Modular battery built from power bank enclosures

[rafalg]

Hi, two weeks ago i was spammed by an ad from Aliexpress and had this sudden idea - build a modular / solderless battery from 18650 power bank boxes. They were sold for less than $3 each, and i thought it might be a good idea to use them just as battery holders at that price - so i ordered 10.
Two days ago they have arrived and i started the build. After inspecting the box I concluded they can still work as power banks, just need to add wires so they can be connected in series.
Each box can hold eight 18650 cells (one more could be added after throwing away the PCB, but it seems like too much hassle with making another pair of contacts for the extra cell. So i stayed with the original design, with the exception of LEDs that were also included - after removing the LEDs i get nice holes for the wires.

After putting the cells in they will still work as power banks, so there's power level indicator working and USB ports can be used for charging (no idea what to do with these yet).
And the plan is to connect everything with WAGO connectors. I'm using 3 way connectors so the third connection can be used for balancing and for the BMS.
The 'bus bars' on the sides seem thick enough for some usable current. I'm worried a bit about the reliability of the spring loaded connections with the cells, but time will tell if they work or not. In case of problems i think these can be improved a little by tinning.
The battery, when finished, will be used to power a lawn mower - there's enough space for 10 boxes, they just need to be fastened somehow to the mower enclosure. I expect the working current to be around 20-25 amps, so it's like 3amp per cell.
The idea behind this build is to have a modular, serviceable battery, where cells can be replaced if any section goes bad. I don't think this would be a good ebike battery, but for the mower it might work.
I'll update you when the thing is ready and put into the mower (some modification on the mower side is needed as well). What do you think, will it work?

 

[harrisonpatm]

There are a lot of solderless batteries out there. It would be wrong to just outright say that all of them won't work. However, you'll always run into the same main 2 issues. First one you pointed out:

I'm worried a bit about the reliability of the spring loaded connections with the cells, but time will tell if they work or not.

I expect the working current to be around 20-25 amps, so it's like 3amp per cell.

3 amps per cell, sure. 3 amps per spring? Not so much. Spring steel is a poor conductor. If you have good calipers, you could measure one section of the spring to find its cross-sectional area, and cross reference it with the carbon steel column in this chart.

In case of problems i think these can be improved a little by tinning.

If you tin the spring, it'll solidify the spring, which won't be as springy anymore, meaning it won't make as much contact with the cell.

So that's issue #1, finding springs that can also carry decent current. Issue 2 is mechanical reliability of the cells in the enclosure; are they going to pop out mid use? You said lawn mower. That sounds like a lot of bumps and vibrations, and cells falling out of their slots a little.

Having said all that. It's still worth a try, as long as you understand the risks and consequences. Let us know what results you find.

 

[rafalg]

Hi, today i did some experiments with battery under load.
First, i came up with an idea of improving the electric contact at negative terminals - by soldering a copper ribbon (desoldering wick) to the springs and connecting them together. You can see the modification at the first pic.
Unfortunately, i can't really tell if it does anything or not.
Then i ran some load tests by connecting an electric motor. It's a fan from a broken battery vacuum cleaner and for sure it draws above 10 amps - i estimate it's around 15 amp. Can't give you a detailed number because my amp meter has 10A maximum range.
The tests were done in two versions - first with unmodified box (just plain steel springs), second one (thermo2.png) - with the copper strip attached. The motor ran around 2 minutes, and there are 3 thermal pics - first one at start, second - after about 30-40 sec, third - around 90 secs. I was looking for any hot spots on battery connections, but it appears there were none. The warmest part was the motor itself and wires connecting it to the battery.
So i think if the contacts are good for all the cells in the box, it should be able to supply the current i want.
However, one thing worries me a little - the contacts are not that reliable: sometimes a cell loses the connection even if seated firmly in place - then some wiggling around is needed for it to become connected again. Anyway, i will continue and hopefully finish the battery, in the meantime maybe i'll be able to perform some further testing with more current and more detailed measurements.
The thermal pictures suggest that in the first test (original springs) the negative terminal contacts were a little warmer than in the second test - so i think the ribbon might make some sense. And more testing, for sure.

 

[A-DamW]

Thermo cam definitely the right tool for testing.
Can't wait to see more!

 

[Guerney]

My worry is those springs will lose their springiness with heating and cooling cycles.

 

[fill]

Just thinking aloud.. could some copper mesh such as a braided sleeve?/solder sucker? or sold as 'stuff-it' soldered to the buss bar and wrapped around/over the springs aid ??..
though loose strands of copper could be a hazard, so perhaps a smear of silicone sealant to fix things in place?? again just thinking aloud so to speak..

 

[rafalg]

yep, this was my idea - you can see copper ribbon soldered to the springs in my second post.
I still need to do some testing of its reliability under heavier load.
And @Guerney, the plan is not to have any heating at springs

 

[Guerney]

And @Guerney, the plan is not to have any heating at springs

I reckon your modular battery will work ok in a static situation, but I'd be inclined to consider inserting and turning in as many springs as possible, for ebike use - double or triple up, in case of broken or weakened springs? On the other hand, perhaps applying too much pressure would be a bad idea... battery cases could deform or break, cell ends could bend, over time? Perhaps consider reinforcing the sides of the battery cases at either ends of the cells? I do hope this plan works out well long term on a bike. It'd be nice if we could make our own inexpensive, easy to assemble and repair, spot-welderless modular ebike battery packs.

https://www.ebay.co.uk/itm/276479921205

 

[gromike]

I do hope this plan works out well long term on a bike. It'd be nice if we could make our own inexpensive, easy to assemble and repair, spot-welderless modular ebike battery packs.

It's for a lawn mower. Still, I'm interested in how this works out. Though, personally, I have gone to spot welding batteries.

 

[docw009]

For added safety, consider buying a sheet of PVC washers, which are intended for placing on the positive end of the cells before they are shrink wrapped. Put them on the cells, and add a second layer of shrink wrap. Gives you a double layer. That will make it harder for vibration to buzz those metal bumps in the battery box past the insulation on the rims of the cells.

 

[rafalg]

Good day!
Did two more interesting experiments today, and some conclusions can be made now. As previously, testing with the fan motor from cordless vacuum cleaner.
Test 1: single battery box, ran the motor for about 10 minutes continuously and registered the temperature at 1 minute gaps. The motor drew around 11 - 11.5 amps. No heating was visible on the battery contacts, however some little warming up might be there. I think it can run pretty long with such currents.
And test no. 2 - in this case i connected two batteries in series, which gave around 8.2volts in total. This is twice the voltage motor is rated for - it ran pretty vigorously and the current was about 31 amps. Had to stop the experiment before 2 minutes have passed because the motor started to smoke and disintegrate a little.
One battery box was from the previous test, with the copper strip added at negative terminals. The other was unmodified, with bare steel springs. It can be seen on the thermal pictures that bare springs were heating up pretty fast, while the box with copper braid did not have such hot spots. So for sure there's some benefit in adding the copper strip.
Now i'll add it to every box and hopefully will be able to power the lawn mower. Not planning any other modifications at the moment.

 

[harrisonpatm]

Thanks for the test. My take away:

It can be seen on the thermal pictures that bare springs were heating up pretty fast, while the box with copper braid did not have such hot spots. So for sure there's some benefit in adding the copper strip.

It seems like you're using the spring not to carry current, but to press copper against the cells. Is this correct? If so, you would only have to watch for vibrations and movement shaking the cells free from their springs.

Also, have you seen this new thread?

My Bullit build using the Repairable and Refillable Gouach Battery

Hi guys! I'm one of the co-founders of Gouach, the Repairable and Refillable e-Bike Battery, we're designers/engineers from France who set up to design the best DIY e-bike battery ever! We'd love your feedback! Here's our custom Bullit which includes the battery, more details and videos coming...

endless-sphere.com

Looking forward to more info on that type of battery as well

 

[rafalg]

Yep i have read about the Gouach battery some time ago. But in my case the goal was not just to have a new battery, it was rather to verify if battery can be built from cheap powerbanks available from Aliexpress.
And my battery will be twice the size of Gouach. The cost so far is $30 for the boxes and $80 for the cells. Not counting wires and connectors.
At this point i think these powerbank enclosures are not bad - not quite ready to be used as solderless battery system, but with little modifications they could become nice and cheap battery building blocks. Just make them stackable/connectable like legos and then anyone could build a battery they need- or expand an existing one when needed.

 

[A-DamW]

Another interesting thing, with the built-in USB power bank interface, you could easily charge/discharge any parallel group for bulk balancing, etc.

 

[E-HP]

I like the concept and goal. Even if it fails the real world test, you could still reassemble each module with a spot welder, and still retain a lot of the benefits by virtue that it's modular, making repair and replacement of a single group much easier. You could even have a couple of spares ready to pop in and still get the lawn mowed. That might require redesigning how the modules plug in though, to make them easy to swap out, or charge up a weak module on the fly.

 

[rafalg]

Yep, some uses for the usb ports could be found. But for now the main problem to solve is reliability of connections inside the battery - for example in the pic from yesterday i only have 7 hot spots on spring connections, but there are 8 cells there. Doesn't mean one connection is so great it doesnt heat at all, rather that the cell is not connected at all.

 

[Guerney]

Perhaps try wiping off any oxide from the cell ends? I used to throw old AA NiMH cells away when they wouldn't charge, until I discovered wiping with isopropyl alcohol solved the problem most of the time. They're cheap from the "Poundshop", where everything costs £1. Or used to. £1 a pair.

 

[Chalo]

So that's issue #1, finding springs that can also carry decent current.

High power brushed motors use copper braid or very fine stranded wire in parallel with the relatively weak springs that push the brushes against the commutator. It would be a mega nuisance to put together 80 of these, but that would circumvent the issue.

Also, beryllium copper springs might be an effective workaround. But at best they'd cost quite a bit more for qty 8 than the price of the power bank enclosure.

 

[harrisonpatm]

Also, beryllium copper springs might be an effective workaround. But at best they'd cost quite a bit more for qty 8 than the price of the power bank enclosure.

I started pricing those for a spring-based powerwall project I'm working on. It's not worth it unless you're ordering 10,000

 

[lnanek]

The springs are there for consumers to easily swap out cells without tools. Feels like it would be easy to rip them out, drill a hole where they were, pop in a rivnut, and just screw a bolt in tight against each cell - maybe with a lock washer if you really need some amount of spring.

There have been successfully batteries with bolt terminals like that featured on these forums before:

The Barncat Battery- No weld/no solder

I designed and built these packs a year ago. After a couple thousand miles of testing and considerable internal debate, as a service to the DIY community and as a social experiment, I'm releasing the design publicly. You use my design and any information contained in this discussion thread...

endless-sphere.com

That Gouach battery that's popular here lately also mostly bolts together with just a little peninsula of PCB with a tin ball for each cell to add a tiny bit of spring:

Gouach Battery Framework

Gouach is a battery design framework enabling fast design and customisation of eco-designed & connected lithium-ion batteries.

docs.gouach.com

 

[Guerney]

Also, beryllium copper springs might be an effective workaround. But at best they'd cost quite a bit more for qty 8 than the price of the power bank enclosure.

I haven't seen any that are conical. I guess the OP could make his own using beryllium copper wire - available on Amazon in various diameters:

https://www.amazon.co.uk/C17200-Beryllium-Copper-Wire-diameter/dp/B0B4ZJYG1S/

If beryllium copper crinkle washers are springy enough, maybe somehow secure small ones onto shims, to replace the steel springs?

 

[rafalg]

Hi, I don't think i'll be researching beryllium springs or more exotic matter here - after all, the goal was to use ready-made, cheap battery boxes for solderless build with minimum modification. So far it's not bad - i just modified every box by adding the copper strip mentioned earlier, which seems to really help with improving current carrying capability. Cleaned everything with alcohol and put cells in. Then all the boxes went into lawn mower - luckily they fit in the original battery compartment without destroying too much of it.
Did some test today, the mower runs nicely - when idle it consumes around 5-6 amps, but when cutting grass the current goes up to 15-25 amps, depending on how much grass it has taken. On average i think the current is around 12-13 amps. I suppose when bogged down on too much grass it could go above 25amps, but then there's an overcurrent switch in the motor controller that would stop it.
However, the battery is not finished yet - i just connected 10 boxes with Wago connectors, but there's no BMS yet. Looks quite clean so far but have to do something not to turn it into spaghetti when BMS is added (apart from that there doesn't seem to be any space left for the BMS board).
When i have the BMS connected I'll do some more extensive mowing test with temperature measurement. No pictures today but i think everyone has seen a lawn mower with wires.

 

[mindgames11]

Hi, today i did some experiments with battery under load.
First, i came up with an idea of improving the electric contact at negative terminals - by soldering a copper ribbon (desoldering wick) to the springs and connecting them together. You can see the modification at the first pic.
Unfortunately, i can't really tell if it does anything or not.
Then i ran some load tests by connecting an electric motor. It's a fan from a broken battery vacuum cleaner and for sure it draws above 10 amps - i estimate it's around 15 amp. Can't give you a detailed number because my amp meter has 10A maximum range.
The tests were done in two versions - first with unmodified box (just plain steel springs), second one (thermo2.png) - with the copper strip attached. The motor ran around 2 minutes, and there are 3 thermal pics - first one at start, second - after about 30-40 sec, third - around 90 secs. I was looking for any hot spots on battery connections, but it appears there were none. The warmest part was the motor itself and wires connecting it to the battery.
So i think if the contacts are good for all the cells in the box, it should be able to supply the current i want.
However, one thing worries me a little - the contacts are not that reliable: sometimes a cell loses the connection even if seated firmly in place - then some wiggling around is needed for it to become connected again. Anyway, i will continue and hopefully finish the battery, in the meantime maybe i'll be able to perform some further testing with more current and more detailed measurements.
The thermal pictures suggest that in the first test (original springs) the negative terminal contacts were a little warmer than in the second test - so i think the ribbon might make some sense. And more testing, for sure.

I bought some copper strip for these kinds of experiments and similar- I'd recommend trying that as a bus bar as opposed to the soldering wick. cheap too. less than five bucks

 

[rafalg]

i'll just add a short update
The mower works just fine on this battery. But I haven't added the BMS yet, it's waiting for the right time - so charging is a bit handicapped now.
The built-in charge level indicators in each box turned out to be useless - they don't show actual voltage of cells inside the box. They do 'coulomb counting' instead by monitoring how much energy passes each way thru USB ports. But, since i'm not using USB for charge/discharge, the power levels shown by the boxes are all incorrect/random.
Regarding the choice between soldering wick and copper strip - I dont use metal strips for soldering because it's too easy to short circuit a cell when soldering a metal strip to the positive terminal. It happened to me with nickel strip - too much solder, instead of sticking to the strip, went down between positive and negative terminal and connected them. Boom.
No such problem with braid/wick - the solder stays in the wick and doesn't flow anywhere else. Here i didn't solder any cells, but still i only use wick.

 

[G8trwood]

Put copper T stud trough spring with T pushing against terminal. Other end soldered to wires? Maybe copper braid pulled through spring would do the same.