Battery Testing

[FIRST JerseyKid]

Does anyone have a good way to test a battery? I know there's a battery load tester, which I think tests for the internal resistance of the battery.

I think, that usually a battery is bad when it has a lot of internal resistance and when it can't hold a 'full' charge anymore, correct me if I'm wrong or missing anything.

Can someone explain to me how the Load Tester tests the battery, I have experience in ac and dc circuits and I think I have an idea of how they do it.

[KenWittlief]

funny you should post this, I just bought a used Marine VHF handheld radio on ebay, and Im sitting here testing NiCad batteries while I type.

A battery tester will only give you some indication of the present charge left in the battery. As the battery charge is used up the internal resistance does increase (as you mentioned). The result is when you hook the battery into a significant load, the voltage across the terminals is lower than it would be if the battery was fully charged.

For non-rechargeable batteries this is about all you can do. Momentarily connect the battery to a significant load, and measure the voltage. The resistance (load) that is significant for a battery depends on its size (capacity).

For rechargeable batteries, the best way to test them is to cycle them (this is what Im doing right now). You charge the battery fully, then you connect a constant load. For example, I am testing a 7.2V 600mAhr NiCad pack. I have it connected to a 10W 10 ohm resistor. I connect it to a DMM, write down the time. When the battery voltage falls to 6V (about 80%) I stop the test and note the elasped time.

The capacity of the battery is the averege current * the time duration. In my case the 600mAhr battery is putting out about 420mAhrs (lasting about 35 minutes). Not bad for an older battery, but not new either.

If you dont have time to sit and check the DMM every few minutes, there are data logging volt meters (like my Keithley 175) that you can setup to record the voltage once a second, or once a minute, or once every 10 minutes.... so you can let the test run and come back later and see how long it took for the battery to drop.

Usually you run the test with the battery loaded at a rate that will discharge it in one hour. For something like the Exide batteries in the FIRST KOP, that would be an 18 Amp load, and the battery should take about an hour to drop 80% of its voltage. (sometimes finding a good load is a challenge. 18A * 12V = 216W. That would be a pretty big resistor (expensive). For large 12V batteries, you can get 12V 50W light bulbs that fit in a regular light socket. They use them in campers and RVs - most RV stores carry them, and some places like Home Depot. Car headlights work well too)

Also, if a battery has not been used much, you might need to cycle it a few times (charge, drain, charge, drain...) to, well sorta wake it up.

[FIRST JerseyKid]

Okay. So my next thing is...

I want to design a board that I can plug battery's in and do a load test on them, take readings using an ADC, feed the ADC to a PIC chip that records the values and does math and outputs to an LCD.

Does it sound like a good idea and are there any special things I should know?

My thought is its mainly a voltage divider with my circuit having a large resistance and the battery having its resistance. The PIC should be able to calculate the resistance of the battery and the capacity...

your thoughts?

[KenWittlief]

so you mean like testing an alkaline battery to see how much energy is left?

Your description is correct if that is what you want to use it for. DuraCell has good spec sheets for their alkaline batteries on their website. You can use the graphs to estimate the energy left in the battery by measureing the voltage with a specific load on a battery.

So depending on whether you are testing a AAA, AA, D cell.... the load or the voltage curve will be different.

Duracell Alkaline website page: http://www.duracell.com/oem/primary/alkaline/alkcapacity.asp

[FIRST JerseyKid]

I want to try to test a battery's capacity and it's internal resistance.

[KenWittlief]

you mean, you want to run it until its dead, so you can say "EverReady AA's have 400mAhr of capacity, and Duracell AAs have 440mAhrs...."

that kind of testing?

also, do you mean rechargeable batteries, or primary cells, like alkalines?

[FIRST JerseyKid]

I want to read its fully charged energy capacity and it's internal resistance.

[KenWittlief]

someone else correct me if Im mistaken, but I dont think you cant take an instantanious reading on a rechargeable battery that will tell you its full-charge capacity.

The only way I know to do that is to run it through a full charge, discharge cycle, like I described in my first post.

For your idea, if you fully charged the battery, and then measured its internal resistance, that might an indication of how worn out it is. I think you would have to do extensive testing to determine the relationship between the internal resistance of a fully charged battery to its remaining capacity. Also the temperature of the battery would probably be a significant factor for your calculations.

And I think you would have to find that relationship for each size and type of battery you want to check: NiCads, NiMH, LiIon... AA, C, AAA...

it sure would be an interesting project!

[Richard Wallace]

West Mountain Radio makes a nice battery analyzer (http://www.westmountainradio.com/CBA_ham.htm) that works the way Ken describes. It came up in an earlier thread (http://www.chiefdelphi.com/forums/showpost.php?p=491818&postcount=6). Several teams have used this analyzer with good results.

[Al Skierkiewicz]

As Richard pointed out the West Mountain tester(CBA-II) will give a graphical display of the discharge characteristic of a battery. It can only discharge the robot batteries at about 7.5 amps but that is enough to show you the state of charge and it gives you the amp-hour rating of the battery. The beauty of these little devices is that for about $100 you get a nice little USB device that will save the data and print battery labels in a 3" cube. West Mountain has a new product (I think at the urging of the FIRST community) that will discharge at much higher currents but runs about $500. The tester will show if individual cells have reduced capacity and has given many teams an indication of a defective battery out of the box. If you are looking for internal resistance measurements there are fairly expensive devices that will do that. I know of a SnapOn product that tests state of charge and internal impedance in a few seconds of testing. It is about the size of a small brief case and the cost is pretty high.

If you are trying to model the discharge of the robot battery in use by somehow comparing or calculating from the amp-hour rating, that will be nearly impossible. There are so many factors that are involved in battery chemistry, discharge rate, peak current, etc. during a match that calculations are fruitless. Wildstang built a current test fixture that we used on several robots during the 2002-2003 seasons. It was able to measure current from several probes on the robot and main battery terminal voltage. The data was ported out through the dashboard port along with time marks provided from the RC so we could correlate current demand against time while watching replays of a the match in which the data was being generated. The data was surprising and we learned a lot about robot current draw but nothing that could lead us into a predictable battery life curve. We could accurately predict that certain robots would fail at some times during a match due to low battery voltage and we could also predict that certain robots would drain the battery by match end. Armed with that info, teams still took no action to correct fatal errors in their design. Thankfully, IFI saw the need to have a backup battery on the RC which didn't correct all problems but reduced the time it took for a robot to recover from a low voltage condition.
One thing to keep in mind at all times, the internal resistance of the battery is 11 milliohms, draw max current (400 amps) from the fully charged battery and the terminal voltage will drop to 7.5 volts. A robot that will draw high current throughout the match will cause that voltage to drop to perhaps as low as 4 volts later in the match when the battery is discharged. 4 volts=robot dead! Take that same battery out of the robot and measure with a DVM and it will read 12 volts, no load, but won't run another match.
Ken, you ought to look into the CBA-II it is a pretty cool device.

[Ben Piecuch]

I'll post this here, and in CD-swap as well... Does any team in the Connecticut area have one of these Battery Analyzers? We have, unfortunately, had some dead battery problems this year, and would love to see how our batteries perform. We're willing to drive anywhere in CT (and maybe even NY and MA) to have our batteries analyzed. Or, if anyone is coming to the Bash at the Beach in October, we could possibly do some testing there.

Please email or PM me here, thank!

BEN

West Mountain Radio makes a nice battery analyzer (http://www.westmountainradio.com/CBA_ham.htm) that works the way Ken describes. It came up in an earlier thread (http://www.chiefdelphi.com/forums/showpost.php?p=491818&postcount=6). Several teams have used this analyzer with good results.

[Richard Wallace]

One thing to keep in mind at all times, the internal resistance of the battery is 11 milliohms, draw max current (400 amps) from the fully charged battery and the terminal voltage will drop to 7.5 volts. A robot that will draw high current throughout the match will cause that voltage to drop to perhaps as low as 4 volts later in the match when the battery is discharged. 4 volts=robot dead! Take that same battery out of the robot and measure with a DVM and it will read 12 volts, no load, but won't run another match.Emphasis added.

Thanks for jumping in here, Al. I think it bears repeating that simple voltmeter measurements do not indicate very much about a battery's readiness. Too many teams neglect battery management and suffer the consequences.

[KenWittlief]

I think Eric has the right idea to make a small device that can measure the energy (charge) remaining in a battery, esp if it was designed to only work with one specific battery (the KOP Exide for example). Measuring the voltage with a load applied does give you a good indication of where you are on the discharge curve.

But now he has me wondering. I dont think you could come up with an instantanious check that will tell you how much capacity is left in the battery (ie, is this a new battery, or one that will only last 30 seconds after its been fully charged)

but I wonder if you could do a short charge, discharge sorta thing to a battery, and get some indication of how well it will hold a charge? Design a tester that would put 2A into the battery for, lets say, 30 seconds, then pull 2A back out for 30 seconds, and see how much the voltage changes?

That might be feasible. Eric might be on to something here.

[Al Skierkiewicz]

Ken,
Again there are too many variables for that. The age of the battery (how many charge discharge cycles) affects the surface composition of the plates. A battery capacity has a lot to do with how much of the plate area is actually exposed to the electrolyte. As the lead oxide builds up, less surface is in contact with the chemical. There are some testers available from automotive suppliers that put an instantaneous load on the battery and measure the current. Unfortunately, many of these devices are attempting to discharge over 100 amps and the meter isn't calibrated for a lower amp hour battery such as ours.
Ben, the CBA at 7.5 amps will take a few hours to discharge the battery to the specified terminal voltage of 8 volts. We will typically put a battery on test when we start the class time and watch it periodically throughout the night. Recent batteries have shown a failure in one or more cells over the discharge. Graphically, the battery will run at normal discharge until one of the cells stops producing current, at that point the output voltage falls by 2 volts. We have had at least one battery that had two cells with reduced capacity. Without doing a CSI on the battery, I suspect defective attachments to the individual plates in those cells. Reduced plate surface area=reduced capacity=reduced charge current/chemical action.
I don't have a graph here at work, but I do believe I have some at home. I will try to post this afternoon.

[Alan Anderson]

...Measuring the voltage with a load applied does give you a good indication of where you are on the discharge curve...
That presupposes that you know what the discharge curve looks like for that specific battery (with that specific load, at that specific temperature, etc.).

You can come close to measuring how much energy is left in a battery by looking at the voltage under load, but only if you have already well characterized that battery's behavior in advance. Devices such as the CBA-II can do that characterization unattended.

[KenWittlief]

That presupposes that you know what the discharge curve looks like for that specific battery (with that specific load, at that specific temperature, etc.).

....

Alan is right. As a rechargable battery is used over and over the discharge curve drops down.

so the voltage for a new Exide KOP battery with a 20A load might be, lets say... 11V when it is 50% discharged

and for a battery that is a year old, it might only be 10.5V when the battery is 50% discharged.

So, yes, you would have to run each battery your team owns through a charge/discharge cycle to get the present discharge curve, and print it out for each battery.

Battery remaining-charge meters are very complex. They have gotten pretty good on newer laptops, but I remember the older laptops: the little battery indicator went something like: 100% for several minutes, then 80% for an hour or so, then it would drop to 10% about 20 seconds before the laptop died.

[FIRST JerseyKid]

The only this I want my device to measure a 'fully' charged battery's discharge curve when putting the battery under a large load. Then if you make some assumptions you can come up with an idea of how good your battery is.

Im still thinking about it. I need to understand more about our batteries.

I know we can't test the plates for how much build up are on them. Does plate build up cause internal resistance and/or less capacity, both rite.

Tell me if my thoughts are wrong on this. A battery trys to keep a constant voltage rite? But when it has less energy and gets hotter some of the voltage drops on the batteries internal resistance, making it even hotter. If you measure the voltage curve couldn't you predict the drop across the battery and find how much build-up is on the plates? I haven't done much with batteris. The only stuff I know is really from chem class and physics class but thats not practical.

[KenWittlief]

the jpeg below it the discharge curve (typical) for a lead acid battery (like the ones in the KOP)

to measure the discharge curve, you have to run the battery through the entire discharge cycle.

as you can see, if you pick a part of any of the three curves, you can find a part on the other curves that look very much the same, so if you only measure part of the curve it would be difficult to tell which one you are on (the new battery curve or the old battery curve....)

but I would not rule out that it could be done with a quick test (a few minutes maybe) to get a general indication of the state of the battery.

You are correct in your understanding. The cells of a lead acid battery put out 1.5V each, and that voltage does not change much as the battery is discharged. What does change is the resistance of the electrolyte, and the contact of that liquid to the plates. In a way, when a battery is used its like the battery is getting smaller and smaller, until the cells start to drop out (electolyte is closer to water than acid) and they can no longer generate a voltage.

[Al Skierkiewicz]

What we have found is that there is little change from new to old batteries. The internal resistance remains relatively constant as does the terminal voltage under load at least at 7.5 amps. Even the amp hour capacity changes little until the battery is well used (several years is typical). When things start to go bad they go in a big hurry. Sudden drops of 1/2 the amp hour rating over a few days is common.

[Al Skierkiewicz]

In looking to upload a picture of the test overlays, I found a previous posting. See...http://www.chiefdelphi.com/forums/showthread.php?t=42284&page=2&pp=15 for a graph and explanation. It is the last post.

[eugenebrooks]

In looking to upload a picture of the test overlays, I found a previous posting. See...http://www.chiefdelphi.com/forums/showthread.php?t=42284&page=2&pp=15 for a graph and explanation. It is the last post.

You can easily make a nice battery load tester with an old robot
controller and a few of spike relays controlling loads to high
power resistors that can be bought from digikey. Use a simple
voltage divider to read the battery voltage directly with the robot
controller. Run a load similar to the average load that you expect
your robot to apply, 20 amps would really be a minimum, and plot
the voltage of the battery as a function of time in the manner that
Al has shown in the prior thread. When this discharge curve starts
showing unattractive behavior compared to your good/new batteries
mark the battery for practice use only. Keep the records on all of your
individually numbered batteries from year to year so that you can
identify a battery that has taken a turn for the worse.

We qualify our growing collection of batteries in this manner just
prior to every season. Yes, it is a lot of work and writing a RC
controller program that automates the process is a good idea.
The effort is worth it. It is good insurance against a battery
going bad during a competition.

You can build/buy a sophisticated constant current load tester,
but you can do a good enough job with simple load resistors.
Don't forget the cooling fans...

Have fun,
Eugene

[DonRotolo]

There are a few things you can test on a battery. Common parameters of interest for FIRST teams are capacity (how much energy can it hold), State of charge (how much energy does it have in it at the moment) and condition (how much of the maximum energy it can hold will it deliver - or in other words how 'good' is this battery). All three use different tests.

Ken is testing capacity with his 600 mAh NiCds. He needs to understand the conditions under which the battery will deliver 600 mAh - batteries of that size are tested at a 10 hour rate, meaning he should be able to get 60 mA for 10 hours (=600 mAh) before the battery voltage drops to some specified value (usually 0.8 volts for NiCd chemistry).

The Lead-Acid batteries used in FIRST are rated at 18 Ah, but large batteries like this are tested at a 20 hour rate, in this case meaning 900 mA for 20 hours. If your discharge rate increases, say to 50 Amps, you might see 5 minutes before the voltage drops to 7.5 volts, or a bit over 4 Amp-hours (Ah). Discharge rate and final voltage are key parameters here. This measurement is not very critical for a FIRST team, since we know what a new battery can do.

Capacity can only be measured accurately by a long-time discharge.

Condition can be measured fairly quickly, if necessary, or through a long-time discharge as above, in both cases by then comparing it to a new battery or manufacturer specification. But, condition is usually fairly constant, it changes over time but slowly. A FIRST team might want to measure all their batteries a week before competition, to pick which ones are for practice and which are reserved for competition. The quick test is to start a discharge test, but at a much higher rate (usually 3x the battery AH rating) for a fixed period of time, and comparing that to a new battery's behavior. There are microprocessor-controlled battery analyzers that do this without a large load, look up the company Midtronics. The aforementioned West Mountain Radio Computerized Battery Analyzer (CBA) is great for this purpose.

This kind of test is what your car workshop does to see if you need a new battery.

The final kind of test FIRST might be interested in is State of Charge, or how charged is this battery. A load test is not the correct method (it also reduces the charge considerably, defeating the purpose), instead a test under a small load is appropriate.

We need to consider a phenomenon called "surface charge", where there is a small amount of 'false' energy in the battery, a remnant of a recent charging. Let a battery sit for 60 minutes and the surface charge disappears, OR put a small load (I suggest a car headlight, 55 watts = about 4.5 Amps) for about 60 seconds to remove the surface charge.

Once there is no surface charge, a simple voltmeter test WILL tell you the state of charge accurately, but only if your voltmeter is accurate !!

12.64 volts is a fully charged lead-acid battery. You will measure up to 12.75 volts, any higher than that is surface charge. 12.50 is a 75% charged battery, 12.35 is 50% charged, 12.00 Volts is a 25% charged battery. (You will often measure voltages far below this - this is the opposite of surface charge, I call it surface discharge. Again. letting the battery sit for an hour will allow that to equalize, and you'll get an accurate reading.)(Note: Leaving a lead-acid battery below 12.50 volts for a long (days) time will damage it)

Notice the problem here? The difference between a fully charged battery and a 50% discharged battery is only a few tenths of a volt. To help differentiate between a 90% charged battery and a 100% charged battery (a difference of a few hundredths (!) of a volt), a small load (that same car headlight) can be used while measuring. I actually use a brake light bulb, 21 watts = about 1.8 Amps. This technique will show a few tenths of a volt difference at high (90%-100%) charge conditions, just what you are seeking. With the light connected and glowing, measure the battery terminal voltage right at the terminals (anywhere else will introduce error).

I could give you the exact voltages versus state of charge, but perhaps you should try this for yourself. Get a nearly brand new battery, or at least one in excellent condition, and charge it fully, allowing it to sit on a FIRST-type automatic charger at least overnight. Measure the voltage under a light (no pun intended) load. Then, pull off some capacity (900 mA for 1 hour brings it down to about 95% - maybe measure in 10 minute increments) and measure again. Make a chart, and you'll be in business. Maybe share with the rest of us...

Some other things about lead-acid batteries:

As lead sulphate forms on the plates, the internal resistance of the battery increases, which we call discharge. The process is not 100% reversible, so there's a little loss of capacity each cycle. Shallower cycles are more reversible than deep cycles. Eventually the battery will not accept a charge, it has then failed a 'normal' failure after a good service life (assuming nothing mechanical in the battery has broken). Allowing a battery to sit for any period while discharged (even a bit) causes damage. This means, a lead-acid batter prefers to be fully charged at all times. Heat hastens the normal failure mode, and is generated by charging and discharging (mostly due to ohmic resistance, not really by chemical action). Self-discharge can be about 20% per month.

So, keep your batteries charged as much as possible, always store them charged, and recharge at least every 2 months, more often is much better.

Hope this helps!

Don

(PS: I do batteries for a living).

[dcbrown]

If your discharge rate increases, say to 50 Amps, you might see 5 minutes before the voltage drops to 7.5 volts, or a bit over 4 Amp-hours (Ah).


According to the Excide spec for the battery, if the discharge rate is 54A then the battery should still be at ~10.8v at 5 minutes. It doesn't look like we hit the knee in the curve until somewhere around 7-8 minutes. Haven't check this on our batteries, but it is on our to do list.

We need to consider a phenomenon called "surface charge", where there is a small amount of 'false' energy in the battery, a remnant of a recent charging. Let a battery sit for 60 minutes and the surface charge disappears, OR put a small load (I suggest a car headlight, 55 watts = about 4.5 Amps) for about 60 seconds to remove the surface charge.

The BCI (Battery Council International) indicates that a C/3 load (6A for the EX18-12) should be applied for 5 minutes followed by a 10 minute rest period for deep cycle batteries like AGM/GEL technology to remove surface charge before measuring SoC (or a rest period of at least 1 hour). For starting batteries such as (automotive type) flooded cells, the recommendation is ½ the CCA (cold cranking amps) for 15 seconds followed by 10 minute rest period. I suspect these are conservative recommendations – we’ll give the 4.5A @ 60s a try to see if that sufficiently removed surface charge.


12.64 volts is a fully charged lead-acid battery. You will measure up to 12.75 volts, any higher than that is surface charge.


I’ve been looking at information that indicates that deep cycle AGM/GEL cells have a different SoC table than flooded-cell technology. It isn’t much, but the temperature compensated table appears to have AGM/GEL cell voltage at a ~1.5% higher voltage (12.79v) than flooded cell which is 12.64v at 70F when 100% charged. This information seems to match what we’re seeing on our bank of 6 Excide batteries from the last couple years. Fully charged batteries that have been off the charger for a week are showing ~12.79-12.84v for batteries from 2005/2004. There is one suspect battery that is only showing 12.49v when fully charged, but it is from 2003.