# Battery capacity testing

• posted

I've finally acquired enough equipment to measure the remaining Amp-Hour capacity of my Lead-Acid and Lithium battery collection. The first result that jumped out is that older batteries suffer from rising internal resistance as they discharge, enough that the automatic low voltage cutoff trips short of rated capacity, and then the battery slowly recovers to well above the full discharge voltage given in the specs.

The 5-year-old 12v 4.5Ah UPS battery I tested this AM delivered 2.45Ah at 3A, which is the average current my laptop draws while browsing. Table 2 shows in the 1 Hour Rate column that it should be good for

2.75Ah at 2.75A current.

Does anyone know a good reason why I can't measure the true remaining capacity in two steps by first discharging to 10V at the fairly high current of my typical loads, then continuing at the 20 hour rate AGM batteries are specified for until the voltage drops to [the appropriate endpoint] again?

The run time for a typical load tells me how useful the battery still is, but it combines the effects of capacity and resistance. I'm wondering if also knowing the Amp-Hour capacity at the 20 hour rate, with less interference from the internal resistance, would indicate how well my long-term maintenance procedures work.

-jsw

• posted

That wasn't a good example. I have a 12V 12Ah AGM battery that is down to 1 useful Amp-hour because its voltage droops to 10V so quickly. Afterwards it recovers above 12.2V.

-jsw

• posted

7-TechManual-Lo.pdf

I think that doing the 20 hour test my provide you with academic indication of the condition of the batteries, but only at the 20 hour rate. In my rec ollection (based on 25 years ago designing a 100 station lead acid charger) , there is surprisingly little correlation between capacity at different di scharge rates.

So, I would suggest you test at your normal load and perhaps with a "normal minimum load" assuming that those rates are pretty far from 20 hours. Anyt hing else is, as I said, purely academic.

BTW, while I was buying the voltage reference, I also bought a USB power me ter (Drok). The Amazon add and the user's manual keep referring to "capacit ance" measurement. What they really mean is capacity of USB battery packs. Pretty funny. Sort of. You can actually buy this meter bundled with a USB l oad bank.

• posted

I think that doing the 20 hour test my provide you with academic indication of the condition of the batteries, but only at the 20 hour rate. In my recollection (based on 25 years ago designing a 100 station lead acid charger), there is surprisingly little correlation between capacity at different discharge rates.

So, I would suggest you test at your normal load and perhaps with a "normal minimum load" assuming that those rates are pretty far from 20 hours. Anything else is, as I said, purely academic.

BTW, while I was buying the voltage reference, I also bought a USB power meter (Drok). The Amazon add and the user's manual keep referring to "capacitance" measurement. What they really mean is capacity of USB battery packs. Pretty funny. Sort of. You can actually buy this meter bundled with a USB load bank.

=========================

I want to separate the effects of capacity and internal resistance to see if equalizing etc improves either or both of them. The internal resistance of AGMs has some strangely behaved component reputedly related to an oxide film. Otherwise I discharge them at the current my laptop draws when browsing as I have them for power-outage backup and NWS radar is the best indication of storms approaching my house that I've found. It tells me when to repair roof damage and when to tarp it.

I bought this which has an easily set low voltage disconnect and handles up to +/-30A,

and previously this which is 10x as accurate at low current

The first one measures charge and discharge current separately and counts the Amp-hours up or down accordingly, though the Watt-hours total is the positive sum of both (???). It has a more accurate voltmeter and a better timer that counts seconds and stops when the relay opens, allowing a pause in the measurement and a record of battery run time. Unfortunately the current resolution is 0.1A despite the display, so it doesn't handle small AGMs well.

The second one matches other ammeters to 1 or 2 digits and I use both in series for discharge loads up to 10A. Together they each make up for the deficiencies of the other. The 12V,12Ah battery is discharging on them at 0.5A.

-jsw

• posted

I've read that repeated capacity tests on the same battery don't correlate well.

• posted

I gave up. Never got any predictive information.

Most of my testing was done with Lithium batteries in laptops.

I consider a laptop battery bad when it won't run the laptop long enough. How vague is that? ;-)

Started with bad battery packs and tested cells. At low current, I almost always got something like specified capacity. The electrons are in there, but the laptop won't let you have them.

I don't know what the sampling interval is, but the laptop wants to shut down at some voltage so you don't lose data and call the vendor. Subtract the peak voltage across the ESR from the battery voltage. If it dips below the threshold, the laptop wants to shut down.

Turn off the power features that warn of impending low voltage shutdown.

The symptom is that the battery gauge decays slowly for a while then drops instantly to a much lower number. The laptop senses impending doom, but you've blocked that. I've had laptops run two hours past the point when the battery gauge hit zero. Problem is that when it dies, you lose whatever you were working on.

I've never had any success trying to fix the ESR. That's probably the same problem you have when your car fails to start. Never been able to do anything about that either.

The higher the peak current, the fewer electrons all those protection circuits will let you have. Even with a new battery, capacity is a strong function of load current.

Numbers from browsing the web won't help much when Microsoft decides to do an update and runs all your cores at 100%.

• posted

[ ... ]

Well ... that is not a problem for me. The first thing I do when I pick up a new (to me) laptop is to remove the virus. The last virus was called "Windows 10". I replace it with either a linux or an OpenBSD system, so Microsoft doesn't have a say in when I update, and updates for the others are based on telling me that updates are available, and letting me decide whether and when to install them. :-)

Enjoy, DoN.

• posted

The "fuel gauge" IC counts Coulombs in and out to determine actual Lithium battery capacity, on the assumption that they recharge at 100% efficiency. It resets its capacity estimate if the battery is nearly fully discharged and recharged, but if only partly discharged it can't detect the slow loss of capacity with age and retains the old, overly optimistic number from the last full cycle. That's why the sudden jump when it realizes it's wrong.

When you give the battery another full cycle it can measure and update the battery capacity to its new, lower value.

As an experiment I reduced the low voltage trip as far as possible and got almost as much run time from an old Dell battery below the 5% level as from 100% to 5%. It appears that Li-Ion cells can be discharged down to or even below 2.7V briefly without much damage. The normal settings are above 3.0V.

This gives you the battery voltage:

-jsw

• posted

Hmm, there are 3 different pictures of the back of those. One has a built-in shunt, another a pair of relays, and another is bare. Which is the real meter pic for the "30a w/ relay"?

IF I ever get the weeding done around here, I'll get those panels up and build the control panel to see how those li'l Bayites work. You showed another link for a milliamp/millivolt-resolution meter a few weeks ago, too. How's that working for you?

The former part is cool. Not having proper resolution for decent data is never fun, though.

Did I ever ask you why you didn't use a real battery for that? (real being 12v 35-275Ah) I set one up for use with the 45w HF trio of panels and was able to power a 14" electric chainsaw with the 2kW modified sine wave inverter, also from HF. It would have taken days to recharge it (or more panels if needed for continued use.)

• posted

The 30A model I received has a shunt and blue NC relay on the base module. The display module has a small red+black pigtail for external power if you don't use a USB connection. At first the USB connection on mine was poor and it intermittently shut off, or switched to wireless without losing power. The correct accumulated totals reappeared when it reconnected.

It displays current to 2 decimal places but is accurate only to 1 place +/-, for example 0.478A on a Fluke 8600A reads as 0.48A on the

10A "Electrical Parameter Tester", and 0.65A on the 30A unit.

A layer of Gorilla tape tightened the USB plug in the base unit against the circuit board contacts and it has remained connected when moved.

The 33.00V/3.000A meter is my favorite for recharging and equalizing batteries slowly from my solar panels. It clearly shows when a small AGM's charging current has decreased to 1% of the C/20 capacity, like

45mA for a 4.5A-h AGM battery. Currents around 1% are recommended end points for trickle charging.
Diagram 4 gives 1-3% for flooded, Diagram 5 gives 0.5% for AGM.

As mentioned, the current rises in older batteries and is an indicator of declining condition.

I first learned how to make accurate measurements as a chemist whose results might have to stand up in court, then when building very precise automatic test equipment for the semiconductor industry. Analog Devices' op amps and voltage regulators were tested on machines whose performance I was responsible for.

I do have "real" batteries that will run the fridge for about 20 hours. Once I'm satisfied with my discharge testing setup I'll get to them. For now I'm testing and risking smaller, older, less valuable jumpstarter and UPS AGMs. These tests are too long to watch and if the low voltage disconnect fails the battery could be drained flat before I notice.

-jsw

• posted

Previously I was using a rewired Battery Isolator I bought from Quicksilver Radio for \$5 at a hamfest to disconnect the load when the voltage dropped. This describes the idea:

The rewiring changed it from switching the load to the second battery when the main one's voltage dropped to switching one battery from the load to a charger.

As a discharge controller it has the disadvantages of still drawing current from the main battery to operate the relay after it has discharged to the disconnect voltage, and needing an adjustable power supply to set or check it.

I haven't seen that remaining for an hour or so at full discharge would further harm a battery and want to record the voltage it recovers to without any load as an indication of true remaining capacity and a safety check that I haven't set the disconnect voltage too low and drained the battery too far.

If not for its poor current resolution the 30A Drok unit would be a fine discharge test controller when powered from an external 12V supply that separates its operating current from the test circuit. The circuit board was drilled but not properly connected for an SPDT version of the SPST NC relay it comes with. I'll set the Battery Isolator to a lower disconnect voltage as a backup on the load side of the Drok.

Based on Amazon comments, it seems the 3-wire / 2-wire jumper may select battery circuit or external power to operate the device.

A DC-AC inverter powering a safe resistive load like a crockpot can be used as a discharge test load though you can't set the dropout voltage and it may cycle back on when the battery recovers.

-jsw

• posted

Duct tape to the rescue again.

I've downloaded both guides you've mentioned in the past several days and will have to compare them. Each one will have tidbits of info the other doesn't. That's the true benefit of research: gleaning tidbits.

That sounds like something I should pay attention to.

Cool! Nice legacy.

I forgot to ask what you meant by that. Are they drawing half an amp or showing that discharge rate?

Smart.

That wouldn't be fun. How often do the disconnect programs (or hardware/relays) fail?

• posted

I suspect it means that the self-discharge rate has increased and they may need more frequent topping off. I don't know how it relates to remaining capacity.

I kept the current from the battery close to 0.5A by tweaking the rheostat load. Surplus phone chargers power the meters and the relay independently from the battery, since I won't be wasting my backup by running these discharge tests into a dummy load during a power outage.

Some of my homebrew test equipment runs on AC power for better accuracy, some on DC from the battery being tested or the solar panels so it will still show demand or charging rate during a power outage. The load current varies too much for accurate measurement and battery remaining life prediction, but all I really need to know is if the batteries can run the fridge overnight.

The box whose components I tested this morning will have a switch for internal or external power. The meters have to be on external power to measure single 18650 Lithiums to their discharge endpoint which is below the minimum supply voltage..

The two of these I bought match the Fluke 8800A to 1mV. YMMV.

How often do brakes fail?

• posted

Great price!

Discharge in series? Sounds like a plan.

Those I've seen only engage the relay to switch to the secondary battery. Are you talking about when the secondary battery is discharged/cutoff and the relay continuing to be engaged? I see that as a problem, too. Perhaps rig up a kickout relay to disengage when the cutoff hits on the secondary?

Yeah, that's a fly in the ointment of capacity measurement. Are you saying "full discharge to cutoff point" there?

There ya go!

It didn't look like that was fully populated in the pic I saw. IIRC, it had only one pin/solder joint out of the 3.

Hmm, yeah.

• posted

This one remains powered by the main battery when it switches the load to the second one, perhaps to avoid the glitch while the break-before-make relay contact is moving. It drives the relay with an SCR and won't release and revert to the main battery until the user pushes a disconnect button, regardless of how high the main battery may have recovered or been recharged. This means that connecting NO to a charger won't make the relay turn off when the battery voltage rises.

I cut and jumpered the traces to redefine COM as the battery instead of the load, which is now NC. Originally COM was the load, NC the main battery and NO the secondary one. As you said it would simply allow the secondary battery to die, but retain whatever capacity the trip point left in the main battery.

Maybe running the anchor light as long as possible is more important than preserving a battery that sinks when the boat is hit?

Here's the problem:

"To get accurate readings, the battery needs to rest in the open circuit state for at least four hours..."

The AGM I discharged at a little less than the 20 hour rate (0.5A) tripped at 10.0V (twice) and then recovered to 12.15V, which is over

40% State-of-Charge on that chart.

The point of knowing the full capacity is to find out why I'm not getting it, and see if anything I can do makes an improvement. I can't fix bad interconnects but a discharge - charge - equalize cycle reforms the active material. Only measurements will show how well equalizing and desulfating work. I know I can make them last much longer than usual, but is it worth the effort?

-jsw

• posted

The two-tier DC load method is how the engineer had me test electric vehicle Lithiums, using a programmable electronic load and a much better DC current probe than I'll probably ever own personally.

-jsw

• posted

One consideration for backup is the cost of failure. if the worst that can happen is a gallon of milk has shortened shelf life, that's a lot less serious than a mechanical fridge compressor that stalls when it tries to start under brownout conditions.

If you're serious about these kinds of measurements, wire up a LM317 (maybe with a booster transistor) as a current load.

If the voltage doesn't vary too much, an incandescent light bulb makes a current source that's more stable than a resistor.

A computer controlled dual-output power supply is a useful tool. Use one output to charge the battery and read back the voltage. Use the other output to drive a voltage to current converter for the load. Makes it very easy to control and log and graph and...

If I had a solar system, I'd have an arduino or some such monitoring it at all times.

I have a Palm Pilot monitoring the HVAC system. Had the installer come fix it before it quit completely.

• posted

I'm more concerned with being able to replace the lost food after a hurricane or ice storm when the roads are blocked by fallen trees tangled in possibly live power lines (generator backfeed) and the stores don't have power either. The area I'm in typically stays dark for a week before the line crews get to us.

I had very nice, and expensive, equipment like that when I was the battery tech at Segway. While running the tests I found out what was important and what wasn't, and saw that I could obtain adequate results at home with much cheaper surplus equipment like old Ohmite rotary rheostats. Tubular variable resistors are difficult to adjust when hot, around 600F near their rated power.

The only time I programmed a load to change automatically was for two-level battery impedance measurements. Otherwise the voltages and currents remained constant until I changed them, which can be done as easily with a knob. We used our lab power supplies to recharge vendor sample Lithium cells to the appropriate voltage at a constant current.

The best representation of a real-life active load isn't constant resistance or current but constant power such as a DC-AC inverter draws, and I can just use one with a hotplate on a Variac for the adjustable load. Inverters usually have a functioning low voltage disconnect to protect the battery. However a fixed resistance load is fine for tracking battery ageing and easier than a switching load to measure accurately.

These connected to a laptop make a good data acquisition system for slowly changing parameters like battery voltage.

Since they are optically isolated they can be connected anywhere in the circuit without creating unwanted current paths through their cabling, a major concern with grounded instruments like scopes. You can combine their separate datalog files into a spreadsheet by aligning the timestamps.

A laptop makes it easy to set up in the best place to run the test. For me that's in the laundry room where I have running water to deal with battery acid spills. The laptop itself can be the representative load on the battery and its own internal battery will keep it running when the battery being tested drops out, so it doesn't lose the most important data point. The time that happens, saved in the system Event Log, may be all you need to know. Older, thicker laptops with CardBus or ExpressCard slots for port adapters to read multiple meters are cheap.

My pure sine inverter reports status over a hand-wired non-standard cable to a monitoring program I wrote but I haven't found much use for the results. This is more valuable:

If that link doesn't work it's a Bayite PZEM-051.

This VAC-1030A could be great if they fixed a few minor bugs and wrote a decent manual. It's good enough already.

I haven't tried the 100A model, VAC-1100A, because I normally recharge at a current too low for it to measure accurately, though the VAC-1030A isn't much better.

Monitoring my solar system showed that clouds pass randomly and I still need a backup generator since I can't depend on a consistently adequate solar output, especially in the kind of weather when I'd need it most. I occasionally check the output and wiring drops on bright, clear days. A cheap HF DVM gives the short circuit current.and an RC wattmeter and variable resistor can find the maximum power point.

That's just an example, not the discontinued meter I own. It shows that the power doesn't fall off much on either side of the maximum, meaning there's little benefit from an MPPT controller on a small system.

-jsw

• posted

What happens to the expensive power supply if the AC power fails while the battery is connected, or you accidentally connect the battery backwards???

"Active loads that create a sinking current should not be connected to a power supply."

I've been charging batteries through series diodes, using this meter to monitor the voltage beyond the diode drop:

"With reverse polarity protection to avoid burning out" A spare LM78xx from your junk collection can extend their supply voltage rating above the input limit, an LM7824 increases it the highest.

The one I tested the other night matched a 5-1/2 digit Fluke to 1 mV at 10V in. Notice where it was made.

The output protection I've been adding to my homebrew battery charging variable power supplies is a forward diode to block a meter-busting surge back into the 0.2 Farad output cap and a reverse-biased diode to ground that -should- pop the output fuse or breaker if a battery is connected backwards. The problem with testing breakers is that they have a maximum interrupting rating above which they may work only once, and of course with fuses is that you know they -did- work and hope the next one (from a different source?) is the same.

-jsw

• posted

That would do it.