Li ion 18650 battery discharge

[VEGETA]

Hello,

I have designed a small board which uses 18650 battery to power a very very low power consumption circuit.

I measured the consumption about 0.250 mV across 13 Ohms limiting resistor... meaning about 20 uAs.

The board is designed to have a charger which charges the battery fairly quickly at 100 mA then I disconnected the battery and it is now powering the circuit.

I noticed some rather quick voltage drop for the battery, as follows:

format is TIME -> BATT VOLTAGE

Code: [Select]

Thursday 01-6:
fully charged in-circuit
04:32 -> 4.188 ---> end of charging
05:07 -> 4.184
16:32 -> 4.171

friday 02-6:
11:31 -> 4.145

saturday 03-6:
00:56 -> 4.142
02:21 -> 4.142
04:17 -> 4.142
11:31 -> 4.142
charged for a bit.
14:46 -> 4.144
16:32 -> 4.143
21:41 -> 4.142

sunday 04-6:
18:53 -> 4.137
22:34 -> 4.136

monday 05-6:
01:12 -> 4.136
16:46 -> 4.134

tuesday 06-6:
00:43 -> 4.131

wednesday 07-6:
07:27 -> 4.126
16:43 -> 4.122

I read there is quick self discharge at first but is it correct?


will it settle later on?

[inse]

Have a look at the typical discharge curve of a LiIon  battery.
For most of it’s capacity it is around 3.6V

[tunk]

No expert, but I think it's not initial self-discharge, but initial cell chemical stabilization,
and that it should take seconds or a few minutes. The simplest way to see self-discharge
is to not put a load on it and measure the voltage maybe every day or week. I would think
it's difficult to measure self-discharge in circuit with high accuracy. You could try to find
voltage vs state of charge curves (maybe in the link below), and then calculate how much
the voltage should drop with your load.

https://lygte-info.dk/

[westfw]

You're counting ~0.05V as significant voltage drop?
Did your "charge" cycle include the full constant-current segment, or did it just stop when you reached 4.188V?

[VEGETA]

You're counting ~0.05V as significant voltage drop?
Did your "charge" cycle include the full constant-current segment, or did it just stop when you reached 4.188V?

well, I thought it will probably drop from 4.2 to something like 4 or down to 3.7 over time and settle there for long time until it starts to be depleted from its capacitance.

my circuit taking micro amps should not drop the battery voltage like this each day... is this ok or wrong?


I wanted the battery to last for at least a year in this circuit but if it keeps dropping like this it won't. battery output is delivered to very low dropout LDO to deliver 3.3v then followed by a schottkey diode to final circuit (dreamcast internal cmos clock circuitry to keep time).

[thm_w]

If its a 3000mAh battery, and its drawing 20uA, then it will last years. Self discharge might be 1-3% per month depending on the quality of the cell and the ambient temperature.

As you said you expect it to drop from 4.2 to 4 quickly. Which is what your measurements show. Its not a linear decrease. Plot it out and compare it to graphs you can find online.

[VEGETA]

If its a 3000mAh battery, and its drawing 20uA, then it will last years. Self discharge might be 1-3% per month depending on the quality of the cell and the ambient temperature.

As you said you expect it to drop from 4.2 to 4 quickly. Which is what your measurements show. Its not a linear decrease. Plot it out and compare it to graphs you can find online.

I didn't read about dropping to below 4v anywhere official but rather people writings here and there. battery discharge curves are all for higher currents.

I'd like to confirm that this behavior is normal and expected from 18650 batteries... I mean that after being charged to 4.1 ~ 4.2v they rapidly (or relatively rapidly) drop back to certain level (4v or nominal 3.7v) then stick firmly to it as long as their life is minus when they are so depleted. Is this correct?

I put readings above on nearly 4 days and will continue to log in data.

[amyk]

Is this a new cell, what brand/capacity/model?

100mA is not charging "fairly quickly" for the 1000mAh+ (most are 2000-3000mAh, only high-current cells are lower capacity) that's typical of 18650s--- it's less than C/10. If you do actually charge one at its normal rate of ~C/2 or more (500mA-1.5A), then there is a "saturation" phase where the voltage will reach 4.2V quickly but the cell continues to draw current, and if you don't let it saturate fully the voltage also quickly drops once disconnected. There should be almost no "saturation" at 100mA for an 18650-size cell in good condition, so I suspect this is a used cell with much reduced capacity?

[VEGETA]

Is this a new cell, what brand/capacity/model?

100mA is not charging "fairly quickly" for the 1000mAh+ (most are 2000-3000mAh, only high-current cells are lower capacity) that's typical of 18650s--- it's less than C/10. If you do actually charge one at its normal rate of ~C/2 or more (500mA-1.5A), then there is a "saturation" phase where the voltage will reach 4.2V quickly but the cell continues to draw current, and if you don't let it saturate fully the voltage also quickly drops once disconnected. There should be almost no "saturation" at 100mA for an 18650-size cell in good condition, so I suspect this is a used cell with much reduced capacity?

it is a Chinese cell and it is new. I know beforehand it is not a genuine good brand but for my application it should be nice enough.

charger controller IC is this: https://www.digikey.com/en/products/detail/microchip-technology/MCP73832T-2DCI-MC/1223148

with 10K resistor = 100 mA charging current. I choose this low value to keep cell from being hot + since it is inside dreamcast i didn't want to stress its power supply a lot and consumption itself is in the micro amp so even a brief session of playing should charge the cell a lot. thus there was no need to get it to maximum of 500 mA.

I believe I left it for 30~60 minutes to get from 3.9 to 4.18 or so and it stays at 4.18 right after that. Don't remember exactly how many minutes but it surely wasn't little. I am happy with charging performance since as I mentioned, comparing the charging amount of mere 1 hour of turning it on should give a lot of charge because the discharge is very low to matter.


so once it reaches 4.18v it should be full but you say it should still be connected to charging to make sure it gets full and to never drop fast immediately after disconnection?


that is all nice info but my original issue is about if the fairly quick drop of voltage of 18650 battery in early stage after fully charging... is it normal? i mean regardless of cell being big or small capacity, just in general so i know if what i see is ok or not.

and if it is correct, then what is the "settling voltage"? or the voltage where the cell will drop to then stick to it for most of its life.

[VEGETA]

You've provided effectively 3 datapoints. If you want anybody to say anything meaningful, you need to share more data. Otherwise your post makes zero sense.

well, I asked about the general behavior of li ion 18650 batteries. do they usually lose some voltage rapidly from 4.2v to a certain voltage, then stick to that voltage for most of it life.... is it correct or not?

^ this is my main question which is regardless of my circuit and application.

[wasedadoc]

Is that 20uA measured before or after the LDO regulator?

[RoGeorge]

This is what I've measured on a 10Ah raw cell, has no load and no battery management on it.


Image from https://www.eevblog.com/forum/chat/is-this-a-magic-pot-or-a-self-healing-li-ion-battery/

Today, 52 days after charging, it measures 4.0571V

[VEGETA]

Is that 20uA measured before or after the LDO regulator?

after the LDO.

here are my chips for your reference:

protection IC: BQ29732DSER
charger IC: MCP73832T-2DCI/MC
LDO: MIC5365-3.3YC5-TR
dual mosfet: DMG6968UDM-7
Schottky diode: RB751V40,115


LDO should have 29uA of quiescent current, which is already more than the consumption itself hhh. I measured 0.25 to 0.28 mV across 13 Ohms limiting resistor thus about 20 uA of consumption. protection IC has current consumption of 4 uA so I guess total current consumption could be something like 50-60 uA max.

there are certainly better LDOs which can do nano amp quiescent current such as AP7354-33W5-7 but they are of course a bit more expensive.

Today, 52 days after charging, it measures 4.0571V

well, this is very nice! my current consumption would be close to nothing, so it is safe to assume it will be the same curve right? I mean what is 60 uA to a 2500 mAH battery? should be close to nothing. battery calculator from digikey resulted in about 5 years battery life but this does not factor diode drop and does not factor in that I won't be operating the battery to the lowest cutoff or 2.5v or so... but still very nice. I mean even if it took 2-3 years on single charge it would be perfect.

is there any real gain from switching the LDO to a better one with lower Iq? or would it be just an overkill.

[RoGeorge]

Yes, for such small currents you should see something similar with the self-discharge rate.

Your battery looks fine by the measurements you posted in the OP, might keep its voltage even better than mine.  To double check, you can put your measurements in a spreadsheet, then plot the voltage over time, and the line should look like an exponential decay.

These were the measurements corresponding to the plotted 10Ah cell:

Days   Voltage
0.000   4.1873
0.021   4.1813
0.208   4.1599
0.625   4.1390
3.000   4.1062
5.000   4.0935
19.000   4.0663
34.000   4.0605
52.000   4.0571

[VEGETA]

I will do a proper excel sheet when I get back home for sure. I noticed my measurements are close to yours maybe slightly better. will continue to plot and see how it performs.

till now no one answered the question about what is the settling voltage? for your battery, at 52 days it is 4.0571 volts... interested to see if it will decline more or just stick to this.

[MrAl]

Hello,

I have designed a small board which uses 18650 battery to power a very very low power consumption circuit.

I measured the consumption about 0.250 mV across 13 Ohms limiting resistor... meaning about 20 uAs.

The board is designed to have a charger which charges the battery fairly quickly at 100 mA then I disconnected the battery and it is now powering the circuit.

I noticed some rather quick voltage drop for the battery, as follows:

format is TIME -> BATT VOLTAGE

Code: [Select]

Thursday 01-6:
fully charged in-circuit
04:32 -> 4.188 ---> end of charging
05:07 -> 4.184
16:32 -> 4.171

friday 02-6:
11:31 -> 4.145

saturday 03-6:
00:56 -> 4.142
02:21 -> 4.142
04:17 -> 4.142
11:31 -> 4.142
charged for a bit.
14:46 -> 4.144
16:32 -> 4.143
21:41 -> 4.142

sunday 04-6:
18:53 -> 4.137
22:34 -> 4.136

monday 05-6:
01:12 -> 4.136
16:46 -> 4.134

tuesday 06-6:
00:43 -> 4.131

wednesday 07-6:
07:27 -> 4.126
16:43 -> 4.122

I read there is quick self discharge at first but is it correct?


will it settle later on?

Hi,

In order to make any real decision on what is happening to this cell you would have to know more about the cell, such as model, age, etc., but that would still only be an approximation.  This is why people that deal with batteries on an everyday basis rely on real life tests, not predictions, at least until some initial test results come in.

What this means is that the only way to be sure is to test your actual battery, and if you have another battery of the same type, make, model, etc., you have to test that one as well.  That's the only way to really know what the behavior is.  Batteries are still a bit complicated when it comes to blind predictions, except in a very general sense.

What you have to do then is test your battery.  The usual way is to apply a load known to cause it to discharge at the usual test rate of C/20 amps.  So if you have a 2000mAh battery you test it at 2000/20=100ma.  This current level is not really mandatory though, but that is the standard test current and the resulting capacity measurement should roughly match the cells advertised rating.
If you happen to have a device that draws 500ma, you can test at that just to see how much run time you will get, but the results of that test may not match up with the cells advertised rating.

Now we come to the lower current tests. To know what the capacity is at 10ma, you would have to test at 10ma.  To know what the capacity is for 1ma, you'd have to test at 1ma.  We could go on and on here, but you see the point.  If you want to know what the capacity is at 50ua load current then the only way to be sure is to test at 50ua.

There is another way but it's still an estimate, and you still have to do some tests and measurements.  That is to test for the P factor for your battery.  Once you know the P factor you can estimate the capacity at other test currents.  This still requires at least two tests and two sets of measurement data, so it may be a waste of time after all.  Once you have this factor you can use it to predict the behavior for different test currents, to some degree of accuracy.  The unfortunate thing is that the P factor will change with age, so to get it right you'd have to do these tests all over again as the cell ages.  It's probably not worth it that's probably why not many people know about this technique.

The hard and proven way, as mentioned, is to test it yourself, and log the data which you are now in the process of doing already.  It goes a bit farther than that too though.  The idea is to perpetually log the data over the entire life of the battery, and while you are doing that, analyze the data for different behaviors.  This requires letting the cell discharge using the actual load, then recharging, then letting it discharge again, etc., all the while compiling the data sets.  For each full run you will get a set of data.  The first run will tell you how the cells behaves when it is brand new.  Subsequent runs will tell you how the battery ages.  From that you get a really good idea how long the battery will last and when you have to buy a new one.
This is kinda what we all do with our cell phones too.  We check out the battery indicator now and then, and charge it when needed, and over time if we see it going down too fast we know we need a new battery.

So, it looks like you have a long road ahead of you, but it's an interesting road.  I've followed it many, many times in the past and am currently doing it right now with my automobile battery using some home made telemetry.  The device in the car measures the car battery voltage and transmits the reading to a receiver in the house and the receiver sends it to a home PC computer and that computer logs the voltage 24 hours a day, 7 days a week, creating a very long log of date, time, voltage and some other information.  This is also to monitor the solar charger i have connected to the car battery.  I can see the battery voltage go up when the sun is shining and go down at night, then back up in the morning, etc.  It's interesting in and of itself really, with the benefit is I can see what is happening and predict certain things about the battery.

Good luck with your quest, and when you have more data maybe you can come back and report that and we can talk about what is going on and make some rough estimates until you get a big pile of data at which time we will know quite well what is going on.

Oh, BTW, the P factor comes from Peukert's law which you can look up if interested.

[VEGETA]

as you mentioned, I am doing the actual test and logging data.

the thing is, I am a beginner at 18650 li ion battery stuff. thus i saw the initial rapid drop in voltage and got confused.


the whole thing about this thread is to verify if this behavior is normal or not. it is not about how to predict or so... if what i called settling voltage for my battery is say 3.9v and someone else is 4v it does not matter at all to me... all that matters is that both our batteries dropped the same way first then stuck to this settling voltage.

I remember my battery bought new (don't know how much time it stayed on seller's rack) was about 3.9 ~ 3.95v so I can assume for now that this is going to be the settling voltage and I am continuing to log.

the battery is crappy Chinese... I didn't find genuine battery and being crappy battery also helps seeing the board performance with such bad batteries, to be sure it works with all,.

[amyk]

There is no "settling voltage". Due to unavoidable self-discharge, the voltage will not stop dropping until the cell is eventually empty.

[VEGETA]

There is no "settling voltage". Due to unavoidable self-discharge, the voltage will not stop dropping until the cell is eventually empty.

i know but how much drop daily? monthly? this is the issue here.

meaning, according to my and george's result... there is rapid loss early on then it settles. settling does not mean 0mv loss but rather very slow loss over long time.

i just want to make sure i understand the behavior of such batteries to verify if the circuit will last the long time i need it to.

[alligatorblues]

Ebay has an electronic load for 13.00, which is perfect for testing batterie. I think you can go down to 0.1mA, and up to 12 v: https://www.ebay.com/itm/264782677333?hash=item3da6470555:g:WTgAAOSwhbZe~ZRL&amdata=enc%3AAQAIAAAAwF%2B4I9YM0YnbBUEKdVzdAxmPohqcxMHxFb%2BKvDTmwkthIzS769rLbU9FjIlWosMy4nN2Es3GTOWVlS6jFnfqpLNdGkUxTAJytZ4TsgI6iYvmHfzXR2ObGD0%2BFE7i9WIQ%2B1livOY%2FZXlYvElJfJWMAHO3i8XBgV9skRlbXdOVjsrv1zQlPfHmJlu7R%2FpcLAY8VKtY%2BzkKFNHJPRmU584QTXp%2BT%2BBFyvq41NrS321ZRU5FAvduN%2FmGuLF6qTdVWPNOFw%3D%3D%7Ctkp%3ABk9SR_7ZvN2TYg

[VEGETA]

Ebay has an electronic load for 13.00, which is perfect for testing batterie. I think you can go down to 0.1mA, and up to 12 v: https://www.ebay.com/itm/264782677333?hash=item3da6470555:g:WTgAAOSwhbZe~ZRL&amdata=enc%3AAQAIAAAAwF%2B4I9YM0YnbBUEKdVzdAxmPohqcxMHxFb%2BKvDTmwkthIzS769rLbU9FjIlWosMy4nN2Es3GTOWVlS6jFnfqpLNdGkUxTAJytZ4TsgI6iYvmHfzXR2ObGD0%2BFE7i9WIQ%2B1livOY%2FZXlYvElJfJWMAHO3i8XBgV9skRlbXdOVjsrv1zQlPfHmJlu7R%2FpcLAY8VKtY%2BzkKFNHJPRmU584QTXp%2BT%2BBFyvq41NrS321ZRU5FAvduN%2FmGuLF6qTdVWPNOFw%3D%3D%7Ctkp%3ABk9SR_7ZvN2TYg

I bought this one: https://www.amazon.com/gp/product/B07HB3XK7L/ref=ppx_yo_dt_b_asin_title_o00_s01?ie=UTF8&psc=1

[radiolistener]

I noticed some rather quick voltage drop for the battery, as follows:
I read there is quick self discharge at first but is it correct?

No, this is not self discharge. When you apply charge voltage to the battery it leads to activate internal chemical processes. If you disconnect the charger it will not stop these processes immediately and requires some time.

will it settle later on?

Yes, if you want to measure EMF voltage of the battery, you're needs to wait for at least 2-10 days after disconnection from charger or from high power consumption load.

For example lead acid batteries has standard EMF 12.7 V. And it's stop charge voltage is 14.4 V. When lead acid battery charge is completed and you disconnect it from the charger it's voltage is still keeps above 13 V for about 1 day.

I'm not sure what is standard EMF voltage of a fully charged Li-Ion battery, but this is definitely not 4.2 V. Note, that 4.2 ± 0.05 V is a stop charge voltage, but not EMF voltage.

I wanted the battery to last for at least a year in this circuit

if you want to extend your Li-Ion battery life, just avoid to charge it to 100%, because it will leads to a quick shortening battery capacity and decrease it's lifetime. In order to extend Li-Ion lifetime just do not store battery charged above 4.1 V for a long period without discharge.

The safe voltage to store Li-Ion batteries for a long time is about 3.8 V. It will minimize battery capacity degradation.

Also note that military and long-life durable Li-Ion batteries has a stop charge voltage about 4.1 ± 0.05 V which is less than standard 4.2 V, this is because they needs longer lifetime. 

Some smart HP notebook chargers even detect that notebook is connected to charge for a long time and starts to not charge it above 80% in order to extend battery lifetime. If you disconnect charger and connect it again, it returns to a normal charge up to 100%.

i know but how much drop daily? monthly? this is the issue here.

When you disconnect the battery from a charger, it's voltage drop include sum of both - settle time due to chemical stabilization and self discharge. Both processes depends on battery manufacturer, internal structure, capacity, battery age and other things. So, there is no answer how it will drops down, different batteries have different curves.

But voltage drop due to chemical stabilization will be minimized within 2-10 days and after that you can just ignore it. While voltage drop due to self discharge will continue for a years.

[MrAl]

as you mentioned, I am doing the actual test and logging data.

the thing is, I am a beginner at 18650 li ion battery stuff. thus i saw the initial rapid drop in voltage and got confused.


the whole thing about this thread is to verify if this behavior is normal or not. it is not about how to predict or so... if what i called settling voltage for my battery is say 3.9v and someone else is 4v it does not matter at all to me... all that matters is that both our batteries dropped the same way first then stuck to this settling voltage.

I remember my battery bought new (don't know how much time it stayed on seller's rack) was about 3.9 ~ 3.95v so I can assume for now that this is going to be the settling voltage and I am continuing to log.

the battery is crappy Chinese... I didn't find genuine battery and being crappy battery also helps seeing the board performance with such bad batteries, to be sure it works with all,.

Hello again,

Well, you see, that actually is the art of prediction.  In order to know how a particular battery behaves without doing the right tests you would have to be psychic.  You can't rely on opinions from any source that does not have the battery in hand because there could be something unusual about your battery that no one knows about yet.  The only way to be sure is to test it, and that's the best way anyway.  Once you do the test you will have no doubt whatsoever about how the battery behaves not only in the short term but also in the long term.

There's an old saying that applies here i think.  That is, "One test is worth a thousand expert opinions".

It's always your choice but you can see that after all these replies you still don't know for sure 
Also, I've been testing batteries for over 50 years and currently am doing the same as you and have been over several months now.

Here's another interesting story...
I had a car battery one time that i tested over about two months and found that it had a high self discharge.  Since the battery was still under warrantee, i took it back.  The place where i bought it put a tester on the battery and the tester said the battery was good.  The tester device was made by someone who thought you can test a battery with one three second test.
Here's the kicker:  The guy said the battery was good due to the tester results, but still replaced the battery with a new one for me.  The question is, why did he use the tester if he wasn't going to believe its results.  That's because he knew that you can't test a battery with such a simple test, UNLESS the battery is very bad, and then it will tell you it is bad.  You can test a car battery with a heavy load and see how far the battery voltage drops and that gives you an idea about the ability of the battery to start the car.  That's a quick test that works but only if the battery is really bad.  If it tests 'good', then you really don't know if there is another issue with the battery.

If you prefer though of course you can keep asking questions about this, you'll learn more and more about batteries and how complicated they really are.
It's been interesting for me to learn about these devices too over the years, and i was surprised at first to find that batteries are very different from other components like resistors, capacitors, and inductors, although of those three inductors are not really that simple either but are often taken to be simple due to some silly test devices out there.

Ok, so this is reply #23, let's see how many more this is going to take 

[radiolistener]

The good way to check battery state is to check it's internal resistance. You can do it with measurement it's EMF voltage and voltage under some load:

R = (Uemf - Uload) / (Uemf / Rload)

But note, that internal resistance also depends on discharge current, so you're needs to use load with high enough current for testing.

[tunk]

i know but how much drop daily? monthly? this is the issue here.

Basically, you cannot know. It depends on the cell, if it's a
new cell from a reputable source then it may be a few tens of
mV per month (maybe even lower). An old, abused cell may
completely discharge in days. And you never know with cells
from eBay, aliexpress etc. - they could be reasonably good,
they could be recycled cells, they could have a fraction of the
stated capacity, and/or total junk.

[BILLPOD]

I have found that the best way, (for me), to charge, discharge and test cylindrical batteries is with the SKYRC MC3000 battery charger.   It does ALL the chemistries and most battery parameters can be measured.
Some may consider it overpriced, but for what it does, is worth the price.
I operate it through a USB cable to my laptop, but it can be programmed using the 8 push buttons on it's face.   I especially like it for categorizing my large collection of 18650s which I have harvested from laptop batteries.   Here is the Amazon USA listing, but it may be found cheaper elsewhere

[Peabody]

there are certainly better LDOs which can do nano amp quiescent current such as AP7354-33W5-7 but they are of course a bit more expensive.

is there any real gain from switching the LDO to a better one with lower Iq? or would it be just an overkill.

Well, you would come close to doubling the life of the battery.  The better LDO will cost four or five times as much, but still not a lot of money in absolute terms.  So I guess it depends on whether you will be making a million of these devices, or just one.  The MCP1700 is another one you might look at that might be more readily available.

[VEGETA]

there are certainly better LDOs which can do nano amp quiescent current such as AP7354-33W5-7 but they are of course a bit more expensive.

is there any real gain from switching the LDO to a better one with lower Iq? or would it be just an overkill.

Well, you would come close to doubling the life of the battery.  The better LDO will cost four or five times as much, but still not a lot of money in absolute terms.  So I guess it depends on whether you will be making a million of these devices, or just one.  The MCP1700 is another one you might look at that might be more readily available.

your suggested MCP1700 is very good. ground current is about 4 uA and quiescent current about ~2uA so total would be < 6 uA compared to 29 uA from before. it is 3 times the cost though but still good enough as the board will not suffer from < 0.3$-0.5$ price increase.

I will make say hundreds and sell them, in batches according to demand.

However, in this experiment I am doing, I am seeing if the cheap (and still very good) LDO of 29 uA will be that much effective or not.

if we assume total current consumption from everything is just about 60 uA total, then what will that be for a 18650 battery? my assumption is that it will be literally nothing. if so then changing the LDO won't be much of a change.

yesterday at this time it was 4.112v and now it is 4.110v so a 2 mV drop which is less than before where it dropped by more than 5 mV, and very early was more than that.

to my understanding from explanation above about chemical characteristics... this should be normal and will settle at some point, after that will be mere self discharge + my 50-60 uA which in reasonable assumption won't be dropping battery voltage in any effective matter.

I would be very happy if the battery will last for a year, which by all reason and calculations it should do much more than that.


 

[radiolistener]

yesterday at this time it was 4.112v and now it is 4.110v so a 2 mV drop which is less than before where it dropped by more than 5 mV, and very early was more than that.

This is not a drop but just a noise due to chemical processes inside battery. Don't be surprised if the voltage can rise by 2 mV in some conditions.