sometimes it's hard to keep the leak current at external components down.

One simple way is to have a micro relay over the switch, then the sleep power will be 0.

And then the rest can be designed the most efficient way for when it's running.

Some RTC have a power switch output that can be the enable signal to an LDO, SMPS, or load switch (an advantage is these have current limiting that a PFET doesn't)

Typical RTC current consumption is one microamp with some at about 100nA.

https://para.maximintegrated.com/en/results.mvp?fam=rtc (Features, Kick-Start)

http://abracon.com/Precisiontiming/AB18X5-RTC.pdf

(page 1)

...

Power management features:

- Integrated ~1Ω power switch for off-chip components such as a host MCU

...

IIRC, there's an NXP RTC that switches power but I don't recall its model.

"Dare to be naïve." - Buckminster Fuller

I don't think OP wants an RTC at all...it turns on/off by a magnet

An assumption is the micro is doing something beyond just on/off control---reading a sensor or blinking led, etc

For just "dumb" standby circuits it's hard to beat the old metal-gate CMOS parts CD4027  flip-flop  0.02 microamps   (20 nanoamps )  quiescent typ....much worse over temp...30uA at 85 Celsius, max

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

avrcandies wrote:
I don't think OP wants an RTC at all...it turns on/off by a magnet
Some RTC have a switch input (active pull-up) or a logic-level input (wake)

avrcandies wrote:
For just "dumb" standby circuits it's hard to beat the old metal-gate CMOS parts CD4027  flip-flop  0.02 microamps   (20 nanoamps )  quiescent typ. ...
Can't improve on that with the pushbutton controllers as these are about two orders of magnitude greater in current.

Maxim Integrated

Application Note 4444

Single-Pushbutton ON/OFF Power Control

https://www.maximintegrated.com/en/app-notes/index.mvp/id/4444

Abstract: This application note presents a single-pushbutton power-control circuit. The design consists of an ON/OFF control circuit comprised of a pushbutton, debouncer, and flip-flop. This circuit toggles the power output voltage by controlling an LDO. The design features the MAX6816 debouncer and MAX6484 LDO.

...

(debounce, D FF, LDO)

...

http://www.linear.com/products/Pushbutton_Controllers

"Dare to be naïve." - Buckminster Fuller

check out TPS22860 load switch ...a few nano amps leakage!!

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

... at 25C and 50nA max at 85C.

Finding low leakage FETs is difficult; common PFETs have a max leakage of 100nA at 25C.

A TI Nanotimer can be used to one-shot a PFET for up to two hours :

ONE_SHOT is active low, Rext for up to 2 hours duration

from

Texas Instruments

TPL5110 (ACTIVE)

Ultra Low Power Timer with MOS driver and MOSFET Power ON

...

4 Simplified Application Schematic

http://www.ti.com/product/TPL5110/datasheet/abstract#SNAS6501062

via

http://www.ti.com/lsds/ti/clock-and-timing/nanotimers-products.page#p2192=One-Shot%20Feature

Edit : via

"Dare to be naïve." - Buckminster Fuller

Last Edited: Sun. Sep 24, 2017 - 03:22 AM