18650 battery pack design for a powerbank

Hello :smiley:
I would like to make an external battery (powerbank) in wood with 18650 PANASONIC NCR18650B 3400 mAh batteries. The goal is to have a high-performance and reliable battery over time. I would like to use 6 batteries to build this Power Bank, however I have several questions about its realization.

Specification: I need the battery pack to deliver a voltage of 5V under 2-5A (I’ll probably need a buck converter to raise or lower the voltage).

My questions:

What is the best configuration in order to have optimal performance for autonomy, 1S6P, 2S3P?
What is the more efficient method, step-up or step-down the voltage?
Which reliable BMS do you recommend me for this project?
Thank you in advance for your future answers and wish you a good day!

You can find USB power bank modules on aliexpress, ebay and co, generally for a 1S battery, with protection, USB charging and 5V2A USB output, which in your case wont be enough…

For 5V out I’d go 1S, even if the efficiency of a boost converter might be slightly less than a buck converter, 1S means no balance needed, and cheap “balance” BMSs are absolute trash at balancing…, that simplifies the choice of BMS and any 1S ~10A BMS will do.
Assembling is also simplified, can use a parallel cell holder.

Thanks for your answer :smiley:
I want to build the powerbank without use finished modules and buy each individuals pcb to build it, i find it’s more modular and easy to fit to the case i will build.
Do you have suggestion for the BMS? Because lot of bms on aliexpress have a over discharge protection cutting lower than 2.9V this looks like bad for battery life.
Does is exist 2S BMS high quality board?

Alright is a good idea, I can use theses Holders:

you can also buy power banks without cells that directly work with 6 18650s already

wle

Yeah, sure,
But I’d like to buy the protection system and charge separately, so that I can understand the mechanism and do other similar projects.
Do you have any recommendations?

i have bought several cheap BMS boards for various configurations, 4s to 15s, most use the DW03 chip for doing the BMS functions and control the discharge and charge FETs. i reverse engineered all of these to compare the circuits and would recommend that you do the same with whichever you choose. Then you can modify it as needed, or re-do it, to suit your needs. But i wouldn’t trust any of them with my life.

After doing all that, i built an 8s nominal 30V power bank with 18650s where the BMS is ME. i always measure the voltage when in use, and control the charging using a current regulated power supply.

i looked at several boards used in various electric cars. The best design i found is the one used in the Tesla S packs. They have a unique feature that nobody else has, or likely even understands why it is important in critical applications.

Thanks for the tips, I will try to do the same!
You entire made yourself the BMS? It seems to be complicated to design it.
Do you have a suggestion for 1S and 2S BMS board according to my project and the NCR18650B?
Alright Tesla S is high end pack quality. What is the unique feature you talking about ?

Efficiency of modern boost converters is absolutely top notch. Concerning the balance thing I have to agree, and it is because balancing is done at a rather high and harmful voltage, and also because the actual balance voltage isn't very accurately set.

The majority of BMS boards cut at ≈2.5V (the 1S DW01A based, for example), or maybe a little bit higher.

Why is cutting lower than 2.9V bad for battery life? Can you find me a reliable, non-biased and trustful source which somehow can support such a claim?

In my experience I cannot find any evidence telling me that discharging below 3V is bad for li-ion. I've heard that below 2V things can start going wonky, and even considering this I once dismantled a laptop pack in which I found two Samsung ICR18650-26C cells “sleeping” below 2V (1.6ish - 1.7ish volts as far as I can remember) and the rest barely or a little above 2V. I restored all the cells, and they still delivered above 85% rated capacity on average. I set their voltages close to 3.9V, and let them rest a few weeks to observe their self-discharge ratios. Once the time passed I observed a couple cells which had lost a few centivolts, and decided to discard them.

You can continue believing such bullsheesh if it pleases you to do so. In any case, the actual discharge time which cells attached to a BMS spend at or below 3V is pretty insignificant.

On the contrary, letting cells dwell at high voltage is proven to be detrimental for their life. Or maybe I should say that charging up to a lower voltage enhances lifespan.

Here's a good looking DIY board solution, Injoinic IP5328P based:

With two boards you will be able to deliver 5A. According to datasheet a single one goes up to 3.1A. Two boards also allow battery charging and discharging at the same time.

If the above doesn't sounds enticing, the solution is BMS plus boost converter to discharge, and some specific buck converter (or power bank board LoL) to charge.

As BMS boards I would choose a couple of the above running in parallel (one would suffice, but I like extra reliability, less heat and a tiny little bit extra efficiency). For sale, for example, here and there.

Check the boost converter board at ZQC Module and ModuleLive stores.

Hello,
Thanks you so much for your help on my project. :blush:
It’ a good remark, indeed i don’t have scientific source to prove that cutting below 2.9V is not good for the cells, but i read in severals forums and thread who recommand this to preserve cell life.
Is it true? It’s a good question, but I prefer to have a battery pack that is secure and durable over time. What do you advise me?
I found this graph from US Army RDECOM

2 years ago i built a 4S2P battery pack with cutting voltage 3V an actually is running perfectly fine.

Thanks for the DIY board solution is and good idea, but in my project i would like to have seperate board.

You mean something like this ?

About charging, the boost converter board is secure ? Does-it stop automatically to charge when the 4.2V of battery is reach?

Of course you are free to believe as you choose to. I advise you to choose wisely, though.

See BU-808: How to Prolong Lithium-based Batteries @ Battery University

I can understand your concerns with low cut-off lifespan wise but, in my experience, I believe that the actual impact of a low cut-off is negligible. High voltage stress, on the contrary, is perceptible.

If you look at the right side of the graph (most cycles), from most to least harmful: 50% to 100%, 0% to 100%, 25% to 75% and 0% to 50%.

3V cut-off in a 4S pack? That's a too low cut-off LoL. Unless you mean ;-) 3V per stage, of course.

Of course not! The boost converter is meant to provide the output, nothing else. That is presuming you are going 1S6P, that's why I chose such boost converter which can work with a rather low voltage (down to 2V), so its efficiency at close to 3V and up should be nice.

If you go 1S6P, to charge the battery pack TP5000/TP5100 do nicely and support a wide input voltage range (up to 18V for the TP5100 I think, go check their datasheets). TP4056 boards will do also, but they are linear so input voltage should remain at 5V or barely above.

Sorry if I missed your reasoning but for something as straightforward and well catered for as a USB power bank, why not just buy one?

It’ll be safer, easier and probably cheaper?

I will also say that, unless cell holders are a must for some required reason, they should be avoided. The impact of contact resistances in a holder can't be dismissed because cells have relatively low internal resistances in comparison. Thus, losses in the contacts account for a perceptible reduction in energy conversion efficiency.

Hello thanks for for the article, it's very interesting.
About your experience, what is the best over-discharge and over-charge voltage cut-off to increase the battery life ?

According to this table from batteryuniversity.com , if we cut at 4.10v instead of 4.2v we already gain lot of cycle from 300–500 to 600–1,000 it is not negligible.

Of course per stage haha sorry my mistake

I have question about TP5000/TP5100, there is not problem for charging 1S6P battery pack with cells in parrallel ? If i understand well If charging at 4.2V 2A each cell will receive 0.333 A right? Total charging time = 3400/330 = 10.3 hours ?

Yes of course , but I find that you learn a lot of things to do it yourself and it allows you to have something flexible and above all sustainable and from an ecological point of view, if something is not right you can fix it without having to throw everything away.

Oh wow, you learn me something
Thanks for this remark.
If i use battery holder, What will be the impact of the resistance in term of loss in %? It's better to soldering ?

I think you interpreted it backwards

Cells in parallel equal to a single combined cell with the capacity sum and increased power delivery. No problem using a single TP5000/TP5100 to charge them. You can also use multiple modules in parallel to increase charging speed. Those modules come stock configured for 1A, and I would just buy additional modules to increase charging current instead of messing with the sense resistor (TP5000/TP5100 modules get hot at 2A anyways).

TP4056 modules are even cheaper, although you will need a DC/DC converter stage if higher than 5V input is required or if you are interested in reducing the maximum charge voltage to the cells. I think that TP4056 modules should still work with lower input voltage, let's say 3.95 - 4V, while still cutting off charging but can't say for sure (I didn't test this myself yet).

According to feedback info on some DIY power bank efficiency (≈80%) in AliExpress, with cheap metal sheets and springs as battery holder, the reduction in efficiency is probably ≈10% (I built one same model with soldered cells, estimating no less than 91% efficiency).

Soldering cells is another story and involves using aggressive steel flux, plus I recommend and use low temperature solder (Bi50Pb32Sn18, a Rose's Metal variant).

Oh ok, I see, I really want to understand something, in theory, if we have 6 cells Panasonic NCR18650B 3400mah in parallel in the same battery pack.
The total battery pack capacity will be 20400mah, charging it with one TP5100 at 2A, each cell will receive 0.333A (2/6) right?
Total charging time = 20400mah/2000mah = 10.2 hours? Is my reasoning right?

About the optimization of the heating, you advised me to use the charging module in parallel:
It should be connected like this?

Earlier we are talking about the tips to preserve the battery cycle and life.
And above all, stopping the charge before 4.2V would be beneficial based on the studies.
After several hours of searching, I found I very interesting module, the MAX745.

Thanks to this module we can adjust the current charging:

- Voltage adjustment range: 4.1-4.4V

  • Charge current adjustment range: 0-3.6A
    What do you think about?

Another question, can we eventualy custom the TP5000/TP5100 to reduce the charging voltage?

Thank you for your future answers :smiley:

I use a 4s or 5S pack to power some marine USB ports made for panel mounting in a dashboard.
These things run off a 12v-24v source.
Supposedly they are QC 3.0 compatible.
I currently have no way to test them that high.
I need to get a better USB monitor and a Load Tester.

Don’t cost a heck of a lot.
Easy to make.
To charge, I just pop the cells out and replace them with fresh ones.
All the Best,
Jeff

Sort of. NCR18650B cells in practice don't reach 3400mAh, though. It's datasheet specifies 3350mAh typical and 3250mAh minimum, which hardly does. Check HKJ's old review here, and its review along the LG F1L at Thunderheart Reviews. At a relaxed 0.2C discharge rate (5 hour discharge) it's more like 3300+mAh if discharged down to 2.5V. The actual capacity (and energy) you will get mainly depends on your average discharge rate.

Dividing the total capacity by the charging rate won't give you a correct charge time. Charging li-ion is done with a CC/CV profile; this means that once maximum battery voltage is reached, current flow to the battery starts tapering down. This current tapering lenghtens the CV phase duration, as the average current flow gets reduced. The CV phase ends once the charging controller cuts-off, usually at a fraction of its rated current. As general rules, the slower the charging speed and the lesser the battery and contacts resistances, the shorter the CV phase becomes.

Optimization of heating LoL!

Yes, that is right. Note: B+ and P+ in the PCM are common.

Interesting MAX745 module:

Output voltage adjustable, interesting. I am also sure that fiddling with the onboard 1003 chip resistor connected to the output voltage trimpot will could result in a extended voltage range adjustment.

It's far from cheap, though (found it slightly cheaper here, but still quite a bit of money).

Concerning output voltage adjustment in TP5000/TP5100, it could be but no idea.

I am right now testing my TP4056 charging module theory:

With my power supply feeding 4.017V into a TP4056 module, I measured the module output at 3.942V.

The TP4056 is feeding current into my cell holder with an VTC5A clamped in. Output from the holder goes into an old homemade shunt resistor made with 5x R050 resistors in parallel and alligator clips to hold my multimeter probes. The multimeter shows 10mV in the above screen (calibration: 10.65mV for 1A of current measured with the multimeter in current range).

The EBD-M05 is there just monitoring cell voltage.

Thanks to this setup I will know if the TP4056 terminates when fed with a lower input voltage. If it does, this will be very neat. :-)

Sun, 06/28/2020 - 04:24

Once input voltage is lower than 4.25V the blue led shuts off, and while the module still limits current to charge, it doesn't terminates (no current cut-off at 1/10 of the set current).

Hello,
Thanks for this precision about the NCR18650B cell total capacity, this is good to know.

Although the many factors to consider, is there a mathematical formula to estimate the charging time of a battery pack?
I found this little tool, what do you think?

I have a question about the TP5000/TP5100.
When the board detect the battery reach 4.2V, Does it continue to send a voltage of 4.2v 0A or everything is cut off?
Another question, I really want to understand the system, what is difference in operation between TP5000/TP5100 module and an CC/CV module like this one?
Does the CC/CV module automatically stopping delivering current when the set CV voltage will be reached ?

Alright, according to the tp4056 datasheet the minimum Input Supply Voltage is 4V, according to your test, does it work propely ? What about the temperature of the board?
If the tp4056 is feeded with 4.017 it stop the charge at 3.942V ? Or you need to control it manually ?