Jump to content

DIY LifePO₄ Battery


Gnome

Recommended Posts

Hi all

Thought I start a topic to go through my journey of creating my own Lithium Iron Phosphate battery pack.

Overall my objective is to create a battery pack that is equivalent and/or exceeds the quality of pre-built battery packs at a lower price point.  All prices include shipping cost

Parts list:

  • JBD BMS @ R2700 total (once off expense, including shipping)
    • I've used JBD BMS before.  They are without a doubt the best quality you can go while still paying very little.  Plenty of info around about them, compared to Daly BMS for example, would not recommend Daly.
  • 18x Cells @ R750 each @ R13 787 total (including shipping)
    • I see the company already put the price of the cells up.  It might be a good idea to try find another supplier.  These are 2nd life cells.  Meaning they were first automotive grade (highest grade) used in cars and after losing roughly 15-20% of its capacity is no longer suitable for a car but is still suitable for storage.  The advertised capacity already includes the loss of capacity.  In other words, they've tested the cells and re-sell the good ones
  • 250A Class T-Fuse @ $35 (ouch) + Fuse holder - $24 (need something more cost effective here)
    • I already had this fuse.  These fuses are much bigger than the picture makes it seem.  It is about the size of the palm of your hand.  The have very high rupture capacity, 200kA at 120vDC. Meaning it can break 200 000 amps at 120v safely.  It is important to note that battery packs can put out enormous amounts of amps in a very short duration.  If the short-circuit breaking capacity of your fuse is too low the fuse and holder will literally explode into pieces (best case).  I think this fuse is probably overkill so another fuse may be suitable at a lower cost.  I already had it however.  Open to suggestions for another type of fuse
  • DC Disconnector @ $30 (no suitable for voltage and not cost effective)
    • Do not recommend this disconnector.  It is only rated up to 48V and I can audibly hear it make spark noises as I turn it.  It isn't up the task of these high voltage packs.  Still this is what I have at present.  To compensate, I'm adding an anti-spark push button.  A button you press that will connect the battery across a resistor to prevent a spark but allow the inverter to charge up.  The way I implemented mine is, you press the button and it stays active for a minute so you can turn the disconnector.  But overall I really dislike this configuration, so open to ideas for better disconnectors.
  • Aluminium flat bar 30mm * 2mm * 2500mm x2 @ R280 total (once off expense)
    • To connect the cells with each other I need a busbar.  Initially I was planning on using copper but, the equivalent busbar 30x2x2500 would cost roughly R1500.  Aluminium has roughly 60% the conductivity of copper.  I'm aiming at 70mm² copper equivalent which means double stacking a Aluminium flat bar.
  • M6/6mm threaded rod & bolts ( R25 + R40 + R25 = R90 )

Considerations

One of the considerations is what you want your battery's voltage range to be.  This is determined by how many cells you put in series. 

The number of cells is series is denoted by cell count and s.  In other words, 16s = 16 cells series, 17s = 17 cells series, etc.  There is an additional moniker that is more prevalent in the eBike circles, p.  so 16s2p = 16 cells series with 2 parallel per cell.

LifePO₄ has most of its capacity between 3v-3.5v.  That is the voltage range you want to be operating it.  The "nominal" voltage is 3.2.  Although you can "theoretically" discharge your cells to 2.5v, the last 0.5v is the last 10% of the battery and the zone in which you wear down the cell rapidly.  So just don't do it.

Configurations for 48v inverters

  • 16s = 48v -> 56v
  • 17s = 51v -> 59.5v
  • 18s = 54v -> 63v

Notice that 16s is the closest to your typical lead acid 48v battery bank.  However increasing the number of cells increase the watt hour capacity you have and it also decreases the current through your battery bank.

My Axpert inverter can handle, at most, 18s which is what I went for.  I did it for a bit of extra capacity and lower current which theoretically increases efficiency.

Process

First off I sketched how I imaged I would arrange the cells

image.png.cbb39254dbc0e7ec3e741bb44cffc3ea.png

You can pretty much arrange this in any configuration, you just need time to play some lego.

I ordered my cells:

image.png.679883277e1f121812e88b1a00cc3810.png

Having received my cells I was able to accurately determine the size I need in terms of busbar as well as screw size.  My cells were 6mm screws (M6)

In terms of busbar I decided to go with 30mm x 2mm x 74mm.  I calculated that I would want to double stack my bus bar to give me 30mm x 4mm, so in other words 120mm² (equivalent wire thickness).  But with 18s I needed to buy 2x flat bar and thus decided I'll have extra so may as well go for triple stack 30mm x 2mm x 74mm.  That would be equivalent to 180mm² wire.  At approximately 60% of the electrical conductivity of copper, that is equivalent to 108mm² copper wire.  Definitely enough for virtually no voltage drop.

At first I thought I would solder together the two pieces of flat bar on top of each other.  But even using a blow torch the solder would not adhere (tried different fluxes I have, no success).  So essentially I'll just bolt them down

image.png.f6ba47f32106e28df42c6c258d44d918.png

Yes these aren't perfect.  I cut the holes with a hand powered drill and 6.5mm bit.  And the flat bar was cut with a hack saw.

image.thumb.png.84aae001b8d894bef41d03985a3ff440.png

To clean up the flat bar after I drilled it, I used a deburring tool I have and then cleaned up the plate with steel wool.  I've used the steel wool before to clean up copper pipes and it leaves a really nice finish.  As you can see here, if you overlook how the holes don't look completely center and the deburring was done hastily but the finish looks sort of decent.  Making these bus bars up to what you see here took roughly 6 hours, its slow work.

Equalize the cells

Before connecting the cells in series, the general advise is to connect all the cells in parallel and then charge them up to 3.65v and keep them there until the current being consumed is very low.

image.png.49b57fb9ad7301c4423c7bbdc9606d3e.png  image.png.622a89f57bb5ca724a4a501f807af17a.png  image.png.752c673cc5d3fee4efdd88fa226ddd53.png

I already have a bench power supply that I'm using for this purpose  But progress is pretty slow.  I suspect the batteries were provided to me near empty, I'll probably some more days to go before these cells are anywhere near fully charged.

The BMS I'm using:

image.png

And the positive side disconnector I created with a precharge toggle switch to avoid that spark when you turn it on: (the switch is momentary, so it has to be held in position)

image.png

Fully assembled battery:

image.png

BMS app:

image.png  image.png

The BMS without load versus recharging after I put some load on it

 

Edited by Gnome
Add progress update
Link to comment
Share on other sites

That aluminium bar might be thicker but the surface are which touches the terminals is still the same. So, that’s your bottleneck for conductivity. If that area is not sufficient for aluminium, it doesn’t matter how wide and thick, the rest of the bus bar is.
I have never seen in any videos or DIY guides, anyone using aluminium bus bars. But I have seen people polishing the terminals and the areas of the copper bus bars touching them, in order to improve the conductivity even more. 

What is your calculations for the total cost of the battery anyway?

Link to comment
Share on other sites

22 minutes ago, PowerUser said:

That aluminium bar might be thicker but the surface are which touches the terminals is still the same. So, that’s your bottleneck for conductivity. If that area is not sufficient for aluminium, it doesn’t matter how wide and thick, the rest of the bus bar is

Thanks for commenting.  The thickness of the busbar is not about the maximum current capacity.  That contact point can carry a lot of amps but at some loss. However loss is calculated over distance.  Thus thicker busbars will have lower losses.  So yes the point of contact has a maximum current rating (hundreds of amps).  But the overall capacity of the busbar is still limited by size because of the voltage drop across the conductor for the given length.

22 minutes ago, PowerUser said:

I have never seen in any videos or DIY guides, anyone using aluminium bus bars. But I have seen people polishing the terminals and the areas of the copper bus bars touching them, in order to improve the conductivity even more. 

The busbars I've seen sold on Ali, etc. seem to be aluminium.  There really isn't any reason not to use it.  But if there is something I'm not thinking about let me know.

22 minutes ago, PowerUser said:

What is your calculations for the total cost of the battery anyway?

Because I already had some parts to begin with (disconnector and fuse), my total cost will be R16 767.  The recurring cost is R13 787, as the BMS, busbars and so on have no replacement interval.  I could also have used a cheaper BMS, which seems common with pre-made kits (ie 100a BMS), however opted for overkill (300amp continuous, 600amp max)

My battery bank is also 5940Wh apposed to the more common 5280Wh of a 16s configuration.  If I did 16s, same as all pre-assembled kits, my cost would be reduced by R1410.

Edited by Gnome
Link to comment
Share on other sites

Have you included in the cost a nice metal enclosure for everything to be packed nicely and mounted on the wall securely, not obstructing anything?

in order to be able to compare the cost vs commercial battery, that’s a must

What happens, if your BMS goes faulty? How long, until you are able to use the battery again?

Link to comment
Share on other sites

1 hour ago, Gnome said:

Aluminium flat bar 30mm * 2mm * 2500mm x2 @ R280 (once off expense)

  • To connect the cells with each other I need a busbar.  Initially I was planning on using copper but, the equivalent busbar 30x2x2500 would cost roughly R1500.  Aluminium has roughly 60% the conductivity of copper

Rectangular conductor resistance calculator here  You can use this to calculate the equivalence resistance for Ally bar vs Copper.

Link to comment
Share on other sites

22 minutes ago, PowerUser said:

Have you included in the cost a nice metal enclosure for everything to be packed nicely and mounted on the wall securely, not obstructing anything?

in order to be able to compare the cost vs commercial battery, that’s a must

I've already built a "cupboard" in which my Lead Acid batteries went, so it'll just go in there.  Making an Alu enclosure is not for the faint of heart so I won't even try to go into that here.  But building an enclosure out of wood is pretty easy stuff so that seems reasonable.  Alternatively a plastic enclosure of sufficient size could work too.  Or buy an Alu enclosure.

22 minutes ago, PowerUser said:

What happens, if your BMS goes faulty? How long, until you are able to use the battery again?

I'd probably need another BMS.  I mean I could bypass it in the meantime.  But like I said, I've already used this BMS in my eBike (60a version), my battery once flew off my bike and the enclosure I bought flew apart.  Even with all that the BMS kept going 😂.  It is also one of the most widely hacked and endorsed BMS out there.  There is a LOT of info available on this BMS, including protocols, and how to videos (decent review here).  So I have very little worries about the BMS.  If anything I'm more concerned about the cells themselves, I have no experience with them.

Why do you ask?

Link to comment
Share on other sites

14 minutes ago, Sarel said:

Rectangular conductor resistance calculator here  You can use this to calculate the equivalence resistance for Ally bar vs Copper.

Thanks I used a calculator before this.  I just converted the bus bar size to the equivalent mm² and then compare to Alu wire of the same mm².  I tried a couple of these calculators and they each had a different result and totally off of what cable equivalent ratings are.  One calculator was completely under rating and the other completely over rating.

 

Link to comment
Share on other sites

@GnomeI’m asking to find out, if you have a better plan than me. I considered for a while building my own LFP battery bank but if you add on top of your cost, a nice metal enclosure and a good bench power supply to top balance the cells initially, plus all the time and extra resources to program the BMS, the cost of the whole thing will be the same, if not even higher than a nice commercial battery of the same size. 
 

And on top of that if something breaks - cell or BMS, it will be a long time until one gets a replacement from China. 
 

That’s why at end, I just went and bought a Hubble. Inside the Hubble, the terminals are even laser welded to the buss bars. Something not easy to do DIY in any case. 

Edited by PowerUser
Link to comment
Share on other sites

32 minutes ago, PowerUser said:

@GnomeI’m asking to find out, if you have a better plan than me. I considered for a while building my own LFP battery bank but if you add on top of your cost, a nice metal enclosure and a good bench power supply to top balance the cells initially, plus all the time and extra resources to program the BMS, the cost of the whole thing will be the same, if not even higher than a nice commercial battery of the same size. 
 

And on top of that if something breaks - cell or BMS, it will be a long time until one gets a replacement from China. 
 

That’s why at end, I just went and bought a Hubble. Inside the Hubble, the terminals are even laser welded to the buss bars. Something not easy to do DIY in any case. 

Gotcha, makes sense.  I already had a bench power supply so the topping charge wasn't a big problem for me.  I definitely considered the Hubble for a long time.  In the end what pushed me over is that they don't offer a strategy to just replace the cells at a reasonable cost.  It does seem the initial cost will be close to the commercial solution (if you had to buy everything, for me there will be still roughly R6k saving).  Starting from nothing, you'd probably save no more than R2k (not worth the saving)

When you start talking about 10kWh batteries the cost quickly becomes much more compelling.  A significant amount of the money goes into the BMS.

32 minutes ago, PowerUser said:

plus all the time and extra resources to program the BMS

It is a Bluetooth Android app.  Super easy.  Don't get me wrong, there are some things that are better on prebuilt, but the JBD BMS is definitely better than what you'll get in a pre-built.

Edited by Gnome
Link to comment
Share on other sites

Today the only thing I did was connect the cells in series parallel and top charge them.

image.thumb.png.2d85ba6661e299d9c1764100d97a5b5d.png

I've also discovered another forum where a number of people are making their own batteries.  Having read through the threads there and seeing the advise given, I would not advise anyone to buy 2nd life cells from https://lithiumbatteriessa.co.za/

They are already putting their price up and there is plenty of advise on that forum on how you can get brand new A grade cells for basically the same price.  This isn't a reflection on these cells per-se but rather on a business selling 2nd life cells at a price that is essentially what you would buy new cells at.  It doesn't make financial sense.

Yes you'll have to wait 60 days for a ship to arrive in SA, but this sort of thing shouldn't be rushed.

Edited by Gnome
Fix mistake about cells being in parallel, not series
Link to comment
Share on other sites

9 hours ago, Gnome said:

I've also discovered another forum where a number of people are making their own batteries.  Having read through the threads there and seeing the advise given, I would not advise anyone to buy 2nd life cells from https://lithiumbatteriessa.co.za/

They are already putting their price up and there is plenty of advise on that forum on how you can get brand new A grade cells for basically the same price.  This isn't a reflection on these cells per-se but rather on a business selling 2nd life cells at a price that is essentially what you would buy new cells at.  It doesn't make financial sense.

Yes you'll have to wait 60 days for a ship to arrive in SA, but this sort of thing shouldn't be rushed.

I don't think you have a chance of getting the cells in SA cheaper than from LBSA in the quantities an individual would buy. Shipping of lithium cells is very costly. I'd love to be proven wrong on this though, so if you have contacts, please share.

Link to comment
Share on other sites

10 hours ago, 87 Dream said:

If you told me you had them in Parallel I would say that this will take much longer as the ah goes up & the Voltage stays the same & In this way no BMS is required because the pack voltage will not exceed 3.65V. 

Sorry I can't make out by your pic if series or parallel is paired here but you mentioned series so I have to accept that on face value.

Ugh sorry yes, late night post.  I meant to say parallel.  I'm charging them at 3.6v @ 5amps.

I'm still waiting on my BMS so I don't actually mind too much if it takes a long time to bring them all up to the same level of charge.

Edited by Gnome
Link to comment
Share on other sites

34 minutes ago, 87 Dream said:

Awesome, 😎 I thought as much. Yes it will seem like nothing is happening for many many days. 

I am ordering a larger 30A charger for this very reason. Mine is also the Humble 5A version like yours. 

87

 

I do a LOT of electronics, so I already had the 5a lab power supply.  I don't think I would have bought one just for this 😛 

I'll give your 4 cells in parallel thing a try if it isn't working for me but I can see the voltage slowly ticking up.  2.35v at the moment for the pack.

Since my BMS hasn't even shipped yet, I fully expect to either get the order cancelled or not see it for another month (Ali sellers are just so unreliable), plenty of time to get the cells ready.

image.png.4e979877555b7939f0aa83c9ffb049dc.png

Edited by Gnome
Link to comment
Share on other sites

7 hours ago, P1000 said:

I don't think you have a chance of getting the cells in SA cheaper than from LBSA in the quantities an individual would buy. Shipping of lithium cells is very costly. I'd love to be proven wrong on this though, so if you have contacts, please share.

Yeah right there with you.  Plenty of advise on the other forum I linked to on what to buy.  I have no idea what the shipping will cost.  It comes by boat.  Delivery by plane would be insanely expensive.

Group buys is probably the way to go.

Link to comment
Share on other sites

On 2021/11/14 at 7:20 PM, Raj35 said:

@Gnome how is  the top balancinng going.

Hey @Raj35

Cells finished balancing last night.  All cells are now sitting at 3.65v.  The cells were probably quite empty to have taken this long.  I also tried not to charge during night time when I slept.

Now waiting on my BMS to be delivered.  Once that is done, the next stage will commence which is actually building the battery.  I'll include pics of that phase of course.

Link to comment
Share on other sites

BMS finally delivered today.  What a wait.  It got stuck in China for more than a week.  Even with Fedex.  When I googled, it seemed to be a problem lately, since COVID.

Anyway, first things first, they lied about the amps.  It says 300amp on the sticker but the relay is rated for 200a.  So it is a 200a BMS.

image.thumb.png.dab0a34f0a95dd6ddfaa8e468843bc9e.png

I also went ahead and removed the 4 AWG wire (~25mm²) they had connected to the BMS and replaced it with 70mm²

Lastly the copper busbar had a bunch of finger prints and glue on it, so I took it off and polished it using steel wool (have done this before when I redid the copper pipes in my apartment).  Mostly in places it needs good contact to make sure once I tightened it down hard it would have excellent contact.

image.thumb.png.db1ee1de40a5f2e52ed50976bdea02d1.png

Link to comment
Share on other sites

@Gnome quick question, on what brand inverter/mppts are you planning to run the bms? I got 4 it's way with someone shipping order that order more and lifepo4 batteries.

I had one on a test setup, they work really nice, but when the contactor opens because the safery, say a sell hits the 3.65v, it can cose a bit of a problem. 

 

Louis written a script so that it can work witj Victron equipment. 

Link to comment
Share on other sites

FYI: I've watched a number of videos now, Daly BMS is pretty low quality BMS do not buy

1 hour ago, Gerlach said:

@Gnome quick question, on what brand inverter/mppts are you planning to run the bms?

Your BMS has nothing to do with the Inverter or MPPT.  But I am using an Axpert King.  I'm purely using my setup for backup/UPS.  I don't have space for solar, I live in an apartment in Sea Point.

The BMS is purely to do two functions:

  1. Protect the cells, so that no one cell is over discharged or overcharged
  2. Balance the cells during charging so that all the cells can reach the same charge level

The BMS does provide voltage per cell and Ah/kWh in the battery but that is purely informational.

1 hour ago, Gerlach said:

I had one on a test setup, they work really nice, but when the contactor opens because the safery, say a sell hits the 3.65v, it can cose a bit of a problem.

Yes but why are you charging your cells to 3.65v?  You should cut off at roughly 3.48v per cell.  That is about 98-99% charge level. That last 1% of charge drastically reduces the life of your cells for virtually no benefit whatsoever.

Lastly, I've done a LOT more research on BMS while I was waiting on BMS.  The latest recommendation is to get this BMS: 

Heltec/JK 150a BMS

Reasons:

  1. It doesn't use a relay, so it can separately control charge/discharge
  2. It has actively balancing instead of passive*.  On a set of half decent cells active balancing is overrated but if you plan to run the setup for a long time, it is definitely worth considering
  3. In terms of build quality it is just as good as the JBD BMS, the app is just as good and it can be configured with equally many options as the JBD BMS.

* Active balancing = Power is transferring from over charged cells to the lowest voltage cells.  Passive balancing = Power is dissipated as heat on overcharged cells through a resistor.

In my opinion, the only downside to that JK BMS is that the wires are permanently soldered to the board of the BMS, so you can't swap it out for higher mm²/gauge wire.  And these company are always using really thin wire (with high temperature rating, but that still means you burn some energy as heat in the cable).

1 hour ago, Gerlach said:

Louis written a script so that it can work witj Victron equipment. 

I assume you are referring to this.  Very cool indeed for Victron owners.  The BMS I mentioned above is also supported.

Edited by Gnome
Link to comment
Share on other sites

@Gnome thanks for the feedback. I waiting for your feedback in the  few weeks when you got your setup up and running. 

 

Ooooooo yes. Just to add, I got a 35kwh lifepo4 diy storage storage and in the diy lifepo4 game more then 3 years, so I know my way arround the stuff and how it work. Watching videos and playing with the real stuff is two different games 😉

 

Enjoy the adventure. 

 

Link to comment
Share on other sites

image.thumb.png.0a4151457f9b89dc8af91a95cc6a7d75.png

Fuse going to the negative battery terminal.  I couldn't get any 70x6mm lugs and I had some 50mm cable lying around so used 50x6mm x 2.  These colours are standard SANS colours so my local cable seller (who is also cheapest) doesn't have them in other colours.

Lastly if you are wondering what the bulging under the heat shrink is about: I put hot glue on over the the end with a glue gun where the insulation ends, the bit of copper sticking out under the lug and then the part of the lug where the wire goes in.  When I put the heat shrink over and blow on it with the heat gun it all starts to melt and spreads out very nicely, creates a moisture and corrosion barrier.  You can actually buy heat shrink with glue inside (the same type as glue guns actually) but they are crazy expensive, so this is my cheaper version of that which should be nearly identical except for not looking quite as nice (bought version spreads out a bit more evenly)

I also took the time last night to crimp some lugs on the ends of the balance wires.

image.thumb.png.1c4d64bd9e3f68cd64ec7b44ecb449ba.png

The wires without crimps are extra (BMS is up to 24s IIRC) and I'll see if I need to either connect them to the last terminal or just disable it in the app (probably both I guess)

Edited by Gnome
Link to comment
Share on other sites

1 hour ago, Gerlach said:

@Gnome thanks for the feedback. I waiting for your feedback in the  few weeks when you got your setup up and running. 

 

Ooooooo yes. Just to add, I got a 35kwh lifepo4 diy storage storage and in the diy lifepo4 game more then 3 years, so I know my way arround the stuff and how it work. Watching videos and playing with the real stuff is two different games 😉

 

Enjoy the adventure. 

 

Ah ok gotcha :)  This is my first time doing LiFePO₄ and I only really decided to go this route on a whim. (AGMs were basically dead and thought I'd do a few googles).

Overall it is looking quite positive.  My one minor worry is the Aluminium bus bars, but I see a view videos online where people got Aluminium bus bars with their batteries and even kind that you can buy professionally.  So I doubt I'll have trouble.

To double ensure I won't have trouble I'm also getting some Noalox, a aluminium electric connection corrosion inhibitor.  When you apply it, it kinds of crips under the Aluminium oxide (Aluminium oxide is the immediate outer layer of aluminium that forms near instantly).  It removes that Aluminium oxide and coats it with a supposedly more conductive metal.  Videos of the creeping looks pretty cool and I have read plenty of people use it on LiFePO₄ batteries between Aluminium and copper bus bars (it also acts as an aluminium to copper corrosion inhibitor).

Worst case, if I need to replace my Aluminium bus bars, I'll be out a two days of work I put in and about R250.  Not bad when you consider a copper bus bar is >= R2000 for what you need to for (essentially).  The kind they sell on AliExpress is definitely under sized.

Link to comment
Share on other sites

Box that will go to the positive terminal with a disconnector and precharge to prevent that spark/pop when you first connect the inverter

image.png.9ef88390db15e73ef39c089d53d9c352.png

The precharge is a momentary switch (so if you flip it to on, you have to hold it or it springs back) and a 100Ω 100W resistor.  I usually used 100Ω ohm resistor that I put between the terminals for this purpose so I know it works just fine.  I went with 100w because when I used a 10w resistor it got pretty damn hot and the maximum calculated heat is around 40 watts (so double for safety).  This is really not a big deal, it shouldn't need to be that big, but if someone is an idiot and keeps holding that switch it'll do just fine.

The plastic box is just a Veti box I got for R50 at Communica.  I drilled a hole using a hole saw for the disconnector to fit through.  Used a stepped drill bit to drill the hole for the cables and the momentary switch.  To try make the hole look a bit better I also used my deburing tool going in circles.  Worked well on the drilled holes but the whole saw hole you can probably see is a bit ehh, but I think it turned out well enough

Edited by Gnome
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...