I did an experiment where I used a "regular" smart-charger set at 12A max AGM mode on a 12V 100Ah rated AGM battery to what it told me was 100% charged. The charger then automatically went into trickle charge mode of 13.5V. I immediately took the battery off of that charger, and put it on a laboratory power supply unit (lab PSU), manually set at 13.5V (and allowed to go up to 1.5A max).

What I am wondering is why does it take so long for that battery to "level off" at about 1/2 watt of charge power going in? For example, when I used the lab PSU, it started at about 3 watts of charge power. Over the course of about a week, it slowly went down to about 1/2 a watt then didn't go much lower.

So what I am thinking is it is very difficult to get those last few % of charge in the battery, so many modern smart chargers don't bother and maybe stop at 90-95% charged.

So does anyone know what exactly is happening in the battery and why those last few % take "forever and a day" to go in the battery and "stick"?

Just as a related comment... it is interesting to see what a particular battery will "bottom out" as, as far as charge power. 1/2 watt at 13.5V is only 37mA. I think many smartchargers may be programmed to stop at 1% of the Ah rating of the battery, so in this case, stopping at 1% of 100Ah = 1A. However, at that point, the battery still seems to want to take more charge.

One experiment I want to do very soon is after charging the battery for a week until it is what I think is a true 100%, drain some fixed amount of power out of it (let's say 1/2 KWh) using a reasonable load (such as a 150W incandescent lamp) and a Kill-A-Watt meter, then measure how much AC wall power it takes to get it back to that true 100% State of Charge about a week later, using that same Kill-A-Watt meter. I suspect running the initial charger a few hours plus running the lab PSU for about a week 24/7 will add up to maybe 3x to 4x the power (I am guessing 1.5 to 2 KWh). If that is correct, then something seems "out of whack" that it takes that much more power to get the battery fully charged.

Also let's assume the first charger really does stop at a true 90 to 95% state of charge. At that point, the wall power consumed is probably much less than 3x to 4x the power we got out of the battery, and I will record this number. It might take as much (or more) wall power to trickle charge for a week, then it does to get the battery to "100%" by the smartcharger's "definition". This should be a very interesting experiment.


I've had the battery on the lab PSU for about a week now and it is very close to drawing 1/2 watt (37mA at 13.5V). It is around 41mA. It must be very difficult to get that last little bit of charge in there, however, I suspect keeping the battery at a true 100% SoC is healthier than what a typical smartcharger (that is much quicker) calls "100%". It would be interesting to do a capacity drain test using both charge methods (smartcharger only vs. smartcharger for bulk + 1 week on lab PSU). I wonder if the difference in capacity will be 5% to 10% more using the lab PSU to top it off.

The last few mA that it drops are VERY slow. It looks like 37mA might be the lowest it will go without reducing the voltage. 13.5V is where it stayed. It must be very close to the battery's self discharge rate, but I wonder if the self discharge rate is a function of how much the battery is charged. For example, if a battery has 1.2 KWh of capacity, and the self discharge rate is 1/2 watt per hour, then in 2400 hours is should be totally dead but that cannot be right cuz that is only 100 days (24/7). The self discharge likely tapers based on SoC.

Here is a pic of four large 6V 230Ah rated AGM batteries in series/parallel configuration (so 12V 460Ah rated). 124mA at 13.5V is only about 62mA per battery (at 6.75V) which is LESS than 1/2 watt per battery of charge power going into each battery. The total charge power is 1.674 watts so that is about 0.42 watts per battery.

For the benefit of those of you wondering, that laboratory DC power supply is Instek brand, model GPS-3030DD. It has a range of 0-30V (actually about 31.4) and current limiting from 0 to 3A (actually about 3.1). It is rated at a maximum of about 90 watts output but can be coaxed into about 100 watts. I usually run it at 1.5A max (about 50% of its max rating) and I usually blow a fan across the heat sink if the charge current is 1A or more.

enter image description here

enter image description here

  • 1
    \$\begingroup\$ Could be the charging algorithm's done criteria doesn't match the charge characteristics of the battery \$\endgroup\$
    – DKNguyen
    Dec 12, 2019 at 2:05
  • 1
    \$\begingroup\$ if Ic=CdV/dt , figure out the initial charge rate dV/dt and compute C and expect T=RC to take 10T to reduce the current charge rate to <10% of initial charge rate. Cutoff is usually 10% of CC rate \$\endgroup\$ Dec 12, 2019 at 3:05
  • 1
    \$\begingroup\$ @David, keep in mind that doing good-precision capacity tests is hard, the capacity of rechargeable batteries varies somewhat depending on temperature and past usage history. This introduces an error in your reading, and if the effect you're trying to capture is small, the errors may easily distort it. My suggestion: whatever tests you do, repeat them several times — otherwise you won't know whether your results are significant. \$\endgroup\$
    – anrieff
    Dec 14, 2019 at 7:54
  • 1
    \$\begingroup\$ Float charge with pulse desulfator is best for longevity and normalizing acid SG, monthly burst ESR and dV/dt = I/C tests for aging will help you to measure aging rate with dV*C = Ah * 3600 \$\endgroup\$ Dec 18, 2019 at 16:51
  • 1
    \$\begingroup\$ Keep in mind for LI ion cells they know they can get more life by being more conservative on capacity and just going between 90%Soc to 50%Soc in total lifetime Ah supplied to load * number of cycles this makes a huge difference whil trying to store last few % capacity increases aging rate or reduces MTBF significantly. \$\endgroup\$ Dec 18, 2019 at 16:58

2 Answers 2


What I am wondering is why does it take so long for that battery to "level off" at about 1/2 watt of charge power going in?

As the battery charges its voltage rises, so the charger has to raise its output voltage to put more charge in. However if the voltage goes too high the water in the electrolyte will break down into hydrogen and oxygen. A sealed battery can withstand a bit of this, but if the gas pressure gets too high it will vent and lose electrolyte.

To prevent excessive gassing the maximum charging voltage is limited to ~2.3V/cell. The battery has some internal resistance which drops voltage proportional to current (Ohm's Law), so once the terminal voltage reaches the maximum allowed the charging current must go down as the battery's internal voltage approaches it. The result is an exponential drop in charging current over time, similar to charging a capacitor through a resistor.

At the same time the battery also self-discharges, due to various unavoidable chemical reactions. This is why the current never quite drops to zero. The combination of charge current reduction and self-discharge results in an exponential curve that levels off at a current above zero. At that point the battery is already at 100% charge, and any further 'charging' current is just maintaining the full charge state.

if 2 identical batteries have 2 different bottom out charge rates (let's say one is 37mA and the other is 40mA at 13.5V), what (if anything) can we conclude from that?

We can conclude that one battery has a lower self-discharge rate than the other. This could be caused by fewer impurities, different temperature, different battery size or a different type of battery (eg. Calcium-lead vs. Antimony-lead). In a 100Ah battery the difference between 37mA and 40mA is only 3mA, which is insignificant compared to its capacity.

since it takes to long to bottom out, does that mean that most (if not all) smartchargers don't do it for speed reasons and if so, does that mean that a battery charged with a smartcharger is not really 100% charged?

Yes. The time taken to put in that last 1% (or 5%, or even 10%) probably isn't worth it.

However 'smart' chargers often have a multi-stage charging strategy that raises the voltage a little higher to speed up the 'absorption' stage. This may cause some gas production (which in a sealed battery recombines once the charging cycle is finished), but must not be continued for too long or the battery will start to lose electrolyte. Once the absorption stage is over the charger either shuts off or drops to a lower 'float' voltage which charges very slowly (if at all). With this technique the battery can get very close to full charge in a reasonable time period.

  • \$\begingroup\$ For the difference in bottom out current, I was referring to same type, age, temperature, capacity, and health batteries, so the difference must be in the impurities/construction quality. Another thing I noticed is sometimes the bottom out rate seems to bottom out, then goes up slightly, then "re-bottoms" similar to the previous bottom (or sometimes it goes lower). I wonder if that is a regulation "problem" on the lab PSU or just the charger going below the self discharge rate and then ramping back up slightly to compensate. It is hard for me to tell what is causing it. \$\endgroup\$
    – David
    Dec 15, 2019 at 22:13
  • \$\begingroup\$ Also, if there is an aggressive final bulk charge voltage (such as 14.6V) for AGMs, then the charger quicly drops to a milder 13.5V trickle/top off voltage, there might actually be negative current flow for a few seconds until the battery voltage drops below 13.5V due to self discharge (and maybe a slight load "backfeeding" the charger). This is partially why I am wondering if an AGM charge algorithm that ramps up (say from 12.8V) to 13.5V in the bulk phase should just hold 13.5V and stay there (since that is the trickle charge voltage). If someone is not in a hurry, that could be ideal. \$\endgroup\$
    – David
    Dec 15, 2019 at 22:35
  • \$\begingroup\$ Perhaps a UPS charge algorithm should do this because of the infrequent use of battery power. It wouldn't need a quick charge in a few hours since the average time to the next power outage might be weeks or even months. A good slow charge (as stated in the above comment) might be ideal for UPS applications using AGMs. The rare chance that there would be 2 sustained blackouts in rapid fire succession can be "overlooked" due to the very low probability of that happening. 2 or more short duration blackouts (or even brownouts tripping battery backup) will still be handled properly and well. \$\endgroup\$
    – David
    Dec 15, 2019 at 22:41
  • \$\begingroup\$ An even better charge algorithm would be to sense the ambient temperature and adjust the trickle charge voltage accordingly. That way you can be better assured that your batteries will eventually be in a TRUE 100% SoC and not this BS 100% some charger manufacturers claim. Manually using my lab PSU usually takes about a week to bottom out so that tells me there is quite a bit more charge to take even after the smartcharger reports "100%". I think they lie like a rug and they are not so smart. By not charging to a true 100%, the battery might eventually reject going into that high SoC. \$\endgroup\$
    – David
    Dec 15, 2019 at 22:43
  • \$\begingroup\$ Some chargers are smarter than others, but there must be a balance between 'intelligence', ease of use, and price. The charger for my lawnmower kept a record of charging time in battery-backed memory. Unfortunately there was a power cut, which caused it to restart from the beginning and cook my battery. I disassembled it and found the backup battery was not connected due to an incorrect part loaded on the PCB! Now I only use an expensive multi-chemistry charger manufactured in Germany. It has to be programmed each time I use it, but I know it won't screw up! \$\endgroup\$ Dec 15, 2019 at 23:14

I will attempt to answer my own question for the benefit of readers. I remember reading somewhere that what is happening internally when a battery is charging is somewhat analogous to musical chairs. In the beginning, as electrons are being forced into the battery, there are many positions where they can make what keeps the battery in a state of charge "attach" to the plates. This is why an AGM battery can be "blasted" with high current (sometimes even more than C/1) while in the bulk charge phase. Another analogy for computer savvy people would be a random number generator with a limited range (such as 1-1000). In the beginning, it is quite easy to generate a "non-seen" random number and then mark it seen (such as we draw 567 first, then 123 and mark that seen second...). However when we get more than about 75% of those numbers already seen, chance are the next random number we pick (from 1-1000 still), will be a repeat (already seen), so that slows down the process.

Yet another analogy is would be filling the gas tank of a lawn mower. If it is pretty much empty, you can pour the gas in there rather quickly, however as it becomes close to 3/4 (75%) full, you need to slow down for fear of it splashing out on the hot engine or missing the full mark and overfilling it.

What I don't really know for AGM batteries is the following: Is it better, worse, or same to slow charge a battery more than 24 hours from say a 50% SoC? Since AGMs do not have to gas (to help de-stratify the sulfuric acid solution), it seems any reasonable charge rate (that gets the battery to at least 75% SoC from 50% SoC within the first 24 hours is acceptable). However I suspect bulk charging to 75% within a few hours and then getting it to 90% or more within the first 24 hours is even better (which is why they do it).

I am also thinking that the reason those last few % are so slow is because the person who owned my used batteries before me maybe never charged them up to a true 100%. After a while, by doing that, the batteries might reject that last 5% or so of charge or maybe it takes extra time to desulfate that part of the battery (the part of the plates that rarely or maybe never had charge). I am not sure if by charging an AGM battery to "95%" each time exposes the same part of the plates as uncharged or if it is more random. That is yet another interesting question out of the scope of this question I think.

My final argument is that smartchargers probably don't put in that last 5% because it takes too long to do that and people don't want to wait that long (generally speaking). Since they go to approximately 13.5V anyway in AGM maintain mode, they are effective still slow charging, so if someone leaves it on there for say 12 hours, they are actually getting MORE charge, even though the smartcharger may indicate "100%" charged long before that.

  • \$\begingroup\$ But a question that this answer may bring up in the reader's mind is how can a smartcharger get that last 5% or so of charge in the AGM battery MUCH more quickly than just trickle charging it for a week? There must be a quicker way without damaging the battery. I wonder what might happen if instead of using the trickle charge voltage for that last 5%, I used maybe 0.3V higher such as 13.8V (instead of 13.5V). This is easy since I have many regulated power supplies that output 13.8V so that would free up the lab PSU for other things (such as 24V banks). Any suggestions? Just try it? \$\endgroup\$
    – David
    Dec 18, 2019 at 16:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.