Seeking battery performance data in ultra low-power (uW) applications

Go To Last Post
20 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'm considering a design that would be activated by a capacitive touch switch.  The switch circuitry would be the only thing drawing power most of the time at 34uA and 3V.  I'd like to run it off of a couple of AAs, and, of course, I'd like to know how long the batteries would last, but the manufacturer data I've been able to dig up doesn't provide discharge curves for such low power consumption.  Has anybody performed or run across an analysis of power consumption at this very low end of the curve?  Thanks.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Typical AA alkaline cells have about 2700 mAh capacity.

 

For example, the datasheet for Duracell's Coppertop AA cell shows a service life (down to 0.8V) of 650 hours at 5 mA.  That amounts to 3250 mAh.  At 100 mA, that service life drops to 27 hours, or 2700 mAh.

 

If you consider the service life to end at 1.2V, a 5 mA rate gets you 450 hours, or 2250 mAh, and a 100 mA rate gets you 17 hours, or 1700 mAh.

 

Assuming a minium voltage of 1.8V (AVR minimum) from two AA cells, the service life ends at 0.9V.  At 5 mA, that gets you 630 hours, or 3150 mAh.

 

At 34 uA you might get more than that, but you'll be approaching the self-discharge rate of the design.  Ignoring self-discharge, 3150 mAh at 34 uA will deplete the cell after 10.5 years.  Coppertop cells are marketed with a 10-year shelf life, although it's hard to find out what 'shelf life' signifies.  For some long-shelf-life rechargeables, the 'shelf life' indicates the period after which 80% capacity is still available.  If that's so, this corresponds to a self-discharge rate in the neighbourhood of 7 uA.

 

Of course, much will depend on factors like temperature, peak loads, and other factors.

 

How long are you hoping to run on two AA cells?

"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]

 

Last Edited: Wed. Jan 23, 2019 - 11:10 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks.  A couple of years would be plenty.  Looks like this would get there.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

lautman wrote:
A couple of years would be plenty. Looks like this would get there
Well, I didn't say that ;-)

 

If your average draw is 34 uA, and if your peak loads aren't too many order of magnitude above that, and if you're running at room temperature with normal humidity, and if you don't use inferior brands of batteries, and if your application draw doesn't change for the worse with dropping Vcc, and... and... and...

 

 

"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]

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

joeymorin wrote:
Coppertop cells are marketed with a 10-year shelf life, ...
20y for Li/FeS2

joeymorin wrote:
... although it's hard to find out what 'shelf life' signifies.
Might be in the cell's application guide.

Energizer - Cylindrical Primary Lithium Handbook and Application Manual

Lithium/Iron Disulfide (Li/FeS2)

(page 14)

Shelf Life:

(last paragraph)

[methods : 1. elevated temperature storage, 2. microcalorimetry]

Energizer has tested LiFeS2 cells using all of these methods.

Energizer L91 - Ultimate Lithium (mid page 2 : log-linear service hours versus constant discharge current)

via Energizer Technical Information

 


Hardware and Firmware Issues in Using Ultra-Low Power MCUs

...

 

4 - Leaks and Drains

http://www.ganssle.com/reports/ultra-low-power-design.html#leaksanddrains

...

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The analysis by joeymorin seems fair.

 

I'd be concerned about batteries leaking...

 

The switch circuitry would be the only thing drawing power most of the time at 34 uA and 3V

 

34 uA is a tiny draw.  Are you sure that is correct?

 

Working up a profile for when the unit is active and current draw should go into your model.

 

Expecting the rated capacity of the battery is reasonable.  

The 80% capacity for a ten year battery seems about right.

This article touches on the subject.

https://tadiranbatteries.de/pdf/...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

lautman wrote:
I'm considering a design that would be activated by a capacitive touch switch.
The parts count might be reduced by one by the MCU doing touch; mega328PB has a PTC and Atmel QTouch for a range of Atmel MCU.

Microchip's acquisition of Atmel had one of several partial redundancies (touch controllers)

 

Software Library Support - Capacitive Touch Example Codes - Developer Help

http://asf.atmel.com/docs/latest/search.html?search=touch

ATmega328PB - 8-bit AVR Microcontrollers - Microcontrollers and Processors - Microcontrollers and Processors

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 1

Under a few mA, the discharge curve is pretty much independent of current, WHEN YOU SCALE THE GRAPH FOR CURRENT. So, if the graph is for, lets say, a constant current of 10mA and extends to 200 hours,. then for 1mA, just scale the time axis by 10ma/1ma = 10. Its that easy.

 

At higher currents, loss due to the inherent source resistance of the battery changes the shape of the cure a bit, so this scaling is not so accurate. But, under (approximately) 10mA, you can scale the time axis.

 

It is quite a bit harder to determine what your REAL average current is. Multimeters do not average that well over long intervals. I am now building a current measuring board built around supercaps. These are used because the discharge curve is highly predictable, once you know the actual capacitance. You can run your device for an hour or 10 hours or 10 days, depending on the average current. By measuring the voltage drop, you can then determine the average current over that time interval. May have some of these boards for sale (at my cost) soon.

 

Jim

 

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

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Jim, if you ever are going to sell them, make it public or let me know.

We might want to buy 1 or 2....

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


Just had an idea for power measurement.

If you build a Current to voltage converter with an OpAmp and feed in a reference voltage, you can measure the voltage over the resistor for current consumption without having the voltage droop at higher currents (within reasons). If you want to combine this with a Joule counter by starting with a bucket of Electrons in some super caps, then add a MOSfet to the output of the Opamp and connect that to your bucket of electrons in such a way that they can only go to your target processor (I assumed Electrons have a positive charge here for ease of mind).

(You can also turn it around and put the resistor in the GND trace if you have a negative supply available for the Opamp).

 

Some of the digital scopes have math functions to integrate over time and this is a handy method to count electrons with your Oscilloscope.

 

Lot's of Supercaps have a 2V7 max limit. You can lift the negative side of the super Cap to the output of the reference voltage and then drain it much further than you could otherwise.

(Is it within spec to drain such super caps to -0.5V? That would give you 18% extra lifetime)

 

AA's are indeed around 2Ah as a ballpark figure, and an AVR can sleep for quite some time on 2 of those, but then it can not fully drain them.

Another name for AA is 14500 ( = 14mm diameter, 50.0 mm long) and you can get those in Lithium variants, which have a much higher voltage and you can run your AVR from one of those or put two parralell for twice the lifetime, or use UPS techniques to automatically switch to the second if the fist has been drained.

Doing magic with a USD 7 Logic Analyser: https://www.avrfreaks.net/comment/2421756#comment-2421756

Bunch of old projects with AVR's: http://www.hoevendesign.com

Last Edited: Thu. Jan 24, 2019 - 07:15 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The problem I have run into with simple current sense resistor is one of dynamic range. Idle currents are in the sub-microAmp range with peak currents in the 10s of mA in my gizmo. If you make the resistor big enough to measure idle, then the voltage drop is so large during peaks that the  thing quits operating. 

 

But, then if the resistor is made smaller, the peaks happen so infrequently (every 250ms to 1 second or so), that you cannot tell what a meter reads. My quasi-oscilloscope (Saleae logic analyzer in analog mode) does not have sufficient resolution (without differential input) to give any meaningful results. 

 

Hence, the super-cap solution I described, above. Will describe how to get boards in a few days.

 

Jim

 

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

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

With a simple current sense resistor you can only afford a few hundred mV over your sense resistor.

In the circuit from #11 the voltage drop over the sense resistor is compensated by the opamp.

Therefore you can have more than 25V drop over your sense resistor (Limited by Opamp supply voltage).

(You will need a buffer cap on the output, depending on the slew rate of the opamp).

 

You can also use a series combination of a resistor and diode parralell over the sense resistor.

This will make an automatic auto range function.

You have a lineair range as long as the voltage over the sense resistor is below the diode threshold.

Then you have a quasi logarithmic part.

Then you have a part wich is almost lineair and you can measure the voltage drop over the other resistor because it is dominant in current delivery.

(Or measure both and add the currents).

 

If you use a LED  you can extend the first linear range to around 1.5V before the LED lights up.

Then you also have visual feedback: The led turns on whenever the uC wakes up.

 

For higer dynamic range an 8 1/2 digit multimeter would be nice :)

For EUR 20 you can build an approximation of that with an ADS1220 and a uC with display.

On paper it has amazing capabilities. This 24bit ADC has around 20 bits real resolution, built in reference and programmable PGA.

 

There are some projects on the 'net with an HX711 used to measure voltage drops over shunt resistors

(It has an effective range of around 14 bits from a maximum input voltage of 40mV.

But the specifications of an ADS1117 or ADS1220 are so much better that it is well worth the extra few $ for some breakout boards from China.

And when you add a uC, you can combine it with #10 and add lineairisation in software, serial outputs for logging, etc.

Doing magic with a USD 7 Logic Analyser: https://www.avrfreaks.net/comment/2421756#comment-2421756

Bunch of old projects with AVR's: http://www.hoevendesign.com

Last Edited: Sat. Jan 26, 2019 - 03:05 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Paulvdh wrote:
And when you add a uC, you can combine it with #10 and add lineairisation in software, serial outputs for logging, etc.

CurrentRanger: auto-ranging current meter | LowPowerLab

(below the features bullet list)

It was a bit ambitious and immediately became obvious that this needs to be digitally controlled by a microcontroller to do all that.

...

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:
I am now building a current measuring board built around supercaps.

http://www.ganssle.com/tem/tem354.html#toolsandtips

...

 

Neal Somos is using supercapacitors to measure current of a circuit:

...

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yes, that bit in Ganssle's newsletter inspired the board I am building. Will have a couple available in a few weeks.

 

What is nice about it is that you only need a moderately good digital volt meter. You can use it both for averaging very infrequent high current events AND you can use it to determine the actual input current of a switch mode power supply over a range of input voltages (such as from a discharging battery). The test fixture will have a maximum working voltage of (about) 6.5V. It includes a 1% calibrating resistor so that you can determine the actual capacitance of each super capacitor in the fixture (space is available for up to 3). Actually, there is rather little to the fixture; it simply replaces a jumble of wires and "iffy" connections on the workbench so that you can get reliable, repeatable, measurements.

 

It will be available as a bare board, or fully assembled. A manual provides assembly information (including a detailed BOM) and the details on how to use it. Cost is to be determined. I suspect that anyone on Freaks will be able to assemble one - its easy through-hole construction.

 

Jim

 

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

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Seems like the cap idea is pretty good.  Even better, assuming the cap itself is very low loss...maybe a small bank of Teflon types that can run the device for 5 seconds; let the cap take the peak loads --then you can use a higher ohm resistor (say 10k for a 20uA avg current) on the input side to log average current over a long time (days, weeks).  Surges are handled by the cap bank & the voltage drop on the input resistor (from power supply) stays "flat".   This also keeps your device supplied with a relatively constant voltage (say 5V) during the entire test month.  The only droop is temporary (during the surge) & the amount depends on the cap bank storage.  The charge remaining in the cap should be accounted for at the end of the test (charge bookkeeping), if the cap bank is large.

 

Also, 34uA seems pretty awful...you can get well under 1uA with some good old cmos gates (of course at high temperatures currents rapidly climb).  I've seen many such circuits that operate in the nanoamp range.  The actual driver switch leakage can become the dominating factor.

 

http://www.learningaboutelectronics.com/Articles/Touch-on-off-circuit-with-a-4011-NAND-gate.php

 

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

Last Edited: Sat. Jan 26, 2019 - 06:49 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The whole idea of running from a capacitor (any cap) is that you measure the net charge removed by measuring the voltage change. So, you DO NOT want it to operate as a constant voltage. The super caps I chose have a DC ESR of 720 milliohms max (AC under 200mohms @ 1KHz) and have a peak surge current rating of 720mA (per each). The maximum voltage rating is 5.5V. You can have up to 4 in parallel, reducing the ESR by a factor of 4 and the increase the surge current by a factor of 4. In this fixture, you can choose to use 1, 2, 3, or 4 1F caps for a specific test. With the much lower AC ESR, it should do well for short-pulse current peaks.

 

The effective leakage current is 5uA or 10uA on these caps (depending on the model used). The fixture not designed for really long term use though super caps  are being touted, more and more, as a battery replacement. They do have some "soak" so for most precise measurements, they will need to be "charged" for several days.

 

Jim

 

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

 

 

Last Edited: Sat. Jan 26, 2019 - 10:31 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The whole idea of running from a capacitor (any cap) is that you measure the net charge removed by measuring the voltage change

Well, if one wants to measure how much current a pulsating amplifier, sleeping scanner, solar calculator, etc circuit draws over a long time (say 2 weeks), it would be nice to keep the supply rather steady, rather than letting it get all droopy.  In this case, large droopy is needed since it is the measurement technique.  My suggestion is to measure the avg current flowing in & very little droop is required (the larger the cap, the lower the droop).   The power supply continues to supply the avg current needed by the circuit, within two tenths of the voltage its designed for.   The current is readily logged via a high ohm resistor, the cap bank can probably be a few 100 uF, made from extremely low-loss caps--so you know the measurement hasn't "lost" many electrons over the long run.  The only disadvantage is probably needing to log the monitored current the entire run, unless there is some overlooked gotcha.  Poly caps are also well rated for high pulse apps (like flashlamps).

 

Bob Pease did some testing that shows poly caps have very low leakage rates too:

https://www.electronicdesign.com/analog/whats-all-capacitor-leakage-stuff-anyhow

 

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

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The circuit I am measuring won't tolerate a series resistor more than a few ohms. Average current is in the range of 50uA to 200uA, depending on the operating mode.

 

In my fixture, you have a choice of 1 to 4 1F caps. This way, you can choose a capacitor value to give a tolerable voltage drop that is measurable. I would expect measuring times on the order of hours, at most. Lets see:

 

I = C dV/dT

100uA = 1F * dV/dT

dV/dT = 100uV/second

 

So, over 1 hour (3600 seconds) one would expect dV = 100uV/second * 3600 seconds = 360000uV = 360mV.

 

Many DVMs could measure that to 10mV resolution. Thats about 3uA resolution (if my "sums" are correct). You can juggle the number of capacitors and the measurement interval to optimize resolution and/or voltage change.

 

Jim

 

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

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

That seems reasonable, though each supercap might leak a few uA, which could be measured beforehand & removed in the post-calculations.

https://forum.digikey.com/t/understanding-how-to-measure-supercapacitor-leakage-current/2156

Murata shows about 1uA, once fully charged

https://www.murata.com/~/media/webrenewal/products/capacitor/edlc/techguide/electrical/c2m1cxs-053.ashx

 

At 200uA, even a 1k resistor would only give a 0.2V supply drop (which might be significant, or not).  1K & a 50uF poly cap would give a 50ms "reservoir", though it might drain pretty quickly depending on the peak load...so then the supercap might be the winner.

I wonder if you can feed the supercap with a 10meg pot & adjust before connecting the load, so the cap maintains perfect balance (zero drift over a few hours).  Then any voltage drop would be attributed only to the connected load. 

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