Newbie 18650 powerwall build help please :)

[Solarmax]

Hi All

Firstly thank you to all the information posted on this forum I've been reading, researching and learning a lot from what is contained in the forums nice to have these communities that help each other out.

So to give a bit of background, I'm in the process of installing a diy off grid solar project with roughly 3kw of panels on my summer house roof and 4kw of panels on my garage roof, i have an 8kw hybrid inverter that has two inputs of 4 kw so one array on each is my plan.
i am also planning to build an 18650 powerwall from recycled batteries ( mainly salvaged from dyson vacuum cleaner batteries) I have been ripping these out and testing them over the last 6 months or so and I have just over 1200 all around 1750 - 2000mAh ive filtered out low capacity batteries and self discharging ones. i also have a 250amp daly bms and planning to build a 14s100p pack for around 10kwh of storage.
i have a few questions in relation to the putting together of the pack that i would like some advice on my maths with cable sizes and thing too please

1) Mixing cells : i have found conflicting information on this, my cells are mainly sony and LG but i am around 200 batteries short so was planning on buying those new (3600 mah) is this much of a difference in single cell capacities and potentiall differnt chemistries an issue or is sorting them using repacker ok? i see many people puting them together from various laptops/ powertools etc.

2) my inverter can put out a max of 8000w of power ( i dont think ill ever draw this much) , im building a 48v system so 8000/ 48 = 166.66 so the max amp draw is 166 and a bit amps this divided by 14 packs is 11.9 and that divided by 100 is 0.119 so does that mean even a max draw its only pulling 11.9 amps per pack and only 0.119 amps per cell? or have i got my maths wrong?

3) Wire sizing: i was planning on planning on individually fusing each cell with 5 amp fuse wire, but if my above maths is correct is 3 amp or even 1 amp if there is such a thing not better? also i have some 4mm2 (12 awg i think) i was planning on using for the busbars is this ok or does it need to be thicker?

4) pack testing : i have litokala lii 500 testers for individual cell testers, but i don't really understand how to test ( charge and discharge ) for capacity and heat anomalies etc the 100p packs or potentially the whole thing put together even. please can i ask for some advice as to what i need and how to do this so i can get things ordered

i think that covers everything

Many thanks in advance for your help.

 

[Korishan]

1) Mixing cells : i have found conflicting information on this, my cells are mainly sony and LG but i am around 200 batteries short so was planning on buying those new (3600 mah) is this much of a difference in single cell capacities and potentiall differnt chemistries an issue or is sorting them using repacker ok? i see many people puting them together from various laptops/ powertools etc.

Mixing is really a bad idea with new vs old, and low vs high discharge rates (aka, mixing power tool cells and laptop and similar cells). The reason is that with new/old is that the new ones will discharge much higher current than the others and will shorten their life. Altho this might not be down to the old ones life. However adding only a few new ones may be fine in the long run as long as current load is kept low, don't max out amp draw capacity.
As far as low/high discharge, same reasoning as the prior with the powertool cells trying to output more current than the others. However, same as previous, keeping the amp draw low will help keep this from happening

2) my inverter can put out a max of 8000w of power ( i dont think ill ever draw this much) , im building a 48v system so 8000/ 48 = 166.66 so the max amp draw is 166 and a bit amps this divided by 14 packs is 11.9 and that divided by 100 is 0.119 so does that mean even a max draw its only pulling 11.9 amps per pack and only 0.119 amps per cell? or have i got my maths wrong?

Yeah, math is a little wonky here. It's 8000w / 48V = 167A / (Number of cells in Parallel). So if you have a 20p pack, then 167 / 100 = 1.67A/cell
This is still a little high. However, 8000W is probably it's "peak" power draw, which means is this unit rated as a 4000W (or 6000W) with 8000W surge/peak? If so, this is far less likely to be an issue. 4000W / 48A = 84A / 100P = 0.84A/cell, far better results. This would be more in line what you would use on a consistent basis. But also note that if you did surge to 8kW, the cells would spike in current load and will possibly cause them to overheat, especially recycled/reclaimed laptop cells that are only designed for about 1A max discharge.

3) Wire sizing: i was planning on planning on individually fusing each cell with 5 amp fuse wire, but if my above maths is correct is 3 amp or even 1 amp if there is such a thing not better? also i have some 4mm2 (12 awg i think) i was planning on using for the busbars is this ok or does it need to be thicker?

Cell level fusing is not a protection of overcurrent load to devices. It is a protection against cells vs cells. If a cell goes bad, and goes dead short internally, it will pull a huge amount of current. This load will cause the fuse wire to pop and disconnect that cell from the pack, protecting the other cells from possibly getting roasted from a thermal runaway condition.
If you put too small of a fuse on the cells, and your devices overload that fuse and cause it to pop, you'll get a cascade effect that will pop all your fuses. 10 fuses @ 10A load is 1A / cell. One of those fuses pops and now 9 fuses have to handle 1.1A. Then another pops with 8 fuses left and each will carry 1.25A. See the cascade effect here?

4) pack testing : i have litokala lii 500 testers for individual cell testers, but i don't really understand how to test ( charge and discharge ) for capacity and heat anomalies etc the 100p packs or potentially the whole thing put together even.

Test each cell individually for capacity and monitor for heat output. A single cell should not go over about 120F (48C) under charging conditions. Even this is approaching danger territory. You will usually be able to detect this once a cell gets to about 4.1V the internal resistance starts to go up and it starts converting the energy input into heat.
These are not suitable for your powerwall/pack. They could be possibly used for single celled devices like remote sensors and such that use very little current (like esp32/8266 devices, weather stations, etc)

please can i ask for some advice as to what i need and how to do this so i can get things ordered

Please check out the FAQ section on the main page. A lot of your questions here could be answered there as well, and then some.
One recommendation for cell testing I'd go with is getting a 4-wire Kelvin tester. This will help narrow down what cells are good/bad based on internal resistance before you start charging/discharge testing, saving time/energy. Check out @Wolf 's posts on these as he has load of data on several different brands/types. Heck, just check out most of his topics and he's covered quite a bit

i think that covers everything

No, there'll be more

 

[italianuser]

Hi All

Firstly thank you to all the information posted on this forum I've been reading, researching and learning a lot from what is contained in the forums nice to have these communities that help each other out.

So to give a bit of background, I'm in the process of installing a diy off grid solar project (snip)

sounds good!

1) Mixing cells : i have found conflicting information on this, my cells are mainly sony and LG but i am around 200 batteries short so was planning on buying those new (3600 mah) is this much of a difference in single cell capacities and potentiall differnt chemistries an issue or is sorting them using repacker ok? i see many people puting them together from various laptops/ powertools etc.

Yes, how @Korishan says is correct. In your place I'd try to get cells as similar as possible to the 2000mAh ones you have.
What's important is try to avoid mixing different chemistries. To make it simple: cells should be all high drain (power cells) or all normal drain (energy cells); to be more correct there's quite a few different chemistries, the most common "energy cells" are ICR chemistry.

When you mix different chemistries the main problem is the big difference in the charge/discharge curves which will obviously change the overall voltage of a parallel pack (and BMS perception of the pack's voltage).

A good way when you have different cells (with same chemistry) is to pick them with "round-robin" approach when you make each pack; so all packs (all series) will be more-or-less similar as for what cell distribution will be, for e.g. for a 20P pack:
- 10 NCR xyz; 5 Panasonic abc; 3 Samsung ddd; 2 Sanyo blablabla => output should also have be a similar total amperage in all packs.

For powerwalls the more expensive high drain cells are not really needed because load is distributed between each cell composing a parallel pack.

2) my inverter can put out a max of 8000w of power ( i dont think ill ever draw this much) , im building a 48v system so 8000/ 48 = 166.66 so the max amp draw is 166 and a bit amps this divided by 14 packs is 11.9 and that divided by 100 is 0.119 so does that mean even a max draw its only pulling 11.9 amps per pack and only 0.119 amps per cell? or have i got my maths wrong?

Check the manual because there's quite a few different parameters around "amperage/watts": input from panels, charging to battery, output from battery, ecc.

3) Wire sizing: i was planning on planning on individually fusing each cell with 5 amp fuse wire, but if my above maths is correct is 3 amp or even 1 amp if there is such a thing not better? also i have some 4mm2 (12 awg i think) i was planning on using for the busbars is this ok or does it need to be thicker?

A busbar should be able to carry the fully amperage of the battery, so a 4mm2 isn't enough. The 4A/mm2 should be a quite realistic rule of thumb for your wiring calculation.

4) pack testing : i have litokala lii 500 testers for individual cell testers, but i don't really understand how to test ( charge and discharge ) for capacity and heat anomalies etc the 100p packs or potentially the whole thing put together even. please can i ask for some advice as to what i need and how to do this so i can get things ordered

On Liitokala 500 use the NOR test which will do a full C-D-C cycle: it will Charge the cell; Discharge the cell and give you the Amp capacity; then Charge the cell again; normally total time will be around 4-5 hours per cell.
Measurement should be accurate enough, for many cells it's surely more accurate than discharging at 1000mAh (Liitokala discharges at 500mAh) because datasheets for many ICR cells report a "standard" charging current of 400-500mAh.

Insert the cell and press MODE twice to go into NOR mode.

Keep us updated!

 

[Solarmax]

Hi Korishan

Thank you that helps clear up a few things!

as you say there will always be more ....

Mixing is really a bad idea with new vs old, and low vs high discharge rates (aka, mixing power tool cells and laptop and similar cells). The reason is that with new/old is that the new ones will discharge much higher current than the others and will shorten their life. Altho this might not be down to the old ones life. However adding only a few new ones may be fine in the long run as long as current load is kept low, don't max out amp draw capacity.
As far as low/high discharge, same reasoning as the prior with the powertool cells trying to output more current than the others. However, same as previous, keeping the amp draw low will help keep this from happening

so am i rgiht in thinking that if i stick to all the same battery type (from vacuum cleaners in my case) i would be bettery trying to source more from there ? i guess these are rated for a higher amperage draw than laptop batteries, i know on full power the dyson battery ( 6 cells) will drain flat in 6 minutes!

yeah, math is a little wonky here. It's 8000w / 48V = 167A / (Number of cells in Parallel). So if you have a 20p pack, then 167 / 100 = 1.67A/cell
This is still a little high. However, 8000W is probably it's "peak" power draw, which means is this unit rated as a 4000W (or 6000W) with 8000W surge/peak? If so, this is far less likely to be an issue. 4000W / 48A = 84A / 100P = 0.84A/cell, far better results. This would be more in line what you would use on a consistent basis. But also note that if you did surge to 8kW, the cells would spike in current load and will possibly cause them to overheat, especially recycled/reclaimed laptop cells that are only designed for about 1A max discharge.

i thought it might be! thank you fro the clarification, according to the inverter manual its rated for 8000w cantiouns peak of 16000w for 5 seconds as best as i can tell, so whilst i may not get close to this (i sized up for input more than outout as the sun can be elusive here in the uk!) maybe id be even safer at 120p packs 167/120 1.39 amps per cell or 200p packs at 167/200 = 0.84 amps per cell? or maybe that size of pack is a bit of a monster?! like you have said a max 4000w draw id more realistic, also the inverter has an ac input from the grid and a bypass mode, maybe i can set this to use this if the draw is ever over 4000w for example to prevent this?

Cell level fusing is not a protection of overcurrent load to devices. It is a protection against cells vs cells. If a cell goes bad, and goes dead short internally, it will pull a huge amount of current. This load will cause the fuse wire to pop and disconnect that cell from the pack, protecting the other cells from possibly getting roasted from a thermal runaway condition.
If you put too small of a fuse on the cells, and your devices overload that fuse and cause it to pop, you'll get a cascade effect that will pop all your fuses. 10 fuses @ 10A load is 1A / cell. One of those fuses pops and now 9 fuses have to handle 1.1A. Then another pops with 8 fuses left and each will carry 1.25A. See the cascade effect here?

this makes total sense now thank you,

Test each cell individually for capacity and monitor for heat output. A single cell should not go over about 120F (48C) under charging conditions. Even this is approaching danger territory. You will usually be able to detect this once a cell gets to about 4.1V the internal resistance starts to go up and it starts converting the energy input into heat.
These are not suitable for your powerwall/pack. They could be possibly used for single celled devices like remote sensors and such that use very little current (like esp32/8266 devices, weather stations, etc)

thank you again for the info , i have a read much of @Wolf 's post and did in fact purchase an internal resistance meter based on his post. i was thinking more about how, once i have build the pack (100p for example) do i test that ? and in fact maybe even the whole Powerwall when its together? i dont understand where all that power goes :/

hope that makes sense, thanks again ill keep reading and learning also

 

[Solarmax]

Hi Italian user , thank you for this info

Check the manual because there's quite a few different parameters around "amperage/watts": input from panels, charging to battery, output from battery, ecc.

https://www.photonicuniverse.com/upload/file/Manuals/Iconica/IC-MX/IC-MX8000-48_user_manual.pdf

link for the manual for those that understand it better than i do at the moment! any advice greatly appreciated

A busbar should be able to carry the fully amperage of the battery, so a 4mm2 isn't enough. The 4A/mm2 should be a quite realistic rule of thumb for your wiring calculation.

thank you i didnt know that rule so always learning! for the main connection to the inverter i see hoe that maths works, but for the individual pack that ill solder/spot weld the cells too this would be split between the 14 packs? so 167 amp/ 4 = 41.75 so this for the main cable but this divided by 14 = 3 so my 4mm2 per pack (both positive and negative side is ok?) or have i got dodgy maths again here?!

On Liitokala 500 use the NOR test which will do a full C-D-C cycle: it will Charge the cell; Discharge the cell and give you the Amp capacity; then Charge the cell again; normally total time will be around 4-5 hours per cell.
Measurement should be accurate enough, for many cells it's surely more accurate than discharging at 1000mAh (Liitokala discharges at 500mAh) because datasheets for many ICR cells report a "standard" charging current of 400-500mAh.

Insert the cell and press MODE twice to go into NOR mode.

apologies for my poor explanation of what i was trying to ask, :/ i have tested around 1200 batteries with these, but when i test the (100p for example) packs or even the whole powerwall when finished how do i do that and what equipment is needed?

hope that makes sense! many thanks again.

 

[Korishan]

so am i rgiht in thinking that if i stick to all the same battery type (from vacuum cleaners in my case) i would be bettery trying to source more from there ? i guess these are rated for a higher amperage draw than laptop batteries,

Yeah, if you can source them all from the same devices, sure. But just remember that normally powertool cells (including vac's) are usually abused pretty heavily. Ie, they are discharged to almost flat (turning off the device) and then fully charged, repeated several times a day at times, or every day. Whereas a laptop, or similar devices, are usually kept close to full, and normally put on the charger long before they are "dead". These usually have a longer life cycle count left.
Just something to take into consideration. Powertool cells are great for learning, at least. Don't need as many to accomplish the same tests. But they just won't last as long lifecycle wise

i thought it might be! thank you fro the clarification, according to the inverter manual its rated for 8000w cantiouns peak of 16000w for 5 seconds as best as i can tell, so whilst i may not get close to this (i sized up for input more than outout as the sun can be elusive here in the uk!) maybe id be even safer at 120p packs 167/120 1.39 amps per cell or 200p packs at 167/200 = 0.84 amps per cell? or maybe that size of pack is a bit of a monster?! like you have said a max 4000w draw id more realistic

Ooooo, chonky beast! Yeah, you'd want to go bigger, for sure, on the battery. What I would recommend is instead of going with a single large 120p, or larger, pack, go with 2 strings of 14s100p. Then connect each of the packs with a balance wire from one string to the next to use only 1 bms. So you'd make 2 complete 14s100p batteries, then connect the absolute ends Pos/Neg in parallel, then each pack in series will have a smaller gauge wire or bar that connects to its neighbor in the other string. Electrically you'd have 14s200p, but gives you the option to take a string out of service for maintenance.

also the inverter has an ac input from the grid and a bypass mode, maybe i can set this to use this if the draw is ever over 4000w for example to prevent this?

This is called a hybrid inverter. And yeah, if it's the right kind, you can do load sharing. So if your battery is 100p, and you want to only pull a max of 80A from it, then you can set the inverter to only pull a max of 80A, or about 3800W. This would keep the battery from being overloaded. Then the inverter will pull the rest from grid

i was thinking more about how, once i have build the pack (100p for example) do i test that ? and in fact maybe even the whole Powerwall when its together? i dont understand where all that power goes :/

Each cell tested individually. Don't try to do a full build IR test, the readings won't be accurate. Every cell should be IR tested and capacity tested without being in parallel or series with any other cell. This is how you get the most accurate measurements.

It's a long time consuming process. But it gets the job done correctly and you have a better peace of mind when done.

 

[Korishan]

i have tested around 1200 batteries with these, but when i test the (100p for example) packs or even the whole powerwall when finished how do i do that and what equipment is needed?

You do this when you have the proper BMS ready to go. Or you use one of these nice cheap devices, or similar:
DC 6.5~100V 20A 50A 100A Panel Display Volt Amp Power Watt Meter LCD

You'd hook this up with a heavy load, say connected to your inverter or some other load, and start the test and monitor the throughput and the total used power.

DavidPoz has a good video on how to do this:

View: https://youtu.be/yX094DWI-Mo

Actually, here's a better more updated video:

View: https://youtu.be/sJ-sy8Osqr0

 

[Solarmax]

You do this when you have the proper BMS ready to go. Or you use one of these nice cheap devices, or similar:
DC 6.5~100V 20A 50A 100A Panel Display Volt Amp Power Watt Meter LCD

You'd hook this up with a heavy load, say connected to your inverter or some other load, and start the test and monitor the throughput and the total used power.

thank you! ill give that a watch

as for chemistries, i have done a little digging and i know that majority of my batteries are sony vtc4 (inr chemistry) and LGDAHD2c1865 (icr chemistry) so i shouldn't be mixing these in the packs? are there chemistries that are close enough? or that as long as i only put the sonys with sonys in the parrellel packs and the lg with lg's thats ok or is this just a no go area?

think i might cry if after all my efforts i can use half :/ but safety first of course!

thanks again

 

[Korishan]

Really the best way is to look at their charge curves. In their datasheets you should find those charts. If the curves are close, then no worries. Also another thing is if you only charge to 4.1V, or even 4.0V, and only discharge to 3.4V, then a lot of the differing of the curves doesn't fully matter. Except for maybe if you are charge/discharging at max current capacity.

 

[Solarmax]

Really the best way is to look at their charge curves. In their datasheets you should find those charts. If the curves are close, then no worries. Also another thing is if you only charge to 4.1V, or even 4.0V, and only discharge to 3.4V, then a lot of the differing of the curves doesn't fully matter. Except for maybe if you are charge/discharging at max current capacity.

this is music to my ears thank you! i have had a look and the curves of these two seem practically the same, maybe as the are both high discharge batteries? like you say they change a bit at around 3.4 volts so as long as i stick between that and around 4.1 charged i could be good. ill keep reading

 

[Wolf]

i know that majority of my batteries are sony vtc4 (inr chemistry) and LGDAHD2c1865 (icr chemistry) so i shouldn't be mixing these in the packs? are there chemistries that are close enough? or that as long as i only put the sonys with sonys in the parrellel packs and the lg with lg's thats ok or is this just a no go area?

Well according to what I have in my database the US18650VTC4 and LGDAHD2C1865 are surprisingly somewhat compatible.
I say somewhat only because I have a serious suspicion about Sony 18650 cells.
The numbers actually look pretty good at mixing them. The mAh rating on both is close enough and as long as the IR difference is ≤ 5mΩ you should be OK. Of course the optimal solution is all the same manufacturer and same part#. If you do go the route of mixing the Sony and LG you should pepper the LGs throughout the whole battery so they are evenly distributed within each pack.. The SonyVTC style cells "discharge curve" tends to knee at ≈3.27V at a modest 0.5A whereas the LG does so at ≈3.38V. Otherwise they look pretty much the same. Considering they are in parallel in the pack there will be some juggling of mA output between the Sony and LG as voltage will dictate the rate of discharge but if the IR is relatively close it should not be excessive. See my study on how cells behave in parallel. Also the SOH of each cell should not be overlooked and in this scenario a > 85% SOH should be observed.
Preferably 90%+.

Wolf

 

[Solarmax]

Thank you @Wolf !

this is good news so I am hoping ill be ok with this, observing your guidance here of course, so no need to try and keep all the Sony's together and all the lg's together or is that actually worse I wonder? ill be sure to read your thread thank you for the link!

out of curiosity, why do you have suspicions about sony 18650 cells?

 

[Wolf]

out of curiosity, why do you have suspicions about sony 18650 cells?

That suspicion comes from the early days of my testing Sony cells, mostly from the US18650GR G* series. Although they tested well the IR was way to high for my liking. Also the IR suspiciously changed throughout testing some with a starting IR of 102.9mΩ at 2.32V and ending IR of 84.2mΩ at 4.19V. That much of an IR change (18.7mΩ) cannot be good when combining with cells that have an IR spread of maybe 5mΩ between SOC cycles. Similar to the LG S3 series 2200mAh cells High IR but yet good test results with a spread of between 58mΩ to, yea can you believe it, 467mΩ. Of course I would not use such a high mΩ cell in anything other than by itself.

Check out this thread https://secondlifestorage.com/index.php?threads/green-sony-cells-just-suck.5132/
Also @OffGridInTheCity didn't have a good experience with them.

Your worst lithium cells

I went through my rejected 18650's and came across a couple of weird cells. My 2 most weird cells are a pair of Panasonic CGR18650C cells, they hold very little charge and as soon as they hit 3.7 volt, their internal resistance increaes up to 2 Ohm. What are your weiredst batteries you came...

secondlifestorage.com

Mind blowing realization about IR

I should have known this, and it's so blindingly obvious now, but I simply never questioned it. I apologize if this isn't news to anyone, but... I always thought that the IR value of a cell is a stable fixed number that only changes with age. A convenient value that an IR tester spits out in...

secondlifestorage.com

So to answer your question, I am sure it is a preconceived "bad taste" in my mouth to condemn the Sony cells although I did state that if it was the only cells I had and they where all the same I would build a battery with them. The same goes with the LG S3 series.
However these are the Sony GR G* series and the VTC* by all indications are a much better cell.
Besides I really don't like the color choice of that putrid green.

Wolf

 

[Korishan]

Besides I really don't like the color choice of that putrid green.

You don't like baby poo green?!?!

 

[Solarmax]

So to answer your question, I am sure it is a preconceived "bad taste" in my mouth to condemn the Sony cells although I did state that if it was the only cells I had and they where all the same I would build a battery with them. The same goes with the LG S3 series.
However these are the Sony GR G* series and the VTC* by all indications are a much better cell.
Besides I really don't like the color choice of that putrid green.

intersting insight though, ill have to test my IR ratings and differing voltages, but all the charged ones ive tested are around 15mΩ so hopefully that's not too bad, agree with the colour though, good job i cant see it when it will be in my garage!