Organising salvaged cells in a DIY battery

[ChrisW]

So I've salvaged a number of 18650s from unused laptop batteries. Their capacities range from 1,800 - 2,500 mAh. Their internal resistance varies from 50 - 150 mOhms. I want a 10AH 7S (nominal 24 - max 29.2V) battery.

My question is how best to organise those cells. I've read that you can place differing cells in parallel and add their capacities to make cell banks which may have different number of cells in, but will have similar capacities, then connect those banks in series. But I don't know what, if anything, to do with the internal resistance variations.

Does anyone have guidance for me, please?

And with all this complexity and the stress around possible disaster scenarios, I'm seriously considering just buying two SLA 7.5AH batteries and hooking them in series. It's not that much bigger nor that much heavier - the weight of the books I carry on the bike far exceeds either. It's a commuter bike.

Any thoughts on that also?

Thanks,
Chris

 

[Danidl]

That is a fun project, and if the parts are free probably a good learning experience.
The nominal discharged voltage of a lithium cell is around 3.2 volts and the fully charged version around 4.2 volts. Nominal value 3.6v
In order to achieve the 30 volts you are seeking asking you will need to have 8 to 10 cells in series connection. I would call that 8 to 10 tiers. The capacity of the cell is hopefully 2 amphrs , so a 10 amp hr battery will require a minimum of 5 cells in parallel. .. so you are talking about 50 cells minimum.
The internal resistance of a cell is a fair reflection of its remaining lifetime or storage capacity.. new cells will have lower values and as they wear, their resistance increases, and their capacity drops. However there will be other cells with higher capacity, but also higher initial resistance...so it's not quite foolproof.
If you had a large population of identical cells, but with different internal resistances, you should sort them into piles of low medium and high. Then you should arrange that each tier has as near as possible, an equal mix from all three piles. ... The total internal resistance of a tier is the parallel combination of all the resistances in that tier. Some of the cells will have less capacity than others , but the total capacity of any tier will be the sum of the individual cell capacities.
If the cells are at nominally the same voltage when they are connected in parallel to make a tier, then all is fine, but if one of the cells is discharged and others charged, large currents can flow during the equilibrium phase. .. so charge then all before joining them togetter for the first time.
You need to think on how you are going to join 5 or 6 cells together in parallel, .. spring loaded battery holders would not be the best idea and commercially spot welding is done. Soldering is theoretically feasible, but heat damage to the cells is a concern.
You need to think on how you are going to connect the tiers together to make the series connection, and then how to charge them up to the required voltage.
The safest way of doing that would be to get a 4.2 volt, large current DC supply and charge all the cells in parallel. .. the5volt power supply from a old desktop pc, dropped to 4.1 volts would be suitable, as they can supply up to 40 amps. .. look on line to see how they can be turned on in the absence of a motherboard..
When the entire set of cells are charged to 4.1 volts, theu can then be switched into series mode and then used on the ebike.
The other way is to acquire a battery management system which is a board and chip which has an electrical connection to each tier and after the battery has charged to the full series value, it redistributes and equalises the voltage accross each tier .

The other alternative is as you say to buy two mobility scooter batteries ...lead acid and use them. The problem there is that if they last beyond 200 charge discharge cycles you will have done well, and you will notice degradation after 20 cycles... and they cost money..

 

[ChrisW]

Thanks for all your time and effort Danidl. Very helpful, and the biggest thing I didn't know was the rate of degradation rate of SLAs. So I guess I'll press on with the build. Thanks again.

 

[Nealh]

For 7s it is 7 parallel groups / strings of cell you will need.
You can parallel as many as you like in each string.
Using old lap top cells with cell with high IR will lead to very poor capacity and dire voltage sag, you may be ok with cells at about 50/60 ohms max. The high IR cells crash very quickly down the voltage scale, as a project a worth while learning course but for a reliable battery with range not so.

 

[Deleted member 4366]

7S is 24v or 29.4v fully charged, not 8S. You need at least 15 amps at 24v to power an ebike. With used laptop cells you can't go higher than 1C, so you need at least 15AH.

 

[ChrisW]

Thanks again guys.

 

[anotherkiwi]

If you are commuting on the flat why not. If there are hills involved you are wasting your time with those cells. The only solution is to do what Tesla does and go massively parallel = big battery and lots of weight. By massively parallel I mean at least 10P. That could be 2 x 5P7S batteries connected in parallel.

 

[ChrisW]

Hmmm. With a nominal 100 milli Ohms IR, the 10-ish amps needed to drive my 25V system at its rated 250 watts. Across 0.1 Ohm, 10 amps drops 1V and dissipated 10 watts. That's 70 Watts total in the 7P pack. The bike woks fine at 23V That's where my theoretical analysis runs out of steam. I was wondering abut real-world practice.

If I mix low and high IR cells in parallel within each cell bank, will they "sort themselves out" or blow themselves up? I guess the voltage in the high IR cells would droop more and so the low IR cells would source more current. Difficult to analyse and of course, ideally, you wouldn't need to because you wouldn't do it at all. You'd use new, balanced cells.

BTW these laptop batteries are unused so I'm not clear why I should limit them to 1C if their spec says 3C. Safer?

Then there's the matter of measuring IR. I'm using my iMax B6 mini, but I already know it contradicts itself - measuring very different IR values depending on whether you're measuring cells on their own or in a string. If I had an accurate current-measuring device I'd do it manually. I have one on order.

Then there's the meticulous versus gung ho camps. I'm in the "can't afford to do it "properly"" camp, but it'd be nice not to set fire to anything.

Past experience says the nay sayers should be listened to but not blindly obeyed. Do it anyway, but keep a keen eye out for what they warned you about.

What I really need is a sugar mumma to take me to top-of-the-lines-ville.

 

[anotherkiwi]

From what I have read about making parallel strings it is best that all the cells are of similar resistance, that way all the cells in the same parallel string charge to 4.2v at the same time and stay balanced.

OK do you have 35 cells of identical capacity? Make a 5P7S battery from them. Maybe you have 14 x 2500 mAh cells? Make a 2P7S battery. When you are done making batteries of same mAh then you can charge them all up to 29.4v (they are new cells so they should charge to full) and then connect those batteries in parallel. Your total Ah is the sum of all the batteries connected in parallel, the higher the better.

 

[Nealh]

The high IR cell/s will sag voltage wise much quicker then the better ones, they will either pull the others down to compensate or the weak cell will crash and hit lvc faster.

I have some lipo's with the odd higher IR cell ( 20 -37 ohms) and they pull the pack down quite quickly esp under higher load, the higher IR cell is always about 0.2 -- 0.4v lower which doesn't sound much. But equates to loss of range and unused ah from a pack, as when lvc is set at 3.5v min the other cells still have good usable capacity if they are sitting at 3.7/3.9v. Luckily with lipo's I can disconnect any pair of bad packs and reconnect remaining better ones to eke out the ah.

Lipo cell IR is a bit different to 18650, even 37 ohms is up 10 x the IR of good cells.

 

[Nealh]

Laptop cells usually aren't top class amp cells may be 4a rated though may be a from a good manufacturer like LG or Sanyo etc.

DO USE A BMS IF YOU DON'T WANT A FIRE.
Generally if a cell fails the pack will shut down if charging or discharging.

I'm with AK separate the differing capacity (Ah) cells in to different 7s packs and match the IR as close as possible (bin any very high IR cells or use for torch light/bike light), then they will charge/discharge & balance better. You can then parallel differing capacity (Ah) packs together to form one large Ah pack. Discharge will be more even with voltage drop more linear any high IR cells need binning.
That said a higher IR parallel pack will still bring down the better pack quicker and reduce it's life.

 

[Deleted member 4366]

Hmmm. With a nominal 100 milli Ohms IR, the 10-ish amps needed to drive my 25V system at its rated 250 watts. Across 0.1 Ohm, 10 amps drops 1V and dissipated 10 watts. That's 70 Watts total in the 7P pack. The bike woks fine at 23V That's where my theoretical analysis runs out of steam. I was wondering abut real-world practice.

If I mix low and high IR cells in parallel within each cell bank, will they "sort themselves out" or blow themselves up? I guess the voltage in the high IR cells would droop more and so the low IR cells would source more current. Difficult to analyse and of course, ideally, you wouldn't need to because you wouldn't do it at all. You'd use new, balanced cells.

BTW these laptop batteries are unused so I'm not clear why I should limit them to 1C if their spec says 3C. Safer?

Then there's the matter of measuring IR. I'm using my iMax B6 mini, but I already know it contradicts itself - measuring very different IR values depending on whether you're measuring cells on their own or in a string. If I had an accurate current-measuring device I'd do it manually. I have one on order.

Then there's the meticulous versus gung ho camps. I'm in the "can't afford to do it "properly"" camp, but it'd be nice not to set fire to anything.

Past experience says the nay sayers should be listened to but not blindly obeyed. Do it anyway, but keep a keen eye out for what they warned you about.

What I really need is a sugar mumma to take me to top-of-the-lines-ville.

You need to listen to people that have actually made batteries from laptop cells instead of making theoretical calculations. Firstly, whatever the spec says, you can divide it by two for practical use in an ebike, so 3c cells in a 10Ah battery would match with a 15A (max) controller. Secondly, you need to divide that again by at least another two because they've already been used, so at best, they won't perform like they did when they're new. At worst, they're knackered.

At what current did you measure the internal resistance?

18650 cells really do wear out as anybody with an ebike will see. When they're worn out we sling them. We don't disassemble the battery and use the cells again to make another battery. Laptop cells are normally rated much lower than ebike ones, so they're going to be even worse.

You can buy a pretty reasonable 24v ebike battery for £150 that includes a charger and BMS. To build your battery, you need to buy a BMS, nickel strip, insulators, wire, connectors, heat-shrink, charger, welder and test equipment. You won have anything left out of £150.

If you want to build a battery, use new cells. It'll be a lot more expensive than a ready-made one, but you will have had some fun and learnt a lot, and you can use your equipment to update it when you want.

 

[Danidl]

From what I have read about making parallel strings it is best that all the cells are of similar resistance, that way all the cells in the same parallel string charge to 4.2v at the same time and stay balanced.

OK do you have 35 cells of identical capacity? Make a 5P7S battery from them. Maybe you have 14 x 2500 mAh cells? Make a 2P7S battery. When you are done making batteries of same mAh then you can charge them all up to 29.4v (they are new cells so they should charge to full) and then connect those batteries in parallel. Your total Ah is the sum of all the batteries connected in parallel, the higher the better.

What you have read is in error... . Logically all cells at the same tier can be considered as the same cell, just of a larger capacity. They have the same voltage drop accross them and as the combination contains more mass of material hence has greater capacity. If at an initial point one cell had a different voltage than its neighbours, current would flow between the cells to minimise that and eventually they will come to and remain at an equilibrium... So it does not matter whether the individual cells have different internal resistances. If a voltage is applied to the tier, charging up the cells, charge will flow into each cell proportional to its internal resistance..likewise during discharge.
What happens in a string of tiers is different. The total voltage, is the sum of the tier voltages, the same current is taken from each tier, but the voltage accross each tier will differ depending on the internal resistance of that tier. As the battery depletes, charge drawn from any cell increases the internal resistance of that cell. Once any tier has gone low, the battery cannot supply current, even if other tiers had power. It is better then to minimise the total resistance of each tier, and to try and ensure that each tier has the same capacity.

 

[spanos]

I have built a 36v battery out of salvaged laptop cells.

On the plus side I learnt an absolute ton.

On the downside it took many, many, many hours to disassemble, test and group similar cells.

And more time to solder /weld up

I only hooked it to a 36v 200w motor. It was a 10s 6p btw

It didn't sag too badly for the first few runs on the gentle flat

But the whole thing quickly died as the already worn cells packed up

I'd never do it again !

 

[Deleted member 4366]

Yes, you can put different capacity cells in each string. That could work well if you wanted a 3.7v battery, but as soon as you go to higher voltage with multiple strings, you have to have the same capacity in each string, which is almost impossible to achieve with random used cells. You need to test every cell for capacity and then pray that they stay the same, which they won't.

The other problem with using worn out cells is that they sometimes go short-circuit, which will drain all the cells in the string and render the battery useless.

Let's make it simple. Whatever the theory says, just about everybody I know that tried it, including myself, agree that it was an interesting project, but a complete waste of time, cost and effort to get a working ebike battery.

Sorry to rain on your parade, but I'm just trying to make life easier for you and I can't change reality.

If you want to play with ebike batteries, go to your local ebike shop and ask them if they have any scrap batteries they're trying to get rid of. If you're lucky, some will be new, but have faulty BMS's or shorted cell-strings, so there's plenty of serviceable cells. I just recently got a new 50 cell one with one string drained to zero volts from a shop. I reconfigured it to 40 cells and got a new BMS to make a custom battery for a project, and it's working perfectly. You can never trust the BMS in a faulty battery unless you can clearly see that the cause is elsewhere, which is why I changed it.

 

[ChrisW]

Danidl - thanks for your interesting insights. Spanos - the benefit of experience - thanks. Kiwi, neath - more opinion points - thanks.

The battery I have is a dog's dinner soldered brutally together by a gorilla with a blow torch, from a patchwork of different cells. It has no BMS. The person who sold me my Ebike told me he bought the battery from a man at a car boot sale. So real pedigree.

I've owned the bike since last April and used it most weekdays and been charging it from a simple power brick until recently. Even from that world of theoretical misery the battery has served me well and continues to do so, though it's losing capacity noticeably now, which is why I want to build new one. I don't think it will be worse than the one I have.

Online, I see people building powerwalls, and EBike batteries from salvaged cells, and reporting success. Also, I made a spot welder, and I have the cells. They are from unused laptop batteries. And finally, I simply don't have the money to buy a new battery or new cells to build one with., so I'll build the battery and see how it goes.

My request for advice was about organising the cells. From what I've heard here and elsewhere, I'll worry less about IR, try to balance the capacity of the series banks by using uneven numbers of cells where necessary, and balance manually.

From the other comments you offered outside my question, which I appreciate, I'll expect poor performance and short battery life, but I'll see how it goes anyway.

Thanks again.

 

[wheeliepete]

If it helps you at all with your battery project Chris, I built a 36v 10P10S pack from new generic chinese 2000mAh laptop cells a couple of years ago. I knew nothing of IR, just charged them all to 4.2v, discarding any that were under or over that voltage. Welded them all together with nickel plated steel ribbon(not the best, but very cheap) and fitted a £7 BMS. It worked really well and gave me a range of around 20 miles on a throttle only 250 watt bike. Two years on a friend has it on his bike for his 8 mile commute. I would also advise fitting a BMS to your pack for ease of charging and safety.

 

[Danidl]

Danidl - thanks for your interesting insights. Spanos - the benefit of experience - thanks. Kiwi, neath - more opinion points - thanks.

The battery I have is a dog's dinner soldered brutally together by a gorilla with a blow torch, from a patchwork of different cells. It has no BMS. The person who sold me my Ebike told me he bought the battery from a man at a car boot sale. So real pedigree.

I've owned the bike since last April and used it most weekdays and been charging it from a simple power brick until recently. Even from that world of theoretical misery the battery has served me well and continues to do so, though it's losing capacity noticeably now, which is why I want to build new one. I don't think it will be worse than the one I have.

Online, I see people building powerwalls, and EBike batteries from salvaged cells, and reporting success. Also, I made a spot welder, and I have the cells. They are from unused laptop batteries. And finally, I simply don't have the money to buy a new battery or new cells to build one with., so I'll build the battery and see how it goes.

My request for advice was about organising the cells. From what I've heard here and elsewhere, I'll worry less about IR, try to balance the capacity of the series banks by using uneven numbers of cells where necessary, and balance manually.

From the other comments you offered outside my question, which I appreciate, I'll expect poor performance and short battery life, but I'll see how it goes anyway.

Thanks again.

OK but you really need a BMS, but they are cheap. If the cells are actualy unused, then they may well give you a good life, therefore justifying the money spent on the BMS.

 

[Deleted member 4366]

The battery I have is a dog's dinner soldered brutally together by a gorilla with a blow torch, from a patchwork of different cells. It has no BMS. The person who sold me my Ebike told me he bought the battery from a man at a car boot sale. So real pedigree.

I've owned the bike since last April and used it most weekdays and been charging it from a simple power brick until recently. Even from that world of theoretical misery the battery has served me well and continues to do so, though it's losing capacity noticeably now, which is why I want to build new one. I don't think it will be worse than the one I have.

You're very very lucky. I'm not really one for health and safety and I'm not scared of taking calculated risks, but to run an ebike lithium battery without a BMS is extremely dangerous. I've seen them explode, and I can tell you that your neighbours would be calling the police if one went off in your premises.

 

[ChrisW]

Wheeliepete, thanks for your data point. It does help.

Danidl, I have a BMS - one of those cheap Chinese things, but testing showed problems with it, and putting a cheap Chinese brain in charge and in permanent connection to all that power seems more dangerous rather than less. My bike has an instantaneous voltage readout, so I know I'm never dipping into the danger zone. On-load, it never goes below 3.2 volts per bank, and I balance charge it at least twice a week, so I think I'm OK, but if I've missed something big, please let me know.