WARNING: Autopsy of a Chinese LifePo4 Battery

[Zash]

That's also good point Re: charge voltage, however, you can see that the voltage on the x-axis is different depending on what voltage the pack was charged at, vs. the remaining charge. Further studies are obviously needed.

The Reaktor does indeed seem to be a good deal.

Is there something that will charge 1-16s? I have come across 2s-16s bot nothing that will also charge individual cells. If I am going to be paying a premium for a 1s with a redundant 6s capacity, may as well get one which is able to charge my full 16s pack as well as 1s.

I see something like this, sadly only 2s-16s:

http://www.litrade.de/shop/BMS-Balancer/BMS-Systeme/Akkumonitor-Chargery-BM16-BMS16-V1-0-4-0-bis-16-Zellen-Kapazitaetsanzeige.html

 

[Dauntless]

I got stupid and fell for some of this crap myself.
Here is an article i wrote detailing how screwed i got:
http://neptronix.org/wordpress/dont-buy-unknown-no-name-chinese-batteries/

So the article didn't give the final result. Did PayPal come through? New problems from the guy?

I'm just scared of any serious effort with lithiums. I use them in video and never a problem, but that's never some large collection of cells in use at the same moment. But I bought one new battery for one of my cameras that lasted weeks. Bowyer.

 

[Hillhater]

That's also good point Re: charge voltage, however, you can see that the voltage on the x-axis is different depending on what voltage the pack was charged at, vs. the remaining charge. Further studies are obviously needed.

The Reaktor does indeed seem to be a good deal.

Is there something that will charge 1-16s?

I think you are mis understanding that graph.
The x axis shows the max voltage the cell is charged to ,..which will also be the resting voltage before the discharge test.
The thing to notice is the huge difference in cell capacity (Ah) between 3.3v and 3.4 v, with very little extra capacity above that ,..even up to 4.2 v...that is typical of Lifepo.
16 s charging (55-60v?) is usually done with a bulk charger and using a BMS to balance cells.
You could always use dual 8s chargers if you really want to.

 

[Zash]

@Dauntless
I wanted to ask the same question, but then realised that Paypal would have refunded him outright, as it normally would do in this scenario. It's sad that you have to take this route to get your money back, when the seller could have given a fraction of that in the first place and settled the matter.

@Hillhater
"I think you are mis understanding that graph."

No, I think you are misunderstanding it mate. For example, if you look at 3.4v charge vs the 4.2v charge, there is a 0.1Ah difference. Now, if you move the 3.4v line by 0.1Ah on the Y axis, you do not get the 4.2v line. If what you say was true then you would.

Regarding the chargers, indeed I know they are difficult to come by, giving an example of one which was actually found. Hence why I asked if anyone knows if other variations are available. 2x8s adds the complexity of splitting the balancing wires and having to add charge wires and the effort of managing them.

$50 is a bit much to pay for a single cell charger if I don't want to use any of its other functions

 

[Hillhater]

......if you look at 3.4v charge vs the 4.2v charge, there is a 0.1Ah difference. Now, if you move the 3.4v line by 0.1Ah on the Y axis, you do not get the 4.2v line. If what you say was true then you would.

...i dont understand what point you are trying to show ?
The graph is simply a capacity plot for different charge voltages
Simple fact is, 80% of the capacity is between 3.3 and 3.4 volts, and no point in charging above 3.4v because the is no worthwhile capacty to be gained,
Also, you need to have a very sensitive volt measurement to quantify capacity accurately in that range.
At 3.4 volt charge level , 75-80% of the total capacity is covered by less than 0.1 volts.

 

[Zash]

...i dont understand what point you are trying to show ?
The graph is simply a capacity plot for different charge voltages
Simple fact is, 80% of the capacity is between 3.3 and 3.4 volts, and no point in charging above 3.4v because the is no worthwhile capacty to be gained,
Also, you need to have a very sensitive volt measurement to quantify capacity accurately in that range.
At 3.4 volt charge level , 75-80% of the total capacity is covered by less than 0.1 volts.

You're missing the 3rd dimension mate, and also moving the goalposts. There is no need to be disingenuous.

"The graph is simply a capacity plot for different charge voltages"

Firstly, no; the graph is showing much more than that as a third dimension. It is in fact showing capacity plot for different charge voltages vs. voltage reading of the cell during discharge. This last piece of information is critical to our little puzzle.


Secondly; the question isn't "does charging at a higher voltage increase capacity", as you well know. Rather, the question is, "if charging cells using different charge voltages, can one reliably predict the charge level of one cell if knowing the charge level and voltage of another cell which was charged at a different voltage?". The answer to this question is that different charge voltages will result in different voltage readings during comparable percentages of cell discharge.

Thirdly; even the cheap voltmeters can read up to 3 decimal points with varying accuracy. Accuracy here is less important compared to precision. The actual voltage reading is not as useful as knowing the difference in voltage. Even cheap mutimeters can be inaccurate, but are generally very precise.

Modern Lithium battery indicators which show charge remaining indeed utilise the small voltage difference to show the difference in capacity. Saying that you need high sensitivity is again disingenuous and not accurate to real life as even cheap multimeters go to 3 decimal points. It's not high accuracy, it's fairly standard accuracy. If it's good enough for the electronics industry such as use in mobile phones, it's good enough for me.

Fourth; I have had to explain things in numerous posts to you, but you seem to deliberately overlook my posts or misinterpret what I say, as well as give out bad information. Earlier you implied that all lithium batteries are the same when it comes to safety, and got yourself mixed up with battery chemistry and battery packaging. Now you pretend that battery voltage level does not indicate charge level to the contradiction of the entire electronics industry - Please tell it to your phone's battery level indicator. I regret that I will not be responding to your comments any more in this thread as it is both a waste of your time and mine. :wink:

Good day, sir.

 

[amberwolf]

Is there something that will charge 1-16s? I have come across 2s-16s bot nothing that will also charge individual cells. If I am going to be paying a premium for a 1s with a redundant 6s capacity, may as well get one which is able to charge my full 16s pack as well as 1s.

A useful item that can be used for this is an adjustable-current, adjustable-voltage power supply, commonly called "lab PSU", or "laboratory PSU". They're not a charger per-se, but they work well for monitored charging of either single or series-connected cells.

I have old Sorensons that I use for this; I just set the voltage I want, then set current to zero, then connect to the cell or cells, then bring up the current until either it is at the limit I want for that cell/pack, or the voltage comes up to the limit I'd already set.

I use them most often for single-cell charging either within an existing pack (usually during repairs) or pre-charging and balancing cells prior to assembly.

There's a lot of brands of these things; none of them are super cheap, but you can find used old Sorensons and similar ones on ebay;etc from to time. The cheapest ones are the huge old transformer-based ones, rather than the smaller, lighter SMPS types. They can also be found at old electronics junkyards, but some of those places think they can get a premium for such things, and would rather sit on htem for decades than sell them for cheap and move them out.


Sorry I don't know enough about available individual cells and reliable places to get them, but you migh tcheck with Stortronics
https://www.stortronics.com/searchresults.asp?cat=1865
http://www.a123batteries.com/product-p/amp20m1hd-a.htm
At $80/cell, it's not exactly what I'd call a bargain, though.


There may also be some members here on ES selling these cells, or packs made from them (probably used, but possibly in better condition than what you have?). I'd look around the Items For Sale - New and - Used sections for "A123" and "EM3EV", as the latter company used to make packs from the A123 20Ah cells, and now and then someone sells such a pack.

 

[Hillhater]

thanks for the friendly encouragement, i guess you are having a bad time with those mystery cells !

"The graph is simply a capacity plot for different charge voltages"

Firstly, no; the graph is showing much more than that as a third dimension. It is in fact showing capacity plot for different charge voltages vs. voltage reading of the cell during discharge. This last piece of information is critical to our little puzzle.
.."you can see that the voltage on the x-axis is different depending on what voltage the pack was charged at, vs. the remaining charge."
Secondly; the question isn't "does charging at a higher voltage increase capacity", as you well know. Rather, the question is, "if charging cells using different charge voltages, can one reliably predict the charge level of one cell if knowing the charge level and voltage of another cell which was charged at a different voltage?". The answer to this question is that different charge voltages will result in different voltage readings during comparable percentages of cell discharge.

..?? look at your graph again.
what difference in cell voltage between 10% and 80% discharge is there, between the 3.5v plot and the 4.2v plot ??.....NONE
sure the 3.4v plot is different, but without knowing the source or methodology of these tests, there could be several explanations for that result.
further, if you believe the graph is in anyway representing your mystery cells, ( which incidentally are definitely NOT A123), then you should realise that the cells charge to 3.3v resting are < 20% charged....IE effectively uncharged.!

. Earlier you implied that all lithium batteries are the same when it comes to safety, and got yourself mixed up with battery chemistry and battery packaging....

no confusion here, and i still say all Lithium batteries no matter what packaging shape or chemistry, can be dangerous, especially in the hands of inexperienced users

Now you pretend that battery voltage level does not indicate charge level to the contradiction of the entire electronics industry - ....

did i ?
i thought i said...
""Two identical cells, of the same capacity , condition, and age, with the same voltage (precisely measured) will have the same discharge capacity available..""

.Please tell it to your phone's battery level indicator..

[/quote]
But you are ignoring the difference in discharge voltage profiles for different lithium chemistry's.
my phone, my laptop, my power tools, my Ebikes, etc all use a chemistry with a much wider voltage range between full (4.2v) and empty (2.5v) with a much more linear change, hence much easier to monitor than Lifepo
i wouldnt claim to know how modern devices monitor battery capacity, but the ones i assemble for my Ebikes, use a Watthour meter to accurately show capacity for both discharge and charge. Others prefer to just use an Amphour meter
I would be surprised if anyone using Lifepo for serious use relied on a simple voltage measurement to accurately monitor capacity.

 

[Zash]

^^

...Anyway,

I have received and installed 2x 8s individual LiFe cell monitors, as well as a proper charge/discharge capacity monitor:

As expected, the 2 battery monitors use battery voltage to gauge battery capacity, and show it in a bar graph as well as the actual voltage. These do a good job for that purpose. I have soldered dip switches in-line so as not to cause battery drain when in storage.

I also obtained a touch screen 1-6s LiFe charger for properly charging individual cells:
[youtube]JIXxTHllUrQ[/youtube]

Using this setup, I fully charged the battery pack and ensured balance, and then drained it to BMS automatic cut-off. The charge meter showed ~15Ah of drain at this stage. The individual cell meters showed that indeed the punctured cell at #12 was the culprit at 2.2v with every other cell still retaining good charge of ~3.3v.

Great, so now we know the charge capacity of the punctured cell. But what about the remaining cells, and especially the cell that we replaced at #2?

I then individually charged cell # 12 by ~5Ah, and left the apparatus to drain again.

The picture itself is of the pack about to be depleted after ~19-20Ah total discharge (showing ~4Ah on the drain meter after being cut-off and reset at 15Ah). As hypothesised, the replaced battery #2 is working very well, having been drained the full ~20Ah, as well as, surprisingly, every other cell. The picture shows that cells #10, 11, 15, have the lowest capacity, and will soon cause the BMS to shut down the power supply.

So, through our investigation the following observations and conclusions can been made:

--The original 16s "20Ah" battery pack had an actual charge capacity of ~5Ah due to cell #2.

--Replacing Cell #2 gave us a capacity of ~15Ah, cut short by punctured cell #12.

--Charging punctured cell #12 by ~5Ah allowed a further battery drain of ~5Ah, meaning that aside from Cells #2 and #12, the pack could indeed hold a 20Ah charge.

--Charging voltage does have a significant impact on the use of voltage to indicate remaining charge. Professional charge level indicators are designed to be used to measure the voltage to show the remaining charge as long as all cells were charged with the same charge-voltage. They offer good results, as shown by the 2x 8s LCD screens to the left.

My personal conclusion is that I shall keep an eye on cell #12. If you recall, I had sealed the puncture, so hopefully, further damage has been stopped. If capacity of this cell continues to decline further than 15Ah, I shall replace this cell as well, leaving us with a 20Ah capacity, if and when required.

All in all, we have gone from 5Ah capacity to a 15Ah Capacity by replacing a bad cell at the cost of ~$25. We have also investigated and found that we can further maximise capacity to 20Ah by replacing the #12 cell as and when needed for another ~$25. Of course, this all assumes that the cells won't degrade down the line.

Conclusions regarding overall thread:

1. Don't buy cheap battery packs from China - they are more trouble than they are worth. However, buying individual cells and building your own pack after testing individual cells is good value.

2. If you take the chance, immediately test for capacity, and return to obtain refund at the slightest sign of trouble.

3. It is a good idea to install both a charge meter as well as an individual cell meters for close observation of the cells. This will eliminate any guess-work, and will warn you of problems well in advance to minimise down-time (shipping from china can take up to a month or more).

I'd be happy to answer any questions or do further investigation if anyone is curious about something else.

 

[Sunder]

^^

Wouldn't roll my eyes too much at him. He's right. Li-Ion and LiPo have a steep enough curve, that small variations don't matter that much. So your phone is 80%, or 75%, not a big deal. But for LiFePo4, it might be taking broader stabs like, somewhere between 80% and 40% depending on age and temperature of the battery.

That's because temperature and even air pressure can have tiny impacts on voltages, which would have large impacts on estimated SOC. In other words, in a 25*C, 40% RH lab at sea level, you could have repeatable and usable results. It's just too much guess work and trying to compensate for the local environment outside of it though. The preferred way to measure LiFePo4 SOC is still a energy meter.

Either way, glad you got it fixed.

 

[Zash]

I think you too are confusing accuracy with precision.

I'll illustrate with a diagram:

We are not trying to ascertain charge by measuring voltage alone - this is accuracy, and granted, this might be tricky.

We are trying to ascertain the difference in charge when we know the voltage and charge of one cell, and only the voltage of another cell in exactly the same conditions - the cells are literally next to one another i.e. we are interested in the precision. The 2x 8s LCDs are designed by professional engineers specifically for this purpose and their product works well. It can be argued that it works fairly well even for Accuracy.

Furthermore, above 3V, all the bars of non-defective cells are about exactly the same size and same voltage until reaching the end of full discharge, at which point, of course some bars and voltages start to deteriorate rapidly: cells #10,11,15 as above.

This suggests that the professional electronics engineers know what they are doing, and the product works as per design - If it didn't work, both the bars and the shown voltages would be wildly different compared to one another throughout the discharge cycle. This simply is not true - the bars and voltages are consistent relative to one another, unless one cell is charged at a different charge voltage or the cell is defective. See the above picture of the 16 cell bars and their voltages yourself.

If one wants to argue about the validity of the measurements being out by "somewhere between 80% and 40%" then one would be arguing both with the professional engineers, their working products, and the experimental data as shown above. If you want to argue that, that's your prerogative, and I won't argue with you.

I would however, personally side with the professional engineers and their products which are being used and are well tested world wide, instead of strangers on the internet.

Thanks for the good wishes

 

[miro13car]

Am I missing something here?
Where is "A123" mentioned in description of this battery?
Why do you post picture of A123 20Ah pouch cell?
Correct me if I am wrong.

 

[Sunder]

If one wants to argue about the validity of the measurements being out by "somewhere between 80% and 40%" then one would be arguing both with the professional engineers, their working products.

I would however, personally side with the professional engineers and their products which are being used and are well tested world wide, instead of strangers on the internet.

Great. Show me a LiFePo4 state of charge meter that uses voltage as the primary means of measurement.

Oh. You can't? Funny that.

You claim false authority based on misunderstanding. You confuse LiFePo4 with other chemistries then claim those products prove your case when they do nothing of the sort.

 

[Hillhater]

Am I missing something here?
Where is "A123" mentioned in description of this battery?
Why do you post picture of A123 20Ah pouch cell?
Correct me if I am wrong.

No correction needed. You are spot on.
This whole thread is like a bottle of cheap wine...
.....it looks normal, but actually , when you open it, you realise it was a mistake.
... its so full of crap , its best avoided all together.!

 

[Zash]

Your argument is a "Straw Man" fallacy, and it is called a fallacy for a reason.

Wow :lol:

You have moved the goalposts sunder, and are deliberately misconstruing the information which I have literally just explained to you in my previous post
This is simply disingenuous.

I have never said that the voltage alone can tell you the charge - it is you who state this to win some imaginary argument on the internet when you realised you had been proven wrong. In fact, I go to lengths to educate you otherwise in my previous post.

For what it's worth:

LiFe battery monitors using voltage only:
1. http://www.isdt.co/bc-8s.html
2. http://gwl-power.tumblr.com/page/112
All this from a 2 second google search. The design engineers of these meters are extensively qualified and have far more knowledge than you - I guess the truth hurts

You will also note that I have a charge counter installed too, both of which I show in a picture and mention explicitly in my post. It's literally right there


@ Hillhater:
Literally the first line in my first post:
"48V 20AH consisting of 16 A123 style cells in series with a BMS." I have never at any point said they were A123 cells. I'll edit my first post to highlight "style" as it seems some people have a hard time with reading, and only pay attention to nice colourful pictures posted for reference as opposed to the body of text.

This whole thread is like a bottle of cheap wine...
.....it looks normal, but actually , when you open it, you realise it was a mistake.
... its so full of crap , its best avoided all together.!

Stay classy mate, glad I didn't waste my time on the likes of you :wink:


It's sad that people take offence once shown to be incorrect, but then this is what seems to be standard for internet forums nowadays.

If anyone except hillhater and sunder has any legitimate questions or want some more investigation, I'd be happy to oblige; otherwise, I'll consider my investigation and fix a success and move forward

 

[Sunder]

Wow :lol:

You have moved the goalposts Sunder, and are deliberately misconstruing the information which I have literally just explained to you in my previous post
This is simply disingenuous.

For what it's worth:

LiFe battery monitors using voltage only:
1. http://www.isdt.co/bc-8s.html
2. http://gwl-power.tumblr.com/page/112
All this from a 2 second google search. The design engineers of these meters are extensively qualified and have far more knowledge than you - I guess the truth hurts

1. If the only two people to respond to your writing, both "deliberately misconstr[e]" it in the same way, perhaps the fault lies with the author, not the reader. Not to mention you've confused the hell out of someone else.

2. You've linked a product (BC-8S) that doesn't show state of charge, as far as I can see, only per cell voltages.

3. The product explicitly says that its accuracy is to 5mv. If you look on the table here:

There's less than 5mv difference between 90% SOC and 80% soc.

Not to mention that a 20*C temperature difference can easily account for 5mV difference at the same state of charge:

Mate. You're wrong. It's that simple.

It's sad that people take offence once shown to be incorrect, but then this is what seems to be standard for internet forums nowadays.

It's odd to speak of yourself in the third person...

 

[Zash]

Ok gentlemen, it looks like we have successfully finished. I have recharged the pack, went for a test run, and everything is working well.

I can be reached via PM/email for questions.

AFAIK, this is the only thread dealing with diagnosing and fixing Chinese LifePO4 battery packs on the internet. I hope this thread is useful to someone down the line; I'll leave the images etc up for posterity, and might update if something else happens to the pack.

I bid you all adieu!



P.S., sunder, thanks for the laughs mate :lol:

1. Zash: 2 identical batteries with identical SOC will have identical Voltage reading.

2. sunder: You can't tell the state of charge by measuring the voltage!!111

3. Zash: We are not using the voltage to tell the SOC. We are saying that if battery 1 has voltage X and SOC Y, then an identical battery with voltage X will also have an SOC Y under identical conditions i.e. we are interested in the "precision" of the readings/meter, not the "accuracy".

4. sunder: Bu...bu...but you can't tell the state of charge by measuring the voltage!!111 Also, the meter is not accurate!!111

5. Zash:

This will probably be a waste of my time again, however, I'll try to educate you one more time mate:
1. Take any cheap, inaccurate voltmeter with bad accuracy to something horrible like +/-100mV.
2. Now, measure the voltage on a battery 10 times consecutively, and write the readings down.
3. You will note that the reading is exactly the same on all 10 readings, not 200mv out, and in fact not even 1mv out.

In other words, a 1.500v source might read on the inaccurate meter as 1.600v; but dang nabbit, it will read all 1.500v sources as 1.600v.

You know why this is? This is because the meter may not be accurate; it is, however, perfectly precise, as all modern digital electronics tend to be. This is the difference between accuracy and precision.

Now, I know I'll probably lose you at the next step, but here we go: Regardless of the accuracy of the inaccurate meter, if you measure the voltage on a cell and it says 3.300v, and you know that its SOC is 50%. We can measure another identical cell at the same time lying right next to it and if it also says 3.300v, then there is a good chance that it too will have SOC of ~50%.

Understand?

No?

Oh well. As they say, you can lead an ass to water, but can't make him drink! :wink:

 

[999zip999]

Hey I wasn't thirsty. But can use a drink. Aka jackass please no rocks.

 

[999zip999]

Get every cell to 3.45v-3.50. I have 1,026 cycles 4 yrs.

 

[amberwolf]

AFAIK, this is the only thread dealing with diagnosing and fixing Chinese LifePO4 battery packs on the internet.

You might want to look around this forum some more. There's a number of such threads.

Most of them deal with Ping brand batteries, but there are also some for EM3EV's packs (some of which are pouch cell and some of which are cylindrical cell). There's a few unbranded (or unknown brand) packs, mostly off of specific-model (or unknown-model) ebikes or scooters.

Feel free to read around the forums for them--there might be information you could use.


I'm sure there's threads on other forums and websites as well, but I've not looked to make sure (that's up to you if you want to know).

 

[miro13car]

if you know something about A123 20Ah pouches shown in your first post you should have know at first glance at your Aliexpress battery specs that it has nothing to do with A123 maybe except that cells are pouches.
A123 is not the only one who manufacture pouches.
"max continous discharge 30A - ha , ha not even close to A123 pouch

 

[Sunder]

Oh well. As they say, you can lead an ass to water, but can't make him drink! :wink:

Oh boy you're ignorant, arrogant and obstinate.

Here we go:

If you cannot reliably tell SOC within 10% (5mv), then you cannot say two packs that are within 5mv (let alone 100mv), have the same SOC.

The argument about actual SOC is just a proxy for your argument.

The above argument also assumes identical cells, which don't exist outside of theory. If you look at the chart above, two cells in the same BATCH in the same ENVIRONMENT are at the same SOC with a 16mv difference. So even in the best conditions science can supply, you are still taking "best guesses".

If I was just arguing with you, I'd leave it here. But people (except you it seems), search the forum, and I'd hate to leave bad advice uncorrected.

On that note, for anyone who finds this waste of time in the future the generally accepted advice is that Li-Ion and LiPo can use voltage to give you a reasonably accurate actual state of charge, and can be used to compare remaining capacity for similar cells. For LiFePo4, voltage can tell you whether a cell is mostly charged, or mostly discharged, but even with cells in the same batch and environment, you're taking a fairly wild best guess.