Searching \ for '[EE] Battery life testing' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: piclist.com/techref/power/batterys.htm?key=battery
Search entire site for: 'Battery life testing'.

Exact match. Not showing close matches.
PICList Thread
'[EE] Battery life testing'
2005\08\10@153823 by Mike Hord

picon face
I've been kicking around the idea of doing some battery
life testing for awhile now, since I can never seem to
find the information I REALLY want.

What I'd like to do is set up a test rig which allows me
to constant-current load a battery at, say, 100 uA,
1 mA, 10 mA, 100 mA, and just maybe, 1 A, and then
log the voltage on the battery for later analysis.

In the end, I'd like to have a moderate sample size (say,
four to eight cells of AAA through D cell, of each of the
top few manufacturers, plus the same quantity of 9V
batteries, at each discharge rate).  Lots of batteries,
of course, but I really would like to have some hard
data that I could use to give good estimates of runtime
for batteries.

Making the test rig, etc, is easy enough, but I'd like
to then make this info available online, but I don't know
if that'd open me up to legal action on the part of a
battery manufacturer who's displeased with my
findings.  Or, if anyone knows of a site with hard data
(NOT nebulous graphs made under poorly defined
circumstances, which is what I always seem to find),
I'm all ears!

Mike H.

2005\08\10@233945 by Russell McMahon

face
flavicon
face
> I've been kicking around the idea of doing some battery
> life testing for awhile now, since I can never seem to
> find the information I REALLY want.

I can see a cooperative effort aborning.
I too have an intetest in battery life testing for several reasons.
- Camera battery capacities
- El Cheapo battery performances wrt brand name
- Battery impedance under peak loads throughout discharge life.

> What I'd like to do is set up a test rig which allows me
> to constant-current load a battery at, say, 100 uA,
> 1 mA, 10 mA, 100 mA, and just maybe, 1 A, and then
> log the voltage on the battery for later analysis.

I was most interested in time to end point under various discharges,
but curve shape is also of interest.

As well as constant current I'd also be interested in constant energy
or, even worse, constant energy out of a converter. This better models
modern equipment such as (surprise) cameras where the device wants a
certain voltage and current and the converter produces this regardless
of battery voltage. The load need not be an actual converter but needs
to be able to model this. eg it may increases power aat a certain rate
as battery voltage falls to model a converters falling efficiency with
voltage. eg it may draw 100 mA at 6v in and 250 mA at 3v in (rather
than the expected 200 mA).

> In the end, I'd like to have a moderate sample size (say,
> four to eight cells of AAA through D cell, of each of the
> top few manufacturers, plus the same quantity of 9V
> batteries, at each discharge rate).

I'm interested in both primary cells and rechargeables.

Tests I did a while ago showed that certain clone rechargeables are
well down in capacity on claimed values.

> Making the test rig, etc, is easy enough, but I'd like
> to then make this info available online, but I don't know
> if that'd open me up to legal action on the part of a
> battery manufacturer who's displeased with my
> findings.

Could happen. I've seen a number of comparative battery tests
punblished over the years. Being careful to publish only accurate data
and a few disclaimers may be wise. "Just becaise Ev-O-Vac batteries
performed superbly when I tested them doesn't mean ... "

> Or, if anyone knows of a site with hard data
> (NOT nebulous graphs made under poorly defined
> circumstances, which is what I always seem to find),

I've found Eveready data and graphs better than most. Also, Rayovac
did some good stuff on their rechargeable Alkalines which may be
reflected in their other data.

I had in mind controlled constant current source loads with voltages
monitored. I was thinking of using a (treason) ATmega48/88 as that's
what I'm using most at present. Discharge recharge capability would be
nice in my case but not essential.



       Russell McMahon



2005\08\11@050606 by Alan B. Pearce

face picon face
>> Making the test rig, etc, is easy enough, but I'd like
>> to then make this info available online, but I don't know
>> if that'd open me up to legal action on the part of a
>> battery manufacturer who's displeased with my
>> findings.
>
>Could happen. I've seen a number of comparative battery tests
>punblished over the years. Being careful to publish only
>accurate data and a few disclaimers may be wise. "Just becaise
>Ev-O-Vac batteries performed superbly when I tested them doesn't mean ... "

I suspect that part of "covering your backside" would consist of including
any identifiable batch numbering and such info in any data you publish. This
gives both you and the manufacturer an "out" in that you can claim that you
have not tested any other batches in a manner that allows the result to be
extended to any other batch, and it allows the manufacturer to identify any
process problem they may perceive has caused a problem if they think your
results are not what they expect.

For Russells testing with converters, it may be sufficient to use a
switching regulator as the negative impedance load. A small one such as the
LT1676 or LT1776 from Linear Technology would be a good starting point for
up to about 500mA load.

Another style of test you may want to implement is to test the internal
impedance. Do this by having a constant low current load (say 10mA) and
periodically turn on a higher current load (say 100mA, depending on battery
size). Measure the voltage on both low load and high load during the
lifetime of the battery. A switching converter such as those above would be
good for this as they have a shutdown pin, making the load switching easy.

2005\08\11@053529 by Russell McMahon

face
flavicon
face
> For Russells testing with converters, it may be sufficient to use a
> switching regulator as the negative impedance load. A small one such
> as the
> LT1676 or LT1776 from Linear Technology would be a good starting
> point for
> up to about 500mA load.

I thought that I'd simply model the current/voltage curve I wanted the
load to follow and ther load would alter the current drain
accordingly. This also allows you to program a variable curve.
Essentially you want an "interpreter" that you can give a load
performance specification to and it varies the load accordingly. It
makes some sense to do this in a PC program and have the load
controller slaved to it. (Something I was considering and also
suggested by a colleague). This allows far greater flexibility of
control at the expense of tying up a PC - not a great problem in this
age. That said, if you don't mind proghramming the processor per test
configuration you can still work in an HLL and download the code as
required. For a one off this is not a totally terrible solution but
not what you'd want to do if more than a few were being built.

One of my target applications does have batteries with low average
drain with periods of very high current bursts. The terminal voltage
during the bursts is critical.

For the camera application the pattern is no current when off, medium
current when idling (not low!) and higher current bursts for Flash
storage and photo-flash. Even when idling the camera discharges at
about the C/2 - C/4 rate :-(.

It occurred to me this evening that cheating by using an existing
charger as the battery testbed would ease the discharge / charge
cycling design. While this is not toooo hard, the commercial designs
already do a good job of crash charging (around C/1) and you probably
can't easily do this part cheaper yourself. Worst case,  depending on
charger design, I'd need galvanic (!) isolation which is acceptable if
necessary. (Lest unclear, that's a relay fwiw). FET isolation would
probably suffice and even that may not be needed.




       RM

2005\08\11@082917 by David Van Horn

picon face


I've done this to some degree, with some VB software and a PMD 1208LS
USB data acquisition module.

In my case, I was really looking at the battery charger, but the setup
would be much the same. You could use the PMD's analog out to do a
voltage to current sink, to program the discharge current, and you could
also do non-static current drains, simulating loads that draw low
standby, then want large run-time current.

I did some investigation of "memory effect" in NIMH cells using this
rig, with the results here, including printouts from the PMD software:

http://www.dvanhorn.org/NIMH/Index.php



2005\08\11@095205 by Mike Hord

picon face
> >> Making the test rig, etc, is easy enough, but I'd like
> >> to then make this info available online, but I don't know
> >> if that'd open me up to legal action on the part of a
> >> battery manufacturer who's displeased with my
> >> findings.
> >
> >Could happen. I've seen a number of comparative battery tests
> >punblished over the years. Being careful to publish only
> >accurate data and a few disclaimers may be wise. "Just becaise
> >Ev-O-Vac batteries performed superbly when I tested them doesn't mean ... "
>
> I suspect that part of "covering your backside" would consist of including
> any identifiable batch numbering and such info in any data you publish. This
> gives both you and the manufacturer an "out" in that you can claim that you
> have not tested any other batches in a manner that allows the result to be
> extended to any other batch, and it allows the manufacturer to identify any
> process problem they may perceive has caused a problem if they think your
> results are not what they expect.

Good idea.  I'm not going to turn testing and posting the results of battery
quality into a lifelong hobby, I'd just like to have the info for my purposes
and I'm sure many of you would, too.

> For Russells testing with converters, it may be sufficient to use a
> switching regulator as the negative impedance load. A small one such as the
> LT1676 or LT1776 from Linear Technology would be a good starting point for
> up to about 500mA load.

That's not a bad thought, either- no better way to simulate the decreasing
efficiency of a switching supply than to use a switching supply!

> Another style of test you may want to implement is to test the internal
> impedance. Do this by having a constant low current load (say 10mA) and
> periodically turn on a higher current load (say 100mA, depending on battery
> size). Measure the voltage on both low load and high load during the
> lifetime of the battery. A switching converter such as those above would be
> good for this as they have a shutdown pin, making the load switching easy.

My very simple-minded consideration of this whole idea implies to me that
measuring the constant-current discharge would likely be both the easiest
and the best place to start.  Measure constant current discharge, and every
30 seconds (or 1 minute, or 20 minutes, or whatever the appropriate period
is) remove the load and measure the cell voltage with a very high input
impedence ADC.  From there, it's trivial to calculate the internal resistance.

Furthermore, by testing a type of cell at different discharge rates and
graphing out the data, one can easily find the total energy in the cell
(in Joules, not this mAH business), as well as getting a feel for how
discharge rate affects energy available to the load.

I'm not as blasphemous as Russell; I was going to use a PIC18F2320
to monitor the cell voltage and use PWM to drive a current source,
with some kind of feedback to keep the current constant.  I was thinking
of storing the data in flash for later upload to a PC, but a constant
connection to a PC is possible, too (although my PC client program
skills are wanting).  I figure integrating cell voltage over time times
current would give me a reasonable approximation of Joules deliverable
to load at a particular current draw.  Of course, load currents can
be highly variable, but the system I'm thinking of could easily be
made to slowly ramp up the current draw, or provide any other
discharge curve.

Mike H.

2005\08\11@193028 by Russell McMahon

face
flavicon
face
{Quote hidden}

Thos two aims may be less appropriate than it may appear.
You need to decide what you are trying to measure.
If it's available energy in the cell at constant current then the
above is OK. But that is less useful in most applications than other
measures.

Consider, if a load draws constant current, such as eg a linear
regulator supplying a relatively constant load, then time to end point
at a given current is the key measure. This may be modified by various
pulsing or on off duty cycles but the fundamental measure is how long
you can operate the given load for. In which case mAH is what you want
and Joules isn't relevant.

If you are operating into a switching converter then Joules is
relevant as the converter uses all the available energy until endpoint
is reached. But it doesn't draw constant current, and if the converter
efficiency varies as Vin changes, as it usually does quite
significantly in practice, then unless you model the converter
performance with you load you won't get an accurate result.

Moving the load/time description out into a PC allows easy adaptation
of the program to suit a given situation. The result could be table
driven or even (gasp) hard coded each time. The actual load time logic
can be relatively simple and would seldom occupy more than a dozen or
so lines of code. All the talking to the load controller and reporting
back voltages and logging would be standard and all that would be
changed would be the load/time descriptor. This requires more effort
per new situation but offers maximum flexibility.


eg This implements a load that starts at Imin and ramps up as battery
voltage drops

While Vbat > Vmin
   Iload = Imin + Idelta*(Vbat-Vmin)
   Other stuff ...
Loop

All logging and control are independent of this.

This (crudely) implements a change in load current for 1 time unit in
every 10

While  Vbat > Vmin
   Iload = something or other
   If  INT(Trun/10) =  (Trun/10)
       Iload = some other value
   Endif
   Other stuff
Loop

This applies a constant energy load

While Vbat > Vmin
   Iload = Ptarget/Vbat
   Other stuff
Loop

You could do table lookup or a combination of algorithm and table
lookup or ...

Whether this approach is acceptable depends on how expert your
intended user is.





       Russell McMahon







2005\08\12@011017 by William Chops Westfield

face picon face
>> Furthermore, by testing a type of cell at different discharge rates  
>> and
>> graphing out the data, one can easily find the total energy in the  
>> cell
>> (in Joules, not this mAH business), as well as getting a feel for how
>> discharge rate affects energy available to the load.

I haven't been following this discussion very much, so forgive me if  
this
info has already been posted, but "Silverfox" in the candlepower forums
has done some very nice data collection for assorted batteries at  
"higher
than design" discharge rates.  For instance:

http://www.candlepowerforums.com/ubbthreads/showflat.php?
Cat=&Number=744320

(primarily alkalines, with some info on the alkaline-replacement  
lithiums.)

I don't recall whether he ever talked about what his test setup looks  
like.

BillW

2005\08\12@121602 by Mike Hord

picon face
> www.candlepowerforums.com/ubbthreads/showflat.php?
> Cat=&Number=744320
>
> (primarily alkalines, with some info on the alkaline-replacement
> lithiums.)
>
> I don't recall whether he ever talked about what his test setup looks
> like.

Wow.  That's pretty nearly exactly what I was looking for, less of
course hard data being provided.

I still may pursue it for my own edification, however; he also left
out 9V batteries, which I am quite interested in.

Mike H.

2005\08\12@135555 by Mike Hord

picon face
> Thos two aims may be less appropriate than it may appear.
> You need to decide what you are trying to measure.
> If it's available energy in the cell at constant current then the
> above is OK. But that is less useful in most applications than other
> measures.

I guess maybe I'm confused.  Doing a constant load current drain
gives me two pieces of data:
1.  Battery life at that current and
2.  Joules available at that current (with some math involved).

> Consider, if a load draws constant current, such as eg a linear
> regulator supplying a relatively constant load, then time to end point
> at a given current is the key measure. This may be modified by various
> pulsing or on off duty cycles but the fundamental measure is how long
> you can operate the given load for. In which case mAH is what you want
> and Joules isn't relevant.

Okay, so my constant current load tests have given me some idea
as to how many mAH I will get at various loads.  I can fiddle and fart
around with the curves I have (in their simplest form, distance to flat)
and get a good idea of battery life.

> If you are operating into a switching converter then Joules is
> relevant as the converter uses all the available energy until endpoint
> is reached. But it doesn't draw constant current, and if the converter
> efficiency varies as Vin changes, as it usually does quite
> significantly in practice, then unless you model the converter
> performance with you load you won't get an accurate result.

Again, my data provides me with an available Joules measure, from
which I can calculate battery life in certain situations.

> Whether this approach is acceptable depends on how expert your
> intended user is.

Intended user (so far) is you and me.  ;-)  So the users range from
fairly inexpert to far more expert, BUT generally capable of using
a device with fairly powerful capabilities as far as programmable
load simulation goes.

My thought on this matter was that by collecting constant current
data at, say, 5 discharge rates, it would be possible to characterize
the life cycle of a particular "species" of battery in such a way that
it would be possible to calculate its lifespan based upon the load
cycles/fluctuation.  

Mike H.

2005\08\12@152348 by Bob Axtell

face picon face
The problem with using a fixed point as a "needs charging" or "battery
level" meter is that the battery voltage peak point will slowly decline as
the battery ages (as it chemistry changes), and it varies from unit to
unit..

The ideal batterycharge guage is the scheme devised for "smart batteries",
It maintains a history about the battery, and therefore is able to determine
PRECISELY the charge level, battery quality, battery lifetime.

Intel and others developed the specification. Battery control and monitoring
is done by multibus I2C. Available thru Google.

--Bob


Mike Hord wrote:

{Quote hidden}

--
Note: To protect our network,
attachments must be sent to
spam_OUTattachTakeThisOuTspamengineer.cotse.net .
1-866-263-5745 USA/Canada
http://beam.to/azengineer

2005\08\12@190949 by William Chops Westfield

face picon face

On Aug 12, 2005, at 12:23 PM, Bob Axtell wrote:

> The ideal batterycharge guage is the scheme devised for "smart
> batteries",
> It maintains a history about the battery, and therefore is able to
> determine
> PRECISELY the charge level, battery quality, battery lifetime.
>
I've never had much luck with "smart batteries."  The theory is nice,
but
I always seem to end up with systems that say there's 15 minutes left
right up to the point where they say "no battery."  This is based on
mostly on sony info-lithium cells in a camcorder of some significant
vintage; have things improved significantly?  (Alas, when the meter is
accurate, one tends not to notice very much...)

BillW

2005\08\15@171130 by Brooke Clarke

flavicon
face
Hi Mike:

I've done quite a bit of battery load testing over the last couple of
years using a laptop and electronic load.  For some of the latest
results using mostly rechargeable Ni-MH AA cells, but also some primary
AA cells see:
http://www.pacificsites.com/~brooke/5590BA.shtml#RTTbl

In the table the battery capacity is expressed as run time under a
defined load which is typically some radio receive current for 9 minutes
then the transmit current for 1 minute, with the cycle repeated until
the voltage drops below some threshold.  The associated plots give you a
good idea of the battery internal resistance.

Note that the results depend heavily on the quality of the AA battery
holders.  The tests on the above page were done using custom AA holders
made with very much lower resistance than stock battery holders, see the
current capability table to get a feel for that:
www.pacificsites.com/~brooke/5590BA.shtml#CC
I have extras of these holders for sale if you need some.

The test results for rechargeable batteries vary depending on how the
cells were charged, how long they have been sitting, etc.

Have Fun,

Brooke Clarke, N6GCE
--
w/Java http://www.PRC68.com
w/o Java www.pacificsites.com/~brooke/PRC68COM.shtml
http://www.precisionclock.com



Date: Wed, 10 Aug 2005 14:38:22 -0500
From: Mike Hord <.....mike.hordKILLspamspam@spam@gmail.com>
To: "Microcontroller discussion list - Public." <piclistspamKILLspammit.edu>
Subject: [EE] Battery life testing
Message-ID: <.....88eca9220508101238328f546dKILLspamspam.....mail.gmail.com>> Content-Type: text/plain; charset=ISO-8859-1
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: list
Reply-To: "Microcontroller discussion list - Public." <
EraseMEpiclistspam_OUTspamTakeThisOuTmit.edu>
Message: 9

I've been kicking around the idea of doing some battery
life testing for awhile now, since I can never seem to
find the information I REALLY want.

What I'd like to do is set up a test rig which allows me
to constant-current load a battery at, say, 100 uA,
1 mA, 10 mA, 100 mA, and just maybe, 1 A, and then
log the voltage on the battery for later analysis.

In the end, I'd like to have a moderate sample size (say,
four to eight cells of AAA through D cell, of each of the
top few manufacturers, plus the same quantity of 9V
batteries, at each discharge rate).  Lots of batteries,
of course, but I really would like to have some hard
data that I could use to give good estimates of runtime
for batteries.

Making the test rig, etc, is easy enough, but I'd like
to then make this info available online, but I don't know
if that'd open me up to legal action on the part of a
battery manufacturer who's displeased with my
findings.  Or, if anyone knows of a site with hard data
(NOT nebulous graphs made under poorly defined
circumstances, which is what I always seem to find),
I'm all ears!

Mike H.

More... (looser matching)
- Last day of these posts
- In 2005 , 2006 only
- Today
- New search...