Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
18650 Garden-Powerwall with 3KWh, no 4,6KWh no 10 or better 25, wait 50KWh in Germany

my name is Stefan ( 37, from Germany) and after reading a lot at this wonderful board and watching tons of vids from HB and others at YT since the last few month I'd like to share my building progress with you folks, show my successes and my mistakes too.

The beginning:
in march this year finished my first installation of a photovoltaic at the summer house in my garden

My house is an 115 years old farmers house and the roof isn't very suitable for PV, therefor in the garden

In March and April I was pleased how much energy comes from the sun - and at the same time sad about how much energy will be "lost" during daytimes because of overproduction.

Plus:  I already planed to increase the PV with some more modules on the firewood storage

The point I stumbled over a video on YT where a guy tinkers on old laptop batteries to build a solar powerwall at it's own - amazing  Idea

I decided to try building a 14s40p powerwall with about 3KWh to cover the base load of the house (250W).

My first idea was to buy some new chinese "Supercells" with 9.900mAh or even more for the powerwall. But the "Trustfire" in my DIY bluetooth speakerbox were a flop and my China batterie 18650 test - 508.800mAh in 62 cells showed that it's the same with other "brands".

So I decided to go with used cells from major brands.

I've done some researches on the internet where to get used 18650 batteries. In Germany this is really hard because of the battery recycling laws. In fact it is not allowed to electronicshops / bikeshops and so on to give used batteries away except to a qualified recycling companies. Btw:  actual in Germany, "recycling" of LiIon cells means "burning" them down   Undecided 

So I mailed and phoned round about 100 companies all over the country to get 6 !! positive answers from some ebike shops and some accurepair services. In two weeks I drove 4.500KM to collect them, but at all it was a good deal. All together I got something about 12.000 cells stuck together in different ebike akkupacks (40 - 60 cells in each pack).

Lots of time and work to dissasseble them



for testing I bought a LiitoKala Lii500 Engineer. Then another one. Then a XTar VC8 for pre-charging and 0V recovering (most defective eBike-akkus are 1-time deep-discharged because of an defective BMS, the batteries are still good and within a range of 90% or even more of the original capacity)

because I hate mussy cables next step was a charging station with an ATX power supply and some space left for even more chargers / testers

actual state for testing is: I pre-charge the cells at cheap 10-slot chargers while testing another lot of already fully charged cells in the 12 Liito Kala 

I really like the XTar VC8 8-slot charger (middle, left), but they are not really cheap and handle a maximum of 300mA per slot, so I added 10 super-cheap chinese chargers (top row)

10 independent slots each at 0,6A cost round about 10 - 11 USD incl. shipping

Phomax 10 slot LiIon charger (the same from brand GTF)


quality isn't the best and I had already three defective charging pcbs,
but got three new chargers send out, took the bad pcbs out, replaced them with good ones and now have too functional chargers = 20 slots extra.
Should not be your first-use charging device but for fast bulk charging it's perfect - for me.
Oh, and they can do 0V reactivating as I tested, that's the first picture

To sort the cells by capacity and distribute them on equal packs I decides to use the Excel based RePackR and therefor I prepared an old IBM Thinkpad to use it into my garage / work shed

while processing the cells there were some works outside going on to build a terrace

within this works I did some cable works...

...from the house... the firewood place - which got 8 JA-Solar modules each 340Wp. This will be the system for the DIY powerwall

electric housing with lightning-surgeprotection and an extra 400V connection for heavy equipment such as log splitter

the inverter will me mounted into the cellar of the house in a grid parallel mode. Just had to add an extra 5x 2,5mm² wire

the place for the inverter. It will be an MPP solar / Infinisolar hybrid. Model unclear at this moment

still processing cells. At this time I sorted them by model  into boxes to safe time when adding into the Excel based RepackR

but as my progress goes on, my plans with the total capacity grown too. Actual plan:  build one test-powerwall to learn how to build them best, then build another one and get about 10KWh total.
At this time a decided to not use the Excel because of the huge among of time to fill all the data in. Instead I wanted to sort the cells into steps of 50mAh like HB have done with his huge powerwall (I think he used 100mAh steps)

So I bought 30 or 40 new plastic boxes. This are "Samla" from IKEA. I hate! this shop because I can't find out when I am in... Butr this boxes are perfect for 18650s. ~150pcs fit in the smalles 4L box (as on my first pictures above), ~350pcs in the bigger 11L box


costs me several days to sort the already processed cells into the 50mAh-boxes. I better had done this directly when took the cells out of the charger

short-circuit mistake. Just one of innumerable others...  Confused

after 30 or more different types of eBike akkupacks the best are definetely the ones from Bosch.
40 cells, very quick and easy to dissasseble, just high-capacity cells (LG MG1 with 2.850mAh minimum, mostly LG and Samsung with 3.500mAh), no glue and other shit between the cells, and every single akkupack of 130 I got in total was a defect at the BMS with cells in very good condition. No heaters, no dead cells, no water, no mechanical damage. Perfect for my use.

after the capacity-tests I checked the inner resistance with the well known Vapcell YR1030. But it's annoying to chek them laying on the table, rolling around so I builded a little helper-thing

now I can handle 150 cells 

Timelaps video

I test them one by one and the bad ones (70mOhm or more) are pushed out a little bit, so when finished all cells I can sort them out for trash or other purposes

of course I had some cells which need a new heatshrink

purple is not my favorite colour, but I killed the complete stack of green shrinks within one unfocused moment while pan the heatgun   Dodgy

finally!! it's time to start the test-build

I am using the not-so-good laptop cells with round about 2.300mAh.
Laptop cells? In German internet forum you mostly can read about builds with laptop cells and how easy this is to disasseble, while eBike-akkupacks aren't easy to open and process, so I've given them a try and collected some laptop batteries from recycling companies.



to be continued
daromer, nz_lifer, OffGridInTheCity And 3 others like this post
Nice read!
Stefanseiner and hbpowerwall like this post
NOTE! My links supplied in this message may be affiliated with Ebay and by clicking on them you agree on the terms.
YouTube / Forum system setup / My webpage  Diy Tech & Repairs

Current: 10kW Mpp Hybrid | 4kW PIP4048 | 2x PCM60x | 100kWh LiFePo4 | 20kWh 14s 18650 |  66*260W Poly | ABB S3 and S5 Trip breakers
Upcoming: 14S 18650~30kWh
I like the resistance testing rig Smile
So far I have had 1 major short, quickly rush it outside to let it finish burning.
Many cuts. At least the nickel strips are sharp and heal fast.
Stefanseiner likes this post
Fantastic progress - you're efforts should pay off nicely!
Stefanseiner likes this post
because the powerwall will be inside the house I will use a fire safe cabinet. And as I tested if the 40p packs will fit - they seemd to be too small   Tongue 

so I added another 4x5 cellholder and 20 more cells

14s60p. Now they don't look like some toy batteries anymore   Sleepy

for security I want to use fuse wire and therefor I will soldering the cells

but my old Ersa soldering iron with 40W is too weak, and the new 200W chinese iron seems to be bad quality, after 120 soldering points this is how it looks like. OK, can be grinded.

my first pack is done. Busbars made of 3x 2,5mm² copper wire

SC16-6 ring-terminals at the end for solid connection between the packs.

mini-voltemers includes

while soldering another pack the fuse wire of one cell burned two or three times untill I noticed, that this cell is getting hot and hotter.
What happened? A soldering wire marble roled into the gap between the positive pole and the negative shell. So quickly disconnected the cell and throw it into the "Fire bucket". I have two buckets in my shed for this, one with the boddom covered with sand where I can throw hot cells and cellpacks in, and another bucket full with sand to pour into the first.

Balancing. I did some cabels with alligator clips to balance all of the 14 packs in parallel. In the meantime I can modify the metal cabinet

It's an old metal box too. I sanded and painted it new

then some works for thermal care.

I dont want the box to be 100% airtight, but I want to keep the fire into the box if there will be one, so I don't want to use an open ventilation system.

Using some computer cpu cooler (Intel boxed cooler) at the inside...

...and some 100 x 100 x 18mm passive radiators...

...with thermal paste... the outside

will be tight fitting for the 14s60p system

fermacell at the boddom and 40mm mineral wool WLG032 (fireproof 1.000°C +) at the walls

getting even tighter but fits

7 packs in parallel...

...intermediate floor with fermacell...

...and 7 more packs

all packs are separated with fermacell separators with 12cm in high = 2cm more then the packs to ensure ventilation.
In front some space left for cables and connectors

BMS / active balancer, 100A DC circuit breaker, voltmeter and fan controller (not on this picture, now it's mounted at the left of the bms)
It is an JKB2A24-15P which means it is the version with 2A balancing current, 150A continous charging/discharging current, 300A max.
I like this BMS / active balancer module and you will find more infos at this thread ->  adjustable Cheap Chinese Active Balancers

another 40mm of mineral wool on top

now I am running some tests to evaluate the thermal developement.
The inverter I finally bought is an Infinisolar E5.5k It's the same like the MPP Solar MPI 5.5k

But as far as I can say at this time, the temps inside the box are OK and stay below 30°C at a maximum discharge current of 30 amps = 0,5A per cell. If tests are going well I will raise the limitation to 60A = 1A per cell.
nz_lifer likes this post
Very nice work; good thing you have a large supply of cells to work with.
this first test-powerwall is about 4,3KWh and works grid-parallel


this is the day I "activated" the powerwall for the first time (Green is consumption from the grid, orange is PV-production which is going into the grid). You see the green line at the morning which is the permanent consumption of our house, round about 250 watts.
Over daytime the PV produces more electricity as we need and feeds into the grid. The green peeks between are coffee machine and microwave. After 6pm when the sun goes down there is, now with powerwall, no permanent consumption   Smile   at 8pm the big peek is the oven / stove while cooking

an exemplary summer day with PV and powerwall   Smile 
0,06KWh consumption from the grid and 21,72KWh feeded back into the grid

summary of August, you easily can identify the first day with powerwall

26th of September:  the heatpump working for the first time in this year and the consumption raises

this is a typical day now

the bad days are coming   Sad 

here is detail view of Oktober 1st. You see that the powerwall is zeroing the consumption till 7am, then it's empty and you see the thirsty heatpump working

the first 4,3KWh are finished and working well so I can do some other work

The planned 3KWh are 4,3KWh now and completed, next step is a second powerwall

this will be a 14s60p system too, but with some improvements because of the knowledge I have from building the first ones


the first pw was build of the "bad" cells I had, now this one will be build out of the "nonplusultra" cells

a minimum of 3.450mAh

most of the cells are LG MJ1 (green) and Samsung INR18650-35A (pink)

trying to sort by color - and of course by capacity. Equal spreading in steps of 50mAh

btw:  this is my favourite step witin the whole building process:  taking long-tested cells with both hands out of full boxes and stick them into the holders   Sleepy

the tower of power

the next steps are as from the last one:  making busbars outof 2,5mm² wire


twist 3 of them


in total one busbar has 15mm²

for a better contact and to be sure, the ringterminals won't move I'm ironing before crimping

when the flux is still warm you can easily crimp and the flux will a little bit self-adjust



before adding a new pack to the balancing-row I use a cable with a 10A fuse to avoid that the little fusewired will be blown

because the cheap chinese 200w soldering iron is annoying me I got a new one, an Ersa 150S with 150W. Costs round about 75USD

the soldering tip is lots thinner and - it's a permanent one

If I knew earlier how much time and nerves this will save...   Rolleyes  and solder wire too!

this time without extra voltmeters. Inside the firesafe cabinet I can't see the displays anyway

ready to build them togehter

I bought another metal box, identically to the first (just a little bit wmore width), sanded and painted it with some anti-rust paint

improvement:  I want to integrate a voltmeter with battery-capacity-display and a temperature-display into the case top for a quick look while walking by so I will not have to start the app, connect, check every time just to see the main-condition

I'm loving the little cutting discs I have left from my PC-modding era, these are from an dental lab and a lot better then the standard cutting discs you will find for this multi tools

sits and fits



then the fire safe stuff




forgit the holes for the cables


as from the first one I am using 10mm Fermacell plates for the bottom, as separatores between the packs and for the intermediate floor

some SC16-6 ring terminals ans massice screws for connection

before adding the second row I have to connect the BMS because later I will not have enough space to reach the bottom row

for easy and fast changings / mantainance I will not solder / screw the BMS cables, I am using alligator clips


the bottom row has 2cm space under this intermediate floor for air circulation



mainly the same BMS as at the first 4,3KWh powerwall. The other is a JK-BD6A24S-10P (0,6A balancing current) which I changed later on to the 2A version. And it's the older, edged version without the big barcode on the top.
Now I am using directly the 2A version (JK-B2A24S-15P)

it will be mounted on a piece of fermacell plate

because there will be some mineral wool directly on the top of the BMS I glued some nuts as spacers below the BMS for better airflow and cooling

adding a radial cooler at the side (where a few air holes are situated) and choose the high of the cooler so that the airflow will go half to the holes and half below the BMS to cool the case

adding two 18650s and a dip-switch to start the BMS. This is a bad thing of this BMS that you have to put a short impulse of 5V between the main B- and P- connectors, and not having a simple switch or so to start this thing up

I need 12V for the temperature sensor display in the case top, and for the aircooler. In the first build I took the voltage directly from the akkupacks (main negative and three packs away the positive) and this caused in a permanent balancing of this tree packs.
Now I am using a little buck converter


adding the main wires




because with a max. current fo 1A per cell the whole powerwall isn't getting warm inside the box although it is isolated, with this build I didn't add some extra coolers to the box as I did with the first one
Hottest temperature I got (with the coolers not spinning) was below 30°C with an temperature at the outside of 22°C

I added some amored tape around the top wool so I can better lift it up for checking without the need of wearing gloves (or wearing not and getting prickled by little glas needles)


ready for the sun   Cool

I forgot to mention, this time I was able to check the real capacity of the packs with an electronic load

though I tested every single ceel with the LiitoKala but the test is from 4,2V to 2,8V andwith this results I have to estimate the usable capacity.
Now I can test the real capacity between 4,0V and 3,3V - which is the range I will use for the powerwall

It's a total of 6,8KWh
ajw22 likes this post
now that the Infinisolar E5.5k is up with ~11KWh I can turn to the second photovoltaic on the garden shed and change the Fronius Symo inverter (no battery support) to an MPP Solar MPI 10k.

another change has been done into the main circuit breaker box. To get two inverters running in netparallel mode at the same time I have to use two SDM630MCT and set the inductive sensors to a false value so the two inverters only see a part of the actual consumption. The smaller 5.5k inverter will get 40% and the 10K the other 60%

because I don't want new modbus cables through the house and through the complete garden to the MPI 10k, I tried some modbus-to-ethernet adapters to chare the signal over the existing CAT7 installation. Costs me a bunch of nerves until this works...

for testing this is the actual setup. The 6,8KWh powerwall and an addition 3,3KWh I build for my mother's house I've done as a side-project in the last week (14s40p).

in the meantime a startet another side-project

because I want to test if my fireproof cabinet is rock-solid firesafe I plan to build another cabinet and provoke a fire

therefore I am using 840 heater cells to imitate my real 14s60p system

most of them I will only place into cellholders but without connecting them electrically, but to provoke overload and overtemperature I build two 60p packs


the packs won't have to handle high current so I will use just 1x 1,5mm² for busbars

and alligator clips so I can connect them easily

looks strange...

the next steps are easy. Drill holes, some zip ties,


and fusewire

why fuses? Because for me it's the easiest way to solder cells and busbars. And:  this are lots of part-defective heater cells, I just want some fuses on it...

because I plan to do this experiment in spring when temperatures are higher, I am sure the pack will self-discharge. I add some voltmeters so I can see when I need to recharge them a little bit


others try to avoid singel heater cells - I build whole packs with them. Why not?

balancing till spring. In the meantime I am collecting some more heater cells. At "my" German DIY powerwall forum I started a "Search heater cells" thread and people help me get 840 bad cells for this.
Now, with the two 14s60p powerwalls I already finished with a total of 11KWh I am hooked - and want some more power   Angel 
Fortunately I have some more already-tested cells left

But no more tiny steps, I want to do it the right way. I will switch to 120p packs and... two systems with 14s each. Plus 1 pack for spare = 29x 120p packs to builld

trying to sort a little bit by color. The green ones are LG MJ1...


the pink Samsung 35E


needs a lot of space to build this up

here is a timelaps of the building process. In realtime this is about 5 hours






tower of power no. 1


cell capacity reaches from 2.750mAh to 3.500mAh, average is about 3.000mAh

tower of power no. 2

same values as the green

I am still using fuse-wire and soldering iron. Trying to calculate how many soldering points I have to do...
- pre-soldering both sides of each cell = 2
- fuse-wire on each side = 2
- fusewire to the busbar = 2 per 4 cells = 0,5 per cell
total of 4,5 soldering points per cell

29 x 120p packs = 3.480 cells
3.480 x 4,5 = 15.660 soldering points

OK... before that I worried about what to do with my freetime while corona lockdown - now I have an idea   Confused 

pre-soldering is very easy and quick with the Ersa 150s



now I have "professional" cable dispenser  Blush

for the new pack layout my bending template will not work. It fits for 60p and 40p

but there is some free space on the back side



what it will looks like

this time I will use 4mm² wire

85cm of cable


I twist 4 pieces = 16mm² or because busbar is twice that's 32mm² in total



a little hard to bend but it works well

unfortunately the 35mm² rinterminals I bought didn't fit the two 16mm² busbar endings. So I decide to take two 16mm² ring terminals

soldering the busbars did not work well with these 4mm² cables

so just crimping

one heatshrink for both endings






timelaps of the fuse-wire soldering



I think these big packs will need some time until they are fully balanced

I give them 4,05V from the DPS8005

yaih, the sun comes out so I pause for some non-battery-work

btw. my first project with akkus and the reason why I started with this whole thing - DIY speaker with bt. Loud enough for some wood works




very rebellious fir wood. It's everywhere in the garden. A lot too much and I have to cut some down time by time

after two weeks of ticker and try and error and grumbling I finally got the two inverters working together - via modbus-to-ethernet adapter and with the two SDM630MCT energy meter

now I am soldering the packs one by one, hook them together for balancing...

...test the real capacity from 4,05V down to 3,3V with the electronic load...

they all have about 950Wh. This is a total of 13,3KWh per system, together 26,6KWh. Added to the already existing 11,1KWh this will be a total of 37,7KWh   Smile

this one was falsely-charged up to 4,10V

after discharge-test I charge them again to 4,05V with the DPS5020

that's the actual status. The packs are just all done and now I have to work at the old locker I will use as a cabinet.

  • weld some more shelfs into the locker
  • drill some holes for cable outlets
  • sanding and painting it
  • install some firesafe isolation
  • prepare the inverter-places for the bigger powerwalls (thicker wire, new DC breaker)
  • getting the locker down the muddy way to the garden shed without dying...
  • installing the packs
ajw22 and OffGridInTheCity like this post

Forum Jump:

Users browsing this thread: PAF, 2 Guest(s)