Sealed Lead Acid Battery Tester Strategy?

I've designed and built a prototype PIC based Sealed Lead Acid Battery Tester, but I'm having confusing and conflicting results and would like some input on whether the strategy I'm using is sensible ...

It's based around a constant-current load, controllable by the PIC chip. The current can be set over a range of C/40 C/20 C/10 C/5 C/2 and C (where C is the nominal Ah rating of the battery). The unit can monitor the voltage of the battery. All the voltage measuring/current measuring and current setting is appropriately calibrated, so any problems are down to approach or duff assumptions on my part :(

The basic idea is a "brute force" test of the capacity of the battery: Load it up at C/n and time how long the battery takes to reach a discharged state. In theory, it will take "n" hours for a 100% battery (aside from the obvious that discharging faster will reduce apparent capacity etc.)

From my research so far a fully charged SLA cell should read 2.16v at rest, and 1.75v when discharged. So for a 12v battery, the useful capacity is found between terminal voltages of 10.5v and 12.96v. Right?

The capacity of the battery can be estimated using this voltage range, which I'm assuming to be a linear function (it nearly is...). So the first feature on the tester is a battery capacity estimator, based on offload voltage.

Then loading the battery at (some constant current) begins the discharge process, and the voltage starts to fall. I time how long it takes for the voltage to reach 10.5v, and then stop the timer, and compute the capacity actually achieved. It should be that simple, however ...

I have a dilemma: When things say "don't discharge the battery below 10.5v" do they mean on-load or off-load? There is a difference!

I noticed that loading the battery, especially at higher currents, causes the terminal voltage to fall, due to the internal resistance of the battery.

So I added a feature to try and measure this resistance at the start of the discharge process. I did this by taking the off load voltage, the on-load voltage at C/n, and working out the voltage being dropped across that resistance. This gives me a figure of = capacity 1. This seems to give reasonable figures, of capacities that are

Reply to
Mike
Loading thread data ...

Depends on discharge rate--the final voltage at the end of discharge at the 8 hour rate is usually 1.75, at the 1 hour rate 1.6-1.65 taken while the cell is still delivering current. Unloaded voltages at 0 charge should run from around 11.5 @ 7 degrees F to about 11.68 at 97 degrees F.

Voltage is not linear but is close during the first 60-65% of discharge--after that point good luck.

Maybe....the internal resistance of the cells is best measured after discharging at a constant rate long enough to obtain a steady voltage at the terminals,and then quickly interrupting the discharge and noting the instantaneous rise of voltage. The instantaneous rise divided by the value of the current at the instant of interruption gives the internal resistance. Another way is to suddenly reduce or increase the current by a definite amount, and divide the instantaneous change in voltage by the change in current.

Final cell voltage on load at 0 charge depends on discharge rate--also, what size wire are you using to attach your test equipment--one foot of #8 @50 amps will drop another .02V.

The value of voltage drop does change--increasing somewhat during discharge. The weakening of acid in the plate pores accounts for another 20-30 percent to the internal resistance drop.

You've done some good work--give Yuasa-Exide a jingle and ask for their Battery Handbook and Technical Guide--good luck. Remember, if it were easy--everyone would be doing it. ;-))

Tut

Reply to
cnctutwiler

some

Load

state.

obvious

rest,

which

feature

10.5v"

battery.

the

load.

above.

reasonable

internal

That

further

causes

cause

function

-

The terminal voltage between fully charged and flat is from about 13V to

12V.

formatting link

Has a really excellent technical write up on their SLA type batteries. Use it as a good reference point for designing any high quality SLA charger/discharger.

regards john

Reply to
john jardine

On Sun, 3 Jul 2005 09:31:32 +0000 (UTC), snipped-for-privacy@posie.local.dom (Mike) wroth:

The better fuel gages actually measure the current rather than applying a fixed load. The amount of charge on the battery is the difference between the current used to charge the battery and the current drawn from the battery, integrated with respect to time of course.

You have to fudge the data to account for energy lost in the internal resistance and the "self discharge" rate if you want maximum "accuracy".

Jim

Reply to
jmeyer

Take a brand new known good 1.5 volt penlight cell, and connect it in series with a 1K resistor in an open circuit. Now, read the voltage, placing the voltmeter from the unconnected end of the resistor to the unconnected end of the battery. What do you read? You still see 1.5 volts, but the cell + resistor can provide only .0015 amps instead of the MUCH higher current the cell could provide without the resistor. Imagine that the resistor is the internal resistance of the battery. That is what happens when a battery discharges - its internal resistance goes up.

Next: don't discharge the battery below 10.5 - in fact, don't discharge it to 10.5. Each discharge takes a little life from the battery that CANNOT be retrieved. The deeper you discharge, the more life you take.

Someone recommended a site for some reading - it would be very helpful to you to look at that, and other sites, as well. There's a lot of information on the care and feeding of batteries out there.

Ed

Reply to
ehsjr

The variable current is a good idea, and C is probably as high as you want to go for small sealed lead-acid batteries. Somewhat along this line, I have tested larger absorbed glass mat batteries (12 V, 25 - 40 Ah nominal) being used in hybrid cars. At first the test was done something like the test for a flooded lead-acid car starting battery: apply a very large load for a relatively short period of time (less than a minute) and make sure the terminal voltage doesn't drop below x. This test is also pretty quick to do. However, we had batteries that would pass this test, but perform poorly in the car. Testing them at a smaller load of 10 to 20 A or so, to some final under-load voltage, and computing the Ah produced proved to be a much better way of identifying weak batteries.

I think this is right. I'm not sure to what extent SLA batteries have this problem, but you might look up "surface charge" in relation to flooded lead-acid batteries. Basically, measuring the voltage of a battery that has just been charging, even if the charging current is switched off, won't always give you the right answer.

I'm pretty sure they mean on-load. If they specified an off-load voltage, then if you wanted your battery-powered device to be able to know when its battery was dead, you'd have to design it to switch itself off (or to a very low load) occasionally and measure the off-load voltage. This is probably trickier than just putting a comparator across the battery and shutting down when the voltage gets to some low level.

I understand that commercial testers do almost the same thing, but use on-load measurements at two different currents instead of one on-load and one off-load voltage.

This seems reasonable. Some SLA manufacturers do mention a range of internal resistances for their batteries; you might look those up.

If you try to draw any current, though, it will quickly crater to down around 10.5 V. You might discharge a battery and then experiment with how little load is required to knock the terminal voltage back down to

Reply to
mroberds

some

Load

state.

obvious

rest,

which

feature

10.5v"

battery.

the

load.

above.

reasonable

internal

That

further

causes

cause

function

There seem to be two possible themes in your post about "battery testing".

1) Measuring how charged is a battery. 2) Measuring if a battery is worn out. Very different things. In this post I just talk generally about both issues.

RATE OF DISCHARGE

We also use constant current drain and measure time to reach a trip voltage.

There is fast discharge, like cranking an engine, where the battery is hardly discharged, but terminal voltage plummets because of reaction products "choking" the plates. Then there is slow discharge, where reaction products have time to disperse where you really get the nameplate Ah. Manufacturer specs give different capacities at different drain currents.

BATTERY METERS BUILT INTO EQUIPMENT :

I think about it this way : I drain the battery at the current drawn by my application, and set the trip voltage at a value which does not discharge the battery too deeply and reduce life excessively.

If your application has a fairly constant drain, you can calibrate your "meter" function to be meaningful at that particular drain. Simplest to get a new charged battery and measure voltage under load - that is "charged" point. Then discharge battery gently to your endpoint then measure voltage under load to get "discharged" point. Connect the two voltages linearly and you have your meter.

If your drain varies dramatically, it is more difficult, but you have to fudge something. Battery meters can never be taken too seriously. Also, users get used to the meter and soon work out that say "two bars left on the meter means I have got 3 more tests in the unit", etc.

BENCH BATTERY METERS

To get consistent results: Fully charge the battery on a voltage float charger or other programmed quality charge process. Then you rest it 24 hours. Then you perform the discharge test. You do it all at a fixed temperature (say 26.5C water bath). Other procedures give you results dependent on history, temperature.

The lower current discharge takes longer, which may be a problem. C/10 takes about 10 hours, and is a pretty fair measure of capacity.

I would not worry about the bounce effect. Once the voltage hits your low trip point, time is up. That is how Ah tests are performed. A full discharge meter by definition performs a one shot measurement, then you must charge the battery and start again.

If you settle a battery for 24 hours at your standard temperature, terminal voltage is an excellent measure of how charged the battery is.

DEPTH OF DISCHARGE

Going down to 10.5 V may be OK for a tester, but in use I would recommend

11.8V minimum for a decent battery life.

INTERNAL RESISTANCE

Varies between brands, types (eg gell to VRLA), time on the shelf and "how worn out" and depth of discharge. Resistance is useful for judging battery deterioration, but with small batteries you have to calibrate against a particular brand and model as well as state of charge, so it can mean a lot of work for you to get it right. Measure resistance at small signal - say

200 mA pp 100Hz squarewave forced thru battery - resistance is then proportional to AC terminal pp voltage.

The discharge to flat time at C/10 is a good cross brand universal test of health of small lead acid batteries and can be your standard against which all other tests are calibrated.

Roger Lascelles

Reply to
Roger Lascelles

Thanks for the replies ...

I am aware of "surface charge" and I am resting the batteries after charging to make sure that they've settled down, so my start point seems fairly well defined. The voltages for a charged, off load battery are mostly well clustered around 12.7-12.9v for both new and used batteries.

Exceptions are duff batteries e.g. 10.78v (shorted cell?).

"Roger Lascelles" wrote :-

I've got time to burn, so C/10 is acceptable. This is not for a production unit, or super accuracy, but I would like it to be in the ballpark, and consistent.

It is, sort of, (save for the estimate function being so over-enthusiastic about my flat batteries!)

Understood. It's just that, when repeating the estimate-my-battery function at the end, I got an unexpected result. Curiosity is a bad thing. :)

That was what *I* thought, as I said, once I've flattened the battery, the voltage comes back up over minutes to say "70% full". After 24 hours, I'd fully expect it to be even higher. Which is even *worse*! Hmm.

Yes this is a tester ONLY, not a battery meter for use in equipment or for spotting when a device should shut down.

I can deal with those variables, by taking a reading at the start of the discharge.

Damn. That shoots holes through that one then :) If it varies, I'd need to retest it, or ignore it.

"john jardine" wrote :-

I'll check out the link. The low point is one of the contentious values that varies. Some manufacturers quote "watts available to 10v" and other voltages up to 11.5v as a minimum point. That's a big range.

snipped-for-privacy@nowhere.net wrote :-

What I'm doing isn't strictly a fuel guage (part of is, just as an estimator BEFORE the test is done). The actual tester applies a load which ensures a constant current (and measures this also to ensure it's really happening).

ehsjr wrote :-

OK, I'm going to have to rethink how and when to measure the resistance if it's expected to change.

Understood: 300-400 cycles for 100% discharges, 3500+ for 20% discharges are the kind of figures I've seen for e.g. Camden Europa. This is not a process to be repeated often.

snipped-for-privacy@wmconnect.com wrote :-

From recording the voltage over a discharge, I got a linear-ish section for the first part of discharge, then a second linear section (shallower) for the remainder. Then a strange plummet ... that looks like the battery is done, and I'm stopping. This may not be true across all batteries though :)

OK, that would be a possibility: Halve the current periodically during the test and then check the voltage change.

The tester has 2 independent banks of transistors. Each bank is rated for

25A max. Each is cabled with 30A cable. So far I'm keeping to load currents to < 20A (so that's 10A per bank). The wire is 50/0.25 copper (2.5sqmm).

snipped-for-privacy@worldnet.att.net wrote :-

In which case I have seen some batteries which when loaded by the target device is almost immediately flat, as the voltage falls below

10.5v. However, discharged at a lower rate, they are still useful batteries.

So are they flat or charged?

An example I'm thinking of is e.g. a 44AH mobility battery: It won't drive the mobility vehicle up a hill without the speed controller kicking in and objecting to the low voltage, even when newly charged.

However, they will power a 12v emergency fluorescent for HOURS. Internal resistance of the cell is high ... but the capacity is OK. Which is why measuring the capacity doesn't tell the full story, and that's why I'm also checking the resistance.

Done this already, which is why I know that feature is reading figures in the right ballpark. But I wasn't expecting it to change with discharge level. That's not mentioned in the datasheets :)

Thanks to all, I shall go and have a rethink and re-write some of the code to try and get this a bit better. At least it seems I'm on the right track.

--
--------------------------------------+------------------------------------
Mike Brown: mjb[at]pootle.demon.co.uk | http://www.pootle.demon.co.uk/
Reply to
Mike

If you are doing EV (or any other highly variable load) get yourself a copy of the Curtis battery book.

formatting link

Robert

Reply to
R Adsett

Having looked at the Yuasa manual link, it's raised some interesting points that I wasn't aware of :-

Yuasa quote their capacities at 20 hour rate to 1.75v per cell (10.5v) so I'm using the right stopping voltage for my tester as "stop discharge" point.

11.0v and higher are "over cautious", in my opinion, and are not achieving the full capacity of the battery (highly relevant in a battery tester). It's probably better to use 10.5v and higher as a stop point for a piece of gear being powered by the batteries. But not for a tester.

My capacities come out slightly under the nominal because I'm using 10 hour and 1 hour rate timings, and higher currents mean lower effective capacities. At least now I have a graph of how the current draw derates the capacity away from nominal.

Yuasa quote discharge levels that go down as low as 1.6v per cell in their calculation tables (9.6v!) which reinforces a figure I'd seen in CSB Batteries and Camden data sheets. I think that's a bit low. However, context is important here ... it would seem that the acceptable "stop" voltage varies with the discharge rate, e.g. :-

0.1C or below/intermittent load=1.75v (10.5v) ... 0.6C=1.60V (9.6v) 3C=1.30v (7.8v)

That caught me out.

So really when discharging at "C" I've not been truly flattening the battery: I've been stopping at 10.5v or thereabouts. I should be stopping somewhere around 9.2v. That's on load, actual voltage. It looks like these figures are already trying to compensate for the internal resistance of the battery, among other effects.

However, estimated battery capacity (off load) doesn't use the same "end point". This is one of the things I got wrong. I thought the discharge termination voltage was the same as the off-load "empty" voltage of 10.5v or less.

The offload voltage should range from 11.5v-about 13v for 0-100%, in a fairly linear way.

This partly explains why my discharge measurement and "remaining estimate" were at odds: I wasn't totally flattening at higher currents, and my estimator was using a range of 10.5v-12.96v, which would bias the estimate readings as too high. It also partly explains the "voltage bobs back up" effect. It seems I'm not the only one to make this mistake, as my Belkin UPS shares my over-enthusiastic estimates, seeing how it relates the terminal voltage to battery capacity.

I will go and code wrangle and see if it improves now. Thanks for the pointers!

--
--------------------------------------+------------------------------------
Mike Brown: mjb[at]pootle.demon.co.uk | http://www.pootle.demon.co.uk/
Reply to
Mike

Happy to see you're on target now. ;-)) See my 4 Jul post--although I haven't reviewed the Yuasa manual, little has changed in 30yrs with lead acid batteries. Keep in mind that in addition to discharge rate, plate design will affect to some degree the final numbers for total power-on discharge as will temperatures--manufacturers data will help here.

"Storage Batteries" by Ralph W. Ritter/552B International Textbook Company, Scranton, Pennsylvania--Copyright in Great Britain, 1939--will provide additional data on old time solutions/equipment for your problem.

Again, you've done some good work--best wishes and let us know how it all turns out.

Tut

Reply to
cnctutwiler

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.