testing rechargeable batteries

[gengis]

The idea isn't mine, I'm just passing it along.

I'm running my time-lapse cameras on AA rechargeables. Problem is that the pedigree/condition of the batteries is unknown.

Awhile back someone had the idea to power a quartz Analog Clock from batteries while loading the batteries with a known load. The length of time the clock runs while loaded gives a very good indication of battery condition. So I finally got around to putting it together with a pair of clock costing $3 each. I mounted the clocks to a piece of wood, added battery holders and a 5 ohm load resistor that I can switch in to load the batteries.

To connect to the clocks I cut a piece of wood the size of an AA cell and soldered wires to a few small pieces of tin. The wood holds the tin against the clock battery terminals - so I don't have to solder wires to it.

Works like a champ. I should have done this a long time ago. I was about to go out and get 3 rechargeable D batteries at $18 each to get enough capacity to run the camera for a week. No need. I found I had one battery that lasted 6.5 hours loaded (or about 85% of the rated capacity) and the other two were 3.5 and 2.5 hours respectively. So, basically, I was just running with only one good battery!

I embellished the original idea a bit with the load switch - then I can use the clocks as alarm clocks should the need every arise . . . and I added a terminal strip so I can connect the picaxe to power the clocks directly from outputs - saving time and effort when I need to measure long time delays - just set a spare output to work the clock while the 'axe is sleeping (using some diodes to reduce the voltage to the clocks).

 

[premelec]

Yes! I like this method for its simplicity - a question - do the clocks have a distinct stop operating voltage? The ones I have seem to go down to below a volt... and the mechanical movements vary from unbalanced hands... I just wonder if you got some data on the precision of the method! Thanks...

 

[Peter M]

if you wanted an acurate cutoff point you could always power the clock from the picaxe as suggested above, and actually use an adc to measure the output voltage of the cell under test and switch off the clock via software... of course you would need a stable supply for the picaxe for a consistant reference.

I like the idea though... very simple... and cheap!

 

[gengis]

Premelec: The end comes relatively quickly with NiMH batteries and it is proving to be very repeatable. Two caveats, come to mind - get two of the exact same clocks and mount both at the same vertical angle. Where a particular clock stops depends on the weight of the second hand . . . it gets to a point where the pulses to the motor are still coming but the energy isn't there to swing the second hand past the 9. I've had to replace a clock after the cat knocked it off the table - the replacement has a thinner, shorter, second hand and runs about 3% longer, running repeated tests on one battery. The cutoff voltage is about point eight volts near as I can determine with my power supply.

The charger has four smart independent charging stations - so I'm also looking for differences in the stations (and not finding any) A single battery will run a repeatable amount if it goes from the charger to the clock, waiting a few days will also lose about 5% of the charge.

Of the 5 quartz battery clocks in the house they all have identical movements - ~2-1/8" square X 5/8" thick black plastic housing. The housing snaps apart and can be put back together if you have the patience and inclination - but I don't recommend it. (and mark the position of the hands or the alarm won't be close) presumably machines build these things . . . Has a high degree of "Chinese Difficulty," Lots of little parts held together by the case halves and some swagged plastic boses - built with infinite patience by Chinese women with tiny hands.

The movement is a quartz oscillator with an external crystal. The alarm is mechanical and tied to the hands through gears and an alarm cam. The oscillator chip pulls an open collector low at a high frequency for the piezo sounder. The motor is a two phase stepper the turns a gear with a small donut magnet attached. A one volt pulse appears on the red and black wires to the coil fed alternately like an H bridge - each wire pulses high for 30 milliseconds every two seconds.

The position of the alarm setting may affect repeatability since there's a pair of cam followers that are spring loaded to drop into depressions on a large gear (closing the switch). Seems logical that when the cam followers are riding down the slope of the cam they'd impart some energy to the mechanism - but I just set the alarm so the battery dies before it comes close

Peter: Yeah, I like the idea of a fixed known cutoff voltage too. All too quickly an idea starts to snowball on its own and pretty soon its neither cheap or simple. My gut feeling is increasing the accuracy of the cutoff to a precise value wouldn't appreciably increase the repeatability of the test unless you were to also control the temperature of the battery under test and under charge. It is a chemical system - I'm surprised at how repeatable it is now.

 

[premelec]

Thanks for this detail - you have indeed been thorough! I had observed some of the continued ticks with no movement - However mostly once you've gone low on voltage the cell dropoff is rapid especially when resistor loaded - unlike the clock mechanism which is a light load... Anyhow I'm keeping an eye out for some cheap clocks though I can just use one clock sequentially as I'll discharge at a fairly high rate... The only additional complexity I might try would be constant current discharge rather than the varying current with R - but I think differences would be small for my practical purposes... You've probably boosted cheap quartz clock sales worldwide!

 

[manuka]

This battery clock technique is pretty classic- I recall using it with NiCads over 30 years back. In recent times cheap RS-232 serial data capable DMMs ( such as the esteemed ~US$40 Jaycar QM1538 ) & attached PCs can give much richer insights. The nature of the discharge curve revealed can be very illuminating for a wide variety of electrochemical cell types - plain elapsed time can be too "raw".

FWIW -since this has arisen- in an increasingly portable world, battery assessment/monitoring like this can be very worthwhile, especially for higher end products (laptops,cameras,cell phones,PDAs etc) using Lithium Ion based technology. Off grid users with large banks of deep cycle lead acids charged from solar PV or wind gene. can face $$$$($) to replace their batteries. Yikes! Hybrid car battery banks apparently are a similar woe...

Aside from cheap NiMH AAs (often sold with "inflated" mAh capacities), consumer complaints about the lifetime of more costly Li Ions abound, with endless e-waste (especially laptops) arising just due to failing batteries. Even if they are still available,few users want to replace a battery pack that costs $$$. Presently ~3 years ( as Li Ion age even when stored) or ~300-500 cycles (whichever comes first) is about all you can expect from consumer Li Ions.

Anyway -for skinflints,like myself, content to run them just from the mains (or external batteries- SLA etc), this means quality 3-5 yo. XP capable laptops can be obtained for peanuts. It's the old "½ full, ½ empty" viewpoint I guess.

 

[premelec]

I've salvaged cells from cell phone packs and such so I have no idea how well they'll charge - empiricism is the order of the day. I'd used a comparator and relay to start and stop old synchronous motor mechanical clocks before but somehow had missed single cell testing using the battery operated clock idea so I'm happy to be informed of it even if I'm the last electronic inclined senior to know - or perhaps I should say 'realise'.

I'm thinking of using a voltage divider for Li cells so they will stop the clock but not fully drain the Li cell and use some dioded and constant current drain so the cell doesn't keep draining after the clock stops... anyhow the power supply is both prosaic and fundamental in successful project operation.

 

[gengis]

updated

Thought I'd update the testing results. Turns out that the clocks don't always stop on the 9 the way they were when I originally started testing - they sometimes make it over the hump - but still die on the downward trip. It doesn't look like it skews the testing results any. Even when death is slow it seldom puts more than a minute or two on the clock.

And I've added a new twist to managing the batteries. I had been putting paper stickers on them with the capacity written on the stickers. Problem is, a little handling and the sticker is unreadable. I got one of the cheaper Dymo brand printing label makers - looks like it uses a thermal print head and has a keyboard ($20). That works very well.

I went with the intention of just getting some of the little self adhesive colored dots, planning to use them to indicate capacity using the resistor color code. Cheaper - but not as sanitary as printed labels.

premelec: Like I said, I'd seen the idea some time ago. Then I'd start thinking of ways to make it a more definitive testing system - constant current, voltage monitor, etc. . But with 1.2 volt cells, it almost had to have an external power supply and active load.

Found this for a low drop out current regulator/source:

The LM334N can regulate with not 1.25 V across the sense resistor—but just 64 mV. So a 1.6- resistor will let you source 40 mA quite handily. And the voltage across the load does NOT have 0.7 V in series with it, so it regulates down MUCH better than Figure 1.
http://electronicdesign.com/Articles/Index.cfm?AD=1&ArticleID=4703
http://electronicdesign.com/Files/29/4703/Figure_02.gif

He was using it for a LED current regulator - but it isn't limited to that application.

 

[premelec]

Thanks for additonal info... I make current regulators mostly with an active circuit driving a MOSFET and sense resistor. Now there are MOSFETS with 1 Volt Vgs that could work out simpler - in any case you've got it spot on that it ain't that critical in this application.... The LDOs with 1.2vRef could work by drawing current from the reference pin which goes to current set R to output... I've only done that with LM317s which have much higher [3V+] needs to regulate. I guess if I'm going to test some NiCd 80AH wet cells I have the current element better be hefty to deal with the watts!

 

[premelec]

I'm wondering just what 'clock stop' voltages you are getting - I've looked at some units here and found from .8v to 1.07v end points - I put a capacitor near the movment so it's sure to get plenty of kick for the tick. Thanks...

 

[gengis]

I was getting ~.9-1V but I don't really monitor when the clock stops for most of the tests. The only way I have of telling it just stopped is if when I switch off the load it starts again. Haven't found a battery that met its nameplate capacity yet. Most of what I have are 1800 mah and the good ones drop out at 1600+. A few 300 mah (nameplate 1200) already went in the trash.

I have a clock plugged into the axe measuring time. To drop the 3.6 to 1 volt I used a red led. The clock hangs when the voltage drops low or impedance gets too high - I was seeing the led stay on and the clock stay off. To get around the problem I put a 10 uf cap across the led, then the clock starts normally when the pin goes high and the led flashes once a second. I mention this because it leads me to believe there's a low voltage or high impedance failure mode of the clock that essentially appears as a short to the battery (led stays on). Is that sort of why you include a cap across the battery? Help get the last dregs out?

Where'd you pick up the 80 ah nicads? I had part of a 24 V wet cell when I was a kid. Came out of a diesel locomotive backup comm battery. My OM used half of it and I got the other half. Stainless steel cases and all cells strapped together.

 

[MPep]

Australian magazine Silicon Chip, February 2008, edition has an automated battery discharger shown in the Circuit Notebook. based on a 28X. Also describes how to graph in Excel.

www.siliconchip.com.au.

 

[premelec]

OK we're trying to keep it simple - just rough measure of battery function... thanks for the Silicon Chip ref nevertheless!

My 80 AH wet NiCd cells are in transparent plastic cases and were once diesel starters I think - but I've had 'em many years and haven't tested them... I don't like handling KOH solutions - I actually turned in a bunch of other wet cells from aircraft clusters and such to recycle rather than try new electrolyte... I'm thinking I'd like NiFe cells as I don't much like H2SO4 but that's still cheapest... oh for a good cheap Li cell! They're using 8000 cell clusters of A123 cells in electric autos [www.teslamotors.com]. However if you calculate the cost of the cells they cost about $1 per KWH of energy stored over 1000 cycle lifetime - excluding the energy cost to charge them... Like paying $5 for a one time use 1 liter gas can...

Recently I extracted a large cord rewind spring from an old vacuum cleaner - I'm thinking of winding it up with a PM motor and then letting the motor be run backward to generate just to see what the overall mechanical storage efficiency is - sort of a mechanical capacitor

Energy storage continues to be a problem to solve... conversion of sunlight energy BTUs to biomass BTUs typically is about 1/10 % return on solar investment. Discouraging... I bicycle...

 

[BeanieBots]

Have a look at super caps. Hybrid vehicals are moving towards 'hybrid' batteries combining the rapid/efficient charge/discharge cycle ability of the super cap with the slower higher energy density of chemicals.
Mechanical solutions have been tried in the past. Not very successful due to the extreme speeds required and subsequent dangers of potential (and case study actual) failures.
For lower power devices (eg portable radio), springs have proved viable. Notably the Baylis windup radio. The solar panel could be used to wind up the spring but the additional cost of motor/gear would reduce its market.
Unfortunately, engineering is no longer about producing a 'better' product. It's about beating your competitor to market and selling spares!

 

[premelec]

I've been watching supercaps for several years... I'm too old to compete I'm at the having fun stage! Supercaps are excellent for taking in large current pulses [regenerative braking] and putting out same - but the linear voltage vs. discharge characteristic isn't great - capacitors need to be operated with small voltage changes - 1/2 cv^2 energy considerations... The supercaps are also a bit pricey still - as are the Li batts. I installed 4 zone valves a few years ago [Taco ESP] that use supercaps to shut the valve when power is removed. The main complaint I've had is the motor/gear noise which has nothing to do with the caps. Whatever works... or plays...