Yet another 18650 battery build.

llile

1 kW
Joined
Dec 18, 2010
Messages
457
Well, I'm going down the adventure road of building a battery from scratch. I've built everything else on an E-bike from scratch almost, so why not?

Batteries - I found a deal on some panasonic INR1865015L's. They aren't the highest capacity 18650's, but they'll do. 1500maH. My goal is to add 15-20 amp-hours to my fading existing battery. It is a different chemistry, so this will be a "spare" that I'll kick in by unplugging the old one and plugging in the new one when the main battery falters. I'll set the Cycle-Analyst low voltage cutout up where both batteries are well protected, with plenty of reserve capacity so I'm not reducing life. If the old 40AH nee' 36 AH LiFEPO4 battery ever bites the dust, I'll use all 18650's on the bike and shed some weight. Right now, the old battery will go till it drops.

Charging and discharging an 18650 bang to bang it has a life of about 500 cycles. If one charges say 90% to 20% the life is increased dramatically, possibly thousands of cycles. This corresponds to a charging voltage of maybe 4.0 volts per cell and a low voltage cutout of maybe 3.45 volts per cell. Depends a lot on the load. In a 72 volt battery this would mean 20 cells in series (up to 84V fully charged to 100%). But wait! I've got another E-bike that's 36V. And my buddy has a 36V bike. So I'll make two 36V batteries and put them in series.

I'm looking for about 15 amp-hours derated capacity (70% of full capacity). That's about 19-20 amp-hours full capacity. These are 1500mAh batts, so 13 in parallel would be 19.5 amp hours derated 70% = 13.65 amp-hours. A little light, however when I sorted my battery order, I've got 263 good ones, so that's where I'll start.
 
Charger. Currently I'm using this charger https://www.instructables.com/id/Electric-Bike-Charger-From-Mean-Well-Power-Supply/, it's made from three Mean Well Power supplies. It is set up for a "72V" LiFePO4 to charge at 82V.

Fast forward a couple of years, and most new mean wells will take 240 or 120V without a switch. They simply don't care what you plug them into. I'd love to be able to catch a little charge on some of my longer rides, and I drift by several electric car charging stations on the way home, wistfulling looking as cars charge up cheap, whereas I can't find a plug anywhere that I can beg. What if I put a J1772 interface on the E-bike? One step at a time, I'll set this charger up so it can accept 220V or 120V either way.

Mean well makes a lot of power supplies, some of them pretty expensive. I found some USP-225-24 CC-CV open frame power supplies on ebay for $12 apiece and snapped them up. Datasheet here http://www.mouser.com/ds/2/260/USP-225-SPEC-806469.pdf These are 9.4 amp power supplies, 24V nominal, adjustable from [strike]21.6 to 26.4V [/strike] 18.6 to 26.4 DC (measured). Two of them could be set up to achieve 90% of full charge voltage or about 39V according to the battery curve.

These are open frame, they'll need a box, and a small computer fan. I see that one can buy a ball bearing 24V computer fan for about $12, I'll just hook the fan up to one of the power supplies.

I'll provide two charging outlets each 36V nominal. This way, I can charge my two 36V 18650 batteries, and I have a set of jacks that can stack two 36V chargers to charge my 72V LIFePO4 battery. Or the spare for my other ebike, or my buddies' bike.

Four such supplies would be about all a 120V 20 amp circuit can handle. I looked at hotter power supplies, but i'd be stuck with 220V charging all the time if I went that route. Total will be about 900W output, and 13.23 amps input.

The set should charge my old battery, except for balancing, in 4 hours and the 18650 battery in 1.3 hours. The 18650 could theoretically charge faster, but I'd be blowing breakers.
 
I've decided to solder the pack, instead of spot weld or using Vruzends. Vruzends are great, but I need a little higher current capability, currently they are at 3 amps and I'll hit that. Spot welding is great, but I don't have a spot welder. After studying the famous Spot Welder vs. Soldering wars https://endless-sphere.com/forums/viewtopic.php?t=88110I concluded that I have really good soldering equipment, a lot of skill with soldering (I've been doing it regularly for 50 years) and I can solder the ends of a battery as fast as you can count one-two-three. That's not, according to better minds than mine, going to heat up your cell significantly or damage it. The ends of an 18650 don't actually touch the roll of material inside, and things cool off so fast I can touch the solder as soon as it is hardened.

I'm going to use the fuse technique, with 24 gauge tinned copper wire. My experiments showed this would burn open in the count of three when shorted across an 18650 cell. A 22 Gauge wire would not burn out in 20 seconds. 24 gauge is theoretically a 30 amp fuse, per this chart, but I don't have anything that can test that kind of current accurately. https://www.powerstream.com/wire-fusing-currents.htm

This is a picture http://www.electricbike.com/wp-content/uploads/2017/07/BatteryFuse1.png of what the pack might look like when I'm done.
 
ORdered some Battery pack spacers off ebay, looking for a US shipper so it won't take weeks from China.
https://www.ebay.com/itm/253258381485 They hold 4X5 cells, so the basic dimensions of the pack will involve modules 100X80MM. That's a little over 4" wide. Eventually I need to figure out a case, and it might be I'll just bend up an aluminum box that's about 4X4 and however long.

I plan to cool the batteries, and have them easily removable. Batteries should also be kept from extreme cold. I should really take my batteries inside in the winter when I'm not riding much, they'd last longer, also store them at 50% charge. However I never know when the last day I ride really is, at some point I just reallise I haven't biked in a month. Maybe I shoudl just take them inside when it gets to be 20F outside the first time, riding season is about done then.

Ordered 26 gauge tinned copper wire from Jameco https://www.jameco.com/z/3819-100-24-AWG-Solid-Tinned-Copper-Bus-Bar-Wire-100-Feet_2098478.html
I found some 26 gauge flat copper sheet that I'll use for busbars. An 18mm wide piece (0.015" thickness) has the conductor area just between a #10 and #8 wire, about 0.0106 square inches of copper area. At 35 amps, the most I expect on a sustained basis, this bar should be about 1.6 ohm/1000 ft or .0016 ohm for a 12" stick. At 35 amps, that's a power dissipation I^2*R of 1.96 watts, which would generate a lot of heat. Given this, I will try to avoid having the full 35 amps through a single piece of busbar, maybe switching to #6 solid copper wire when I need to carry more than hallf that current.

I ordered a kit of Tinnit and I will try tin plating any bare copper parts for corrosion. Also some real Kester 44 solder, instead of the generic stuff laying about here, because Kester 44 claims it is good for soldering to nickel and other metals. I believe the flat battery ends are nickel plated.

Not sure how to do a cooling fan. One way would be to look for a 48V nominal rated computer fan, and add a Snap-disc thermostat that closes at 110F https://www.mouser.com/ProductDetai...=/ha2pyFaduhG1ExVzibM2dTBJ3PSxS2tY8vNQZqGup8= to control it. That would only kick in if the battery was generating a lot of heat, not when it was idle, since it never gets to 110F here. I don't know how I'll keep water out of a battery case that's got a fan sticking out of it, but I'll worry about that later, or just put it in an aluminum box and forget about it. The aluminum box should transfer a lot of heat. Maybe a fan is just overthinking this.
 
Here are a couple of possible layouts. I'm going to build two 36V batteries, to be used normally in series. I could accomodate the long skinny version, but the short wide version might be more practical.

I need 15 cells in parallel, but the spacers work out easily to blocks of 16. I can leave one cell out, and use the open space to jump a wire down to catch the negative of the next bank from the positive of the first bank. This allows me to center-tap the busbars, resulting in the minimum current flowing in each busbar. Heavy currents only flow in the jumper wire between banks of cells, which will be #6 solid copper.


View attachment 18650.pdf
 
I don't have a discharge curve for these batteries exactly, but there is one for a very similar battery from the same mfr which should be close enough.

If I read this right, at the lowest measured current, 10% less than full charge is about 4.0 V per cell, and 20% reserve is about 3.45 volts per cell at 2 amps. Setting a low voltage cutout at 20*3.45 = 69 volts with the cycle analyst should make things work just about right.

Here are some similar curves
 
Here are two of the mean well power supplies adjusted to achieve 40V in series. Power and low voltage wires are hard wired. I'll probably add a 30A automotive fuse just in case, and they'll be connected to some three pin mic jack style terminals to interface with the batteries. Ping is a fan of these connectors and so I've just standardized on them.

IMG_20180309_202434742.resized.jpg
 
I didn't make the plots, they came from a website that posts a lot of battery reviews.

The plots likely came from a "Battery Analyzer" and the cream of the crop is the West Mountain Radio CBA series.

Mine is a cheap consumer grade battery tester, that reads out voltage, charge mAh, and can put the batterty through a discharge-charge cycle to tell you it's full capacity. There are a number of these, I have this one, there are a few better models out there that can do the tests faster. There's another thread here where I discuss that tester and another guy recommended a better one, I'll look it up here in a bit. This tester is slow and can do only four at a time. I sampled all the batteries I got, and tested any that had voltages lower than 3.8 volts. Any with under 2.5 volts were thrown out. The seller said they'd sat around for a while so some had self-discharged, so they were returned. All other batteries have tested good for voltage, charge, mass (a quick way to spot fakes is to weigh them!) etc.

Some kind of quick and dirty battery analyzer like the IQ 338 is a must, and I wouldn't mind having 3-4 of them if I do this kind of battery build again.
 
I was meaning what program did you use to produce the battery layout format neatly, then convert to pdf? I am in the process of charging up the lead to start spot welding mine shortly 8)
 
Cross posted from a thread "Batteries and Heat"
Here is an interesting scholarly paper on the subject of Li-Ion capacity degradation and temperature.

Here is a link to the paper Effect of Temperature on the Aging rate of Li Ion Battery Operating above Room Temperature

And the most interesting graph - Charge (Q) curves for a number of cycles vs temperature

It seems the short answer is, under 35C LI-Ion batteries do not suffer appreciably, by 55C the capacity loss is double the capacity loss at 35C, and triple that at 25C.

So my question on my build blog here is - should I bother trying to cool my batteries? I am planning on using them at about 2C discharge rate. It'll be 35C Temperature here on a hot day. I certainly want to design them to achieve a long lifetime, since I've now sunk money into them.
Should I simply wrap them in an aluminum case, with air circulation space, and hope that cools them off sufficiently? That's the easiest thing to do. Does anyone have a good estimate as to how hot your 18650's get at a given ambient temperature at some given C rate?
 
Here is a blog where a guy is DIY'ing an 18650 battery for a motorcycle, and sees about 40C temperature in a test at 2C discharge. He says a fan over it reduced the temperature 10C.

Interesting. This seems to argue that one should consider cooling for a 2C pack ... even if it is passive cooling it would beat stuffing your battery in a foam box (My last build was stuffed in a foam box.)
 
On my charger-tester, which takes 4 batteries and can charge at 500ma or 1000 ma, I am seeing 31.9C measured on the surface of the battery in free air with a good quality K thermocouple instrument at 500ma Charge, and 35C at 1000mA charge current, both in a 26C room. So that's almost 10C over ambient, with no restrictions around the battery, but also up next to a charger which may make its own heat. I've almost got a battery module completed, once the BMS is hooked up and tested I'll cook up a resistive load bank and do a few temperature tests on it. Might be intyeresting. If we're seeing 10C rise above ambient, and ambient is already 35C here in the summer, then the battery could be at 45C, which is beginning to cook it. Measurements will tell. Oh, and pictures! Next post.
 
You are addressing issues related to my project. Thanks. For an electric go kart project, my plan is to use the following:
Brand new Samsung INR18650-25R cells (20Amps maximum continuous discharge)
Battery packs: Using the Samsung cells, I plan to have two separate 20S5P packs connected in parallel
Motor: 72V Motenergy Brushless Motor (ME0201014201). Continuous current - approximately 50 - 60 amps at 72V (Max current 300 amps)
Charge Controller: Sabvoton (svmc72150) (Max current 300 amps)

I have been practicing using a spot welder to connect thin fuse wire (36 swg which blows at about 5 amps) to the cells instead of soldering. I would think that spot welding the fuses improves safety (since thermal runaway would be less likely) but I also want to avoid adding too much resistance into the circuit. Any opinions or suggestions would be greatly appreciated.

Gary
 
So much progress, must catch this thread up with picvs soon.

Garolittle, you are right, spot welding is much better. I'm soldering cuz no spot welder. I also intend that this isn't my first build - get my chops on cheap batteries then upgrade with newer better faster spot-weldier ones. :lol:

That is *awfully small* wire and I am skeptical you can successfully and repeatedly spot weld it without making the weld into a tinier fuse. Also, is it not copper? Will that even weld properly to nickel steel? Or maybe I misread and you are using nickel plated steel wire. You know for sure, you've been practicing. What is your experience? I recommend using a thicker wire, I found 26 gauge to work pretty well, low resistance, yet the cell can still blow it if theres a bad problem. These fuses will only activate if something catastrohpic happens like cell rupture.
 
Every good battery needs a charger. As you may have noticed, I'm building a new charger at the same time.

Although I am using a 72V battery on my bike now, I'm setting up the charger and new battery to use 36V section. Each section is two 10AH 36V modules in parallel. This allows me to use the battery with my other bike (36V) or on my friend's bike (36V) or to charge my friend's bike. I have my 72V battery already set up so that I can use a 72V, or two 36V in series, chargers. I know, this is extra complexity, but really not that much. I like modular design.

So this power supply, as I posted earlier, uses four mean-well power supplies, open frame USP-225-24's. Two of these in series can be adjusted to produce 41 volts, just right for topping off a 12S LiFEPO4 or a 10S 18650. This works out handy, as I'll be running both chemistries for a while until the LiFEPO4's get tired. I wont run two chemistries in parallel, I'll switch between them using Anderson plugs. The old LiFEPO4's have lost 10% capacity already, which is why I'm building a new battery in the first place.

USP-225-24's are rated 225 watts each, have a wide input range of 90 to 305 VAC 50 or 60 hz. Later I may set these up to use 220, and hack things so I can charge from a car charging station, which could get me out of a jam. The power supplies are 9.4 amp max.

One problem with Mean-wells is inrush current. These supplies can pull 15 amps inrush current each, four could pull 60A briefly. Will that blow a 20A breaker? Don't want to find out when I'm out begging for a charge. I've added a switch that puts two 5 ohm 5 watt resistors in series witht he Mean Well input leads, then can be switched to connect them straight to the power lines. I can switch it to charge, which brings the output up to proper voltage in a couple of seconds, then switch to straight through for actual use. The switch on the front is not on-off, but on-charge.

I got some 24V ball bearing fans at Jameco, ball bearing being long-lasting and quiet, 24V to match my power supplies. They run fine on reduced voltage, a little less airflow. These are hooked up to opne of the supplies, running whenever the unit is plugged in.

I used a standard computer three-pin AC socket, like is used in the back of your computer, to allow quick changing from 120 to 240 if I choose to do so in the future. A little ammeter/voltmeter/power meter measures one of the 36V outputs (I was too cheap to buy two, one tells me what I need to know) and used some gland fittings as strain reliefs for the wire. Two three-pin mic jacks are the output cable, compatible with many commercial batteries, all my bikes are standardized on these as charging plugs.

I considered briefly building it watertight, but balked at the cost of Mean Well waterproof power supplies. This open frame supply can run without fans, but not at full power. The unit is not waterproof, a consideration when charging on the road, but usually a solvable problem.

Here is the voltmeter - 0-100V, 20A, internal shunt. Although it states that it saved energy measurement between uses, mine doesn't, which is just fine as I want to know how much energy I put in for each charge. Energy Meter

Next post a few pics.
 
IMG_20180309_202434742.jpg

Two of four power supplies, being adjusted.

IMG_20180413_190508821.jpg

Cases under construction

IMG_20180414_214054875.jpg

Getting ready to close up the can. I riveted this aluminum sheetmetal case together. Had to take the rivets out one time to readjust things, they drill out pretty cleanly. I was afraid to use sharp sheetmetal screws, lest a sharp end hit a wire inside. I was quite careful to place rivets or bolts where they would not hit any live parts, and vacummed the whole thing out with a shop vac several times before closing it up. Would not do to have a chip of aluminum in the works. Power supplies are secured with small bolts.

IMG_20180414_214606034.jpg

Fan endcap riveted in place.

IMG_20180419_175137520.jpg

Completed power supply. Almost 800 watts.
 
Finally a few battery pics:

IMG_20180310_181104034.jpg

Assembling 18650's in plastic battery holder.
I started this project before Vruzends V2 came out. My loss! If I was doing it over again, I'd be using Vruzend V2.

IMG_20180310_182312786.jpg

A little hot glue to hold things together. I've read Hot glue can fail after vibration, but I've never experienced that. Once this batt is in a case, it won't matter, plenty of other material holding things together.

IMG_20180310_183657350.jpg

Copper strips snipped out, ready to tin.
Eventually I started folding these over so they were double thickness and half the width - things fit better.

IMG_20180310_183947535.resized.jpg

Tinnit - a chemical that deposits corrosion-fighting tin on copper. Easy to use, just add hot water and agitate in a plastic tray.

IMG_20180310_190525991.resized.jpg

Busbars ready to tin,

IMG_20180310_192143663.resized.jpg

Ten minutes later - voila! Tinned copper busbars.
 
IMG_20180310_200357057.resized.jpg

Busbars ready to go. Like I said folding these copper strips in half lengthwise worked much better.
I hot glued them onto the battery holders, to keep everything in place during assembly.

IMG_20180310_203733584.jpg

One layer of cells soldered. I'm using a temperature controlled Hakko 907 iron with the largest tip. I melt a large glob of solder on the tip, then wait a few seconds for the solder to heat up beyond the melting point. I apply the tip to the battery/wire, adding more solder. The whole soldering process takes about a count of three. Another three seconds and I can touch the solder joint - it cools fast.

IMG_20180310_205559073.jpg

Busbar
 
IMG_20180310_182139829.jpg

BMS. This is a 20A BMS. There will be two in parallel, giving me 40A capacity. Note the connector for the voltage sense wires.

It is absolutely critical to pull this connector before soldering up sense wires. If you do not, and you even brush a conductor along the wrong voltage, you can blow your BMS board. Don't ask me how I know this. :mrgreen:

IMG_20180317_201505655.jpg

Sense and power wires hooked up. I used #12 Primary automotive wire for power conductors, flexible and durable.

IMG_20180317_221821163-1.jpg

Insulating material over the conductors and bus bars. I'll go into what this is next post.
 
Here is a great video of a guy spot welding fuse wires onto 18650's

Spot Welding Fuse Wires

He's using this 60A spot welder

He's using 35 AWG tinned copper wire, 5 amp fuse, which IMHO is too light. 5 amps could easily blow as a nuisance.
Set at 4 pulses, 30 amps current.

He shows that a proper spotweld doesn't fail at the weld, but a spotweld with too much current weakens the wire. Pretty cool tests.

As much effort as this soldering stuff is, maybe I should have gone all the way and built a spot welder instead. Well, we've come this far soldering, bash on!
 
llile said:
... If I was to start from scratch I'd be wanting to use Sketchup.

Thank you for your advice - got it done in Sketchup. Pretty blown away at how user friendly it was, its been a few years since I've touched a CAD program and that was definitely easier to use than the old versions of AutoCAD.

I've lamented for sometime and I think this layout is about as good as it gets for 13S in a 21 x 5 pack? what do you think?

5qg3G7.jpg
 
That's one way to do it!

I've got to learn sketchup one day - it's too easy to break out the AutoCad clone. 1980's technology.

I'm wondering how the busbars will go on your layout. Are you soldering or welding? Fuse wires or nickel sheets? Man, after spending all this effort soldering, I'm becoming more a fan of welding! There's a lot of tech involved in building a battery, and not a whole lot more effort (for me at least) to build a damn welder out of a microwave transformer and an arduino. And don't forget Vruzends.

One could also just write a check for $164 and have a really nice welder delivered three day ground.
 
Back
Top