Mixing 18650 capacities

[buzzy]

Yep thats what needs to be done if you want a long lasting battery harvested from "laptop batteries".
In my case it took over 10,000 cells to come up with 1120 cells for my 14s80p Frankenstein Battery.
6000+ cells recorded in this spread sheet

Harvested Cell Analysis.xlsx

Database Part Number and Color,Manufacture,Format or SKYRC Program,Wrap Color,Insulator Color,mAh Rating,Nominal V,Max V, Cutoff V,Standard Charging,Standard Dis Charging,Charge Cutout mA,Manufacturer IR (mΩ AC 1kHz) spec,Prefered IR (mΩ AC 1kHz) ,Max IR (mΩ AC 1kHz) APR18650M1A,A123Systems,...

docs.google.com

After learning that IR is a perfect indicator of a cells SOH (State of Health) I was able to discard 4000 + (to the recycler) and giveaway a bunch of marginal cells as they would not (as you say) fit my criteria. I also saved a bunch of time by not testing these out of range /criteria cells.
I am not telling you how to build your battery as you can do whatever you like but you asked the question..............

So if you want to build a successful battery that requires little to no maintenance try to stay within the parameters. If this is not possible at least try to keep you cells SOH to > than 80% and avoid high IR >65mΩ cells. Which if you are pretty much using laptop cells( ICR chemistry) then that IR reading will be the very top margin for good cells as pretty much anything over 65mΩ will have a SOH below 80%. Except the occasional LG**S31865 which can have a reasonable result at 80mΩ The Sony US18650GR G* series cells also can have a very high IR but they do not play well with others so I would stay away from them.
These are my recommendations learned from experience, experiments and many other long evenings researching this.
My Frankenstein battery consists of these cells and their rated capacity as you can see between the highest and lowest there is an 800 mAh difference.
View attachment 26406
This is not tested capacity as by now you surely understand that most used cells will not give you 100% of rated capacity.
That being said if you take a 3000mAh cell at 80% SOH this = 2400mAh, likewise if you have a 2600mAh cell at 80% SOH it is 2080mAh you are within the 400mAh difference so it is not impossible. Does it take a little more work, some more cells, sure. Do you want to build a good battery or something that needs to be serviced on a regular basis. It matters none to me you are welcome to take the advice as you wish.

Best
Wolf

So you spend time opening 10k laptop batteries and then check every single one of them and end up with 1k usable ones? I guess you must value your time to absolute 0 dollars That must have taken months. I even already regret getting the 40 laptop packs that I got. It will take me 2-3 weeks to go through it. Definitely not worth my time, but as I already started, I will finish it. Anyways, thanks for your input. I appreciate it.

 

[OffGridInTheCity]

So you spend time opening 10k laptop batteries and then check every single one of them and end up with 1k usable ones? I guess you must value your time to absolute 0 dollars That must have taken months. I even already regret getting the 40 laptop packs that I got. It will take me 2-3 weeks to go through it. Definitely not worth my time, but as I already started, I will finish it. Anyways, thanks for your input. I appreciate it.

Yes sir - there are trade-offs to consider. You can buy single cells ready to go that are likely very high quality like this ($2.40c/cell - maybe 99% good) - https://www.batteryclearinghouse.co...onic-ncr1865j2-3400mah-18650-high-grade-cells or cells that are likely 'moderate' but would probably make an OK powerwall like this (35c/cell - maybe 70% OK) - https://www.batteryclearinghouse.co...n-lgdas31865-cells-in-large-modem-packs-4-per. Battery acquisition can be *difficult* to navigate. I didn't have the extremes that @Wolf reports. My worst purchase was 30% discard but overall - maybe 10% discard.

Another big issue is the purpose of your battery. If you're building one for an EV where high demand - weight - max performance currents are needed - then that's entirely different than *large* (10, 30, 60kwh powerwalls) with low stress current/cell requirements.

Using the above links you *roughly* might get numbers like this for a 13kwh battery.
The 2.40c/cell, maybe $3000 (cells + other) you might get in the range of $230/kwh for high quality.
The 0.35c/cell, maybe $1500 (cells + other) you might get as low as $115/kwh - 50% more work / moderate quality but OK for low-stress powerwall.

I spent 3 years part-time (maybe 5hrs/week) processing 10,000 cells... (e.g. 20weeks full time) Not sure - just that it was significant work but OK for me as I just did it in the background while in my office thru the day.

HOWEVER - now look at BUYING a solution. Here's just a couple to illustrate - but really, start shopping around....
1) BigBattery - https://bigbattery.com/products/kit-48v-hsky-lifepo4-103ah-5-3kwh/ $2,539/5.3kwh = $479/kwh.
2) EB4 - https://shop.signaturesolar.us/products/48v-100ah-lifepower4-battery-by-eg4?ref=cPwLcVc0SW-BjN $1,499/4.8kwh = $313/kwh.
but with these you need racking/hookup etc.

Just start checking prices of batteries 'ready to go' or even LifePo4 cells - less work than 18650 but still some work - and then you can decide if you have time/interest vs $ etc.

The reason many of us have sent the time is 1) hobby/fun to do and 2) saves really significant $ and we have the time to spare, but 3) I agree it's can take a lot of time

 

[Wolf]

So you spend time opening 10k laptop batteries and then check every single one of them

I did not open 10K of laptop batteries but ≈1500 with an average cell count of 7. Yea about 10K of cells and yes checked the Voltage and IR of every one. As mentioned before once I found a correlation between IR and SOH processing went much faster as I would only test cells that fell within an appropriate IR range for the cell type. I got the laptop packs very reasonable and my overall cell price was ≈ .25¢ for the 1120 that I finally chose. I do value my time but in this case it was more about learning how to salvage cells in the most efficient manner and sharing this with the community to hopefully assist others in their endeavours.
Plus I actually saved a lot of money as it kept me out of the bars .
My first 14s80p took me from cell collection, concept design and completion about a year. In this time I learned a lot and improved my workflow to the point of processing 64 cells every 24 hrs, taking maybe 1hr a day to record results and input the data into my spreadsheets most of which was automated. The addition of 2 Megacell Chargers also helped immensely.
I did purchase Scooter batteries from Battery Clearing House and after initial testing found the yield to be much greater that the laptop packs so I bought a bunch, 230 packs each with 20 cells in them, although at an increased price of ≈.75¢ per cell. Still a bargain for LGM26 cells that tested at close to if not at 100% SOH and the IR was also consistent. At that rate I was able to process my 2nd, packs within a month and started on the 3rd pack while waiting 30 days to tests for SD cells. I have just finished my 4th packs so all in all I have (4) 14s80p batteries with a conservative estimate of 32 kWh capacity.

I didn't have the extremes that @Wolf reports. My worst purchase was 30% discard but overall - maybe 10% discard.

In my case at first I was collecting all kinds of packs many with generic cells that did not pan out. As I gained experience the usable cell count increased considerably but in the final long run I was a bit more stringent with my build parameters hence the difference in the usable cell count. Out of those 6000+ cells I still have a bunch that are just fine being stored for another time and another project.
Also as @OffGridInTheCity says: hobby/fun to do/saves $
Me: /I learned a lot/Arduino programing/Excel/ Python/and much much more..........
Wolf

 

[Wolf]

Depends on how you want to build it. You could have a (let's say 7s20p) battery and add as in figure 1 additional 7sxxp packs.
You are in essence parallelling packs to each pack.


In my case I build 14s80p packs and each one is its own battery. I then parallel the batteries together through a 2 pole circuit breaker to a buss bar.
This allows me to take a battery out of service if it needs maintenance yet keeping the system up and running.
Wolf

 

[fkarabill]

Depends on how you want to build it. You could have a (let's say 7s20p) battery and add as in figure 1 additional 7sxxp packs.
You are in essence parallelling packs to each pack.


In my case I build 14s80p packs and each one is its own battery. I then parallel the batteries together through a 2 pole circuit breaker to a buss bar.
This allows me to take a battery out of service if it needs maintenance yet keeping the system up and running.
Wolf
View attachment 26748

@Wolf if i understand, in figure 1 you need only one BMS and in figure 2 you need 3 separate BMSs, right?

 

[Korishan]

if i understand, in figure 1 you need only one BMS and in figure 2 you need 3 separate BMSs, right?

Yes, correct. Unless you connect balance leads from each of the strings together, which would look like Fig 1, just with thinner wire that isn't designed for carrying load current.