Being the smart consumer he is, [Denis] usually looks at the price per pound when comparing similar products at the grocery store. When it came time to buy a few AA batteries, he didn’t have any data to go on. To solve his little conundrum, [Denis] decided he would test several brands of batteries and see which one gives him the most bang for the buck.
After bringing home a haul of a dozen different brands of AA cells, [Denis] broke out the Arduino and starting designing a circuit. To test how much energy each brand provides, the Arduino measures the voltage across a load every second until the battery reaches 0.2V. The elapsed time, as well as the voltage, Watt hours, Joules, and ambient temperature are logged on an attached LCD screen and sent over a USB serial link to automate the data collection process.
What’s the verdict? Unsurprisingly, words like ‘super,’ ‘max,’ and ‘ultra’ didn’t connotate a better battery. The best bang for the buck came from an off-brand called RS Power Ultra. The worst battery was the Panasonic Evolta cells that came in at about $1.50 USD per watt-hour.
If you’d like to verify [Denis]’ work, all the code is up on Github along with the schematic.
The question is: did he test multiple? Just testing one a piece really doesn’t show significant proof. (I think this way because I’m taking Stats 121 at college right now.)
No, FTA, he only ran a singles cell for each:
In each case, I took 1 battery from the pack and measured its capacity. Obviously, it would be better to take a selection of each and average the results but I didn’t want to waste all my batteries on the test."My though when I read the criticism was you should have burned more batteries 4 batteries someone would say you should have test 8,16 and so on, no way you can win. Perhaps you will start receiving money to do more comprehensive tests ;) Anyway thanks for posting your projects.
To satisfy your stat-biased brain:
If the n of the test is limited based on the cash at hand, he could take a look and see if the different run-times really are statistically significant
I too did a test on ‘AA’ cells. Mine was mechanical in nature though. I also tested four cells of each brand. Check out my results.
It’d also be interesting to see at what point the batteries go below some threshold voltage above 0.2V. 1.0-1.2V is a popular range where stuff stops working with the batteries, though I think I’ve got some down to 0.6v and still working in some remotes.
‘Flat’ is generally considered to be 0.9v.
given for that load they all perform about the same, then you just buy the cheapest.
It would be nice to see this for the non-alkaline 1.5 volt batteries on the shelf as well.
I’m looking at those zinc-carbon batteries that will take just 1 or 2 photos on my digital camera before being flat…
I’ve been very curious about this as well, and it’s long been an idea on my “future projects list”. Now I guess I don’t have to do it.
I’d like to see Rayovac (seen in all the US big boxes) and AC Delco (found in Fry’s) tested as well.
I kinda want to try building my own tester, with a few controls to set a constant-current discharge rate, so you can compare how batteries perform under various loads.
Very nice!
Different loads tend to produce different results, depending on battery type:
Light loads that are continuous (mp3 player)
Heavy intermitant loads (camera, esp with flash)
Heavy continuous loads (flashlights, r/c plane)
It would be useful and interesting to compare the performance under different loads.
A lot of devices will stop working at a certain voltage point, so that can be critical. For example – if that point is 1.2 volts, then these batteries will all perform the same. It does not matter that the battery keeps chugging below 1.0v.
It would be nice to incorporate timestamps into the graphs and data.
The flashlight folks are very into measuring battery output over time. Their application tends to be fairly high drain, and their testing tends to focus on that type of load. So it may not be applicable to other loads.
Though his graph of voltage over time omits a scale for time, a quick glance at the picture seems to confirm that all batteries are the same in practice. Everything he tests, save for one cell, hangs-in neck and neck until one volt and only then do things start to diverge. In the digital age, next to nothing will operate on a cell discharged beyond one volt anyway, so a cell’s performance beyond that point is pretty much irrelevant.
Nice round up, but it would have been interesting to see how internal resistance of the batteries changes over time too. This could be done by putting a mosfet in series with the load to disconnect it briefly and take a second voltage reading.
It might be that the Panasonic Evolta are not nearly so bad with a lighter, better matched load.
@nes I was thinking the same thing. Some batteries are designed to withstand different types of loads. In the same way it’s recommended to use NiMH in some cases and Li in some other cases.
Maybe adding different types of loads (a DC motor, just electronics, passive components) to get a wider analysis.
Great tool! And very well done (finished) as a product.
I agree. Nicely done project.
To follow-up on Ivan’s post, an enhancement might be to switch in different loads using MOSFETs so you could analyze the battery under different load conditions as well as measure the internal resistance change. Generate a family of curves to let one pick the best battery based on a variety of factors.
Of course I’ll bet that over the years — or even between manufacturing runs or factories — any particular brand/model changes.
i think in the end its more likely to be chemistry related than load related.
different chemistry affects power density as well as the rate the battery are able to produce electricity. i thinks theres a good chance that after being “depleted” by quick discharge as was the case here some battery will “recuperate” to a certain point. (due to the reaction rate of the chemicals)
i also like to point out that batterys have expiration dates because they self discharge over time so shelflife should definetly be considered in that analysis. i think that because people tend to stick with well know brands those battery are more likly to be “fresh” that odd branded batterys
Agreed.
A way long time ago alkaline batteries almost never leaked. If you suffered a leaking battery it was because you bought some low budget zinc chloride battery. Now you can suffer the problem with alkalines. Then again, 30-40 years ago batteries were made in Michigan and Ohio not China.
My solution has been to pretty much stick with rechargables only. if something doesn’t work well with rechargables, then I build an external battery pack containing a couple of extra cells, then plug it into the device’s DC in. So far I’ve had no issues.
Cool project but what is a “R” and what does it measure? Is it actually a unit of measurement or a mental fumble where the fellow meant “ohm”?
I think he meant ohm, isn’t the Greek letter omega available for HTML use?
In parts of the world other than the USA, resistors are often labelled with an R, instead of the Greek Omega symbol. In other words, 4.7R = 4.7 ohms.
Ω Ω Ω Ω certainly a browser can display the omega, but character maps are a pain in the ass to use. I believe there is a “more professional way to go about it , but I forgot how, and I believe requires remembering some code. This one Ω I simply copied/paste from elsewhere. My guess is one could make text file of special characters one uses most often to simplify finding the ones they need most often without using a character map filled with god knows how many characters.Placing sticky note to remind myself to make my own custom character map.
http://www.w3.org/Math/characters/html/symbol.html
Not all browsers handle these things properly in my experience tho.
OK, where do you get RS Power Ultras?
Here is the link to the RS Power Ultra batteries. http://radionics.rs-online.com/web/p/aa-batteries/6656198/
RS is a chain on the British Isles, not to be confused with Radio Shack (U.S.A.) IIRC, RS stands for Radio Spares.
I used to believe that Duracell were the best. They always seemed to give a bit more power at the end.
My BIG complaint about Duracell is that they leak the electrolyte. The batteries never had this leaking problem 10 years ago. Now Duracells leak even if they are sitting on the shelf.
Have others seen this problem?
I also found from the Inet that you can use vinegar to clean up the leak from surfaces.
I’ve noticed this, too, though the problem is not confined just to Duracells.
It seems as though, in the 1970’s, battery manufacturers made a big to-do about “leak-proofing” batteries. I’d say they had the problem pretty much licked. Now, they’ve taken a big step backwards.
I have had more equipment ruined from leaking batteries in the last 5 years than in the previous 20. Like you, I have observed swelling and electrolyte leakage even in the case of unused batteries still in their retail packaging.
Last month I lost two expensive Mag lights. The cells inside swelled, leaked and corroded inside the barrel of the flashlights. The bonding was so complete that I couldn’t drive the old cells out even with an iron rod and a hand-sledge. The barrel of the flashlight yielded before the cells did.
I wonder what stake the battery companies hold in companies manufacturing battery-consuming devices?
Sorry to hear that PI. I saved a Mag-Lite that was destined for the trash with the exact problem you had, leaking cells and “stuck” batteries. I soaked the flashlight barrel (minus the bulb, and things that would unscrew) in some vinegar for a day and the cells finally came out.
Took a good bit of cleaning and brushing, but it works now (over 6 years later) and is a decent light. The end cap is a little tight to unscrew, but otherwise in perfect condition.
I’ve had that happen to a pair of Maglites. I was that pissed with the second one that I told them it was ruined and they replaced it.
I can’t believe that the batteries they supply with the actual Maglite burst before their use by date.
Interesting, but most electronics will not work down to 0.2V. If you look at the graph, all but one of the batteries works the same down to 1.0V, which is about where something using the battery will say, “I’m dead”.
I usually just go for the best value per battery, buying no less than 24 at a time. I love the BRICKs of 100 batteries you can find around the holidays. One time I got 100 batteries for $20.
A nice start on battery comparisons, but constant-resistance life ratings are already easily available from battery manufacturers. What is harder to get are lifetimes with constant-current and constant-power loads. Circuits using series-mode linear regulators behave as constant-current loads (until the battery falls below the regulated voltage). Circuits using switch-mode power converters can be approximated as constant power loads. This supply topology is typical of many higher-cost consumer electronics devices such as cell phones, etc. A further complication is that modern designs increasingly do aggressive power management, so they are “asleep” much of the time. The lifetime of batteries in such low duty cycle application is quite different than one sees in continuous load testing.
Quoting the article: “This lets us compare each of the batteries in terms of the cost per Watt-Hour (which is really all you should care about when buying them).”
I couldn’t disagree more.
For intermittent, low-drain devices like remote controls, shelf life is far more important than capacity. As well as whether it expires gracefully, or leaks electrolyte all over the inside of your device; causing a mess at best or damage at worst.
Most other, higher-drain devices, are better served by rechargeables; which are much cheaper in the long run, and have larger capacities as well.
Finally, isn’t it a bit suspicious that the reputed longest-lasting battery came in dead last, and the cheapest battery (by a huge margin, even to the second cheapest) came in first? I’d expect a few minor surprises, or one major surprise. But two major surprises, so completely contrary to expectations? Sounds like a possible testing error to me. That the author didn’t even bother to properly label his time axis certainly doesn’t fill me with confidence.
This was a nice effort, but the data is mostly useless unless you are using a similar resistive load. Battery testing is a bit more complicated – especially when the final usage is unknown. The best cell for a flashlight [which this test came close to doing but I suspect usability is lost long before 0.2v], may not [and probably will not] be the best for another application. When one reads the manufacturer’s technical information [usually a good step to take before one starts to perform testing on a product], it can be seen that two different ‘models’ are not just ‘good’ and ‘better’. Each has a chemistry tweak for a particular type of load. Any two similar electronic devices could ‘like’ different vendors’ batteries for the best performance due to different internal designs.
Since when do panasonic batteries have a reputation for long life? I’ve always viewed them as generic batteries that are thrown into packages so manufacturers can say “batteries included”.
If anything, duracell or energizer are reputed to have long lives, and they scored somewhere in the middle. I don’t see this as being suspicious at all.
Well, http://www.batteryshowdown.com/ basically agrees with his results. Panasonic Evolta appear to suck massively.
I’m thinking of building something to test my nicd and nimh cells. As they age, they don’t hold as much charge. So, I want a way to weed out the dying ones, and give them easier tasks. So how long they can remain above 1.0 volt under the same load is more important to me.
Very interesting. Although the graph shows that they are basically all the same between 1.5V and 1.2V which is where I usually toss them for recycling. So I’ve learned 2 important things: Don’t trust in ultras, max or extras, and go for the cheaper!
I wish they would have tried this with lithium batteries. He didnt include it in the graph, but the one lithium battery he tested came in 2nd place cost wise, and had double the capacity of the alkaline batteries (4.17Wh vs 2.3Wh). Since those RS batteries dont appear to be readily available in the US, it would seem that lithiums are still US readers best bet.
lithiums were also significantly lighter, 14g vs 24g.
The picture at the top of the report includes a Tesco Value battery. The test mentions no results of this battery, which is a shame, because it would be interesting to see how it compares!
Good work though!
He reports on it in the comments, it performed poorly, it was neither alkaline nor lithium, it was zinc-something.
Nice project. For those of us with a wish list, we just have to make the thing ourself and fulfill our own wishes. I have bought some odd brand batteries that wouldn’t be worthy of testing. Because there where dead cells in the package I wouldn’t bebuying them again. I buy Rayovac at wall mart and the seem to perform well, as well as the off brand I was buying there as the same price point
What does “off-brand” mean? RS is one of the biggest pro electrical suppliers with a long history in the UK. Those guys used to only sell to registered businesses and had probably the biggest electronics catalog of any electrical supplier. I used to read those as a kid, dreaming up inventions. Of course they wouldn’t sell anything to me back then…
A useful Arduino application! Thumbs up.
Interesting results.
Like Eddie above, I’ve preferred Duracell Coppertops over the others hands down, but a couple years ago, I noticed that my “Annie” loco that would run all weekend on a set of coppertops would only go a couple hours on a set. I figured it had developed a bind or something and was drawing more current, but as I examined it, it was running as smooth as ever and drew the same current it always had. I too have had some leak lately. They’d never done that before.
I wonder if they cheapened the old coppertop so we’d buy the extras.
Thanks for the comparison.
i have a small boost converter that allows me to use three “completely dead” D batteries in series to power a light load at 5 volts
my converter allows me to get almost a half amp @ 5v from 3 volts and 50ma @5v(100ma @2.5v) from 1.5v.
so 0.5 volts each cell is way better then an mp3-gettoblaster, which works down to ~0.9 volt each D batterey (8 in series) with mp3 and ~0.75v with tuner only.
so my “dead” D batteries from my mp3gehtoblaster can charge(edit: top off) my cell and then once “dead-dead” can continue to provide LED light for hours more! using 3/8, then do it again with the other 3 and two left over for only LED light :)
PS: “recycler” = lol
Statistically insignificant sample size.
STOP BUYING THROW-AWAY AAs and switch to Sanyo Eneloop rechargables (sorry for yelling, but this is a hot button for me). Since switching I have not used an alkaline AA cell in over three years. It is literally a crime against nature to use disposable AAs. Now that cells like Eneloop have minimal self discharge, there is no application I can think of, where they are not a better choice. They are better in cameras, flashlights…..even remotes, and clocks. You will also save a ton of money. End of rant.
When I tell people that pretty much all alkaline are the same, they reliably tell me Duracell are the best. I guess they just had the best adverts.
One issue is that alkaline perform differently depending on drain. There might be an issue with the best rate for each battery. Certainly it’s a waste to use alkaline in very low drain devices like remotes and clocks.
I designed something similar, but I used it for testing the capacity of rechargeable batteries. The testing concludes when the voltage drops below a usable threshold, and the load is automatically removed.
Rechargeable-Battery-Capacity-Tester It was interesting to see your graph of the results. Even though some cells lasted significantly longer, many were about a tie at 0.9 volts.
Needs switch in/out resistors for constant load/constant voltage etc.