The case for SOLDERING 18650 cells.

[Robocog]

I soldered my 18650s. Heres why. Better contact, over a larger area. Thus, less sag, heat and power.

The arguments against, all stem around hamfisted technique.

I sanded the cells, and used nickel strip. I don't think I need to double the strip on series connects, due to this, anyone disagree, pipe up.

I used good solder, paste flux and tinned the cell ends. I shock cooled the tinning, to quell heat.

I heated the nickel strip till the solder flowed, which was about half a second, mashed the joint, to maximize METAL contact.

Then, when it solidified, I blew on it by mouth. This gave a cell end, that was not as hot as a battery in heavy use.

All nickel connects are short.

The parallel joints are heavy wire, soldered to nickel strip soldered tabs.

I personally don't see why this method is not superior, with the possible exception of time and effort.


The alternative is, spot welding, which is usually six, points of contact, comprising all of say .03 thousandths. I say this leads to heat. Sag and less power, heating of motor etc.

Also, I used a linear style of connect. With no mixing of cells. As they do with parallel AND series connects together. It was a lot more prone to mistakes in MY case, as I had a nine cell config per group. Thus, to make 13 was sharing across separators.


BUT, I think the custom config form fitting per bike style shape is the future. Bricks are for kids. Not only that, I will create a rakish front scoop, that will cool the cells, and my 48v to 12 converter.

Also, putting my batt right next to the motor makes SHORT parallel runs. Does anyone think I should use dual xt60 connects for power hookup? Also, what is the best power switch? NONE? Disconnect batt entirely? An external FUSE acting as switch?

[moderator edit to fix title]

 

[Hillhater]

Hard to comment on anything conclusively as you dont mention the pack size , configuration, voltage , max amps, cell model, strip width or thickness, etc etc

 

[Robocog]

48v 14ah 13s 40watt iron, 6.35mm x .25mm Nickle Strip Tape , 3inch around pvc tube, hot glued as nine cells per segment. Panasonic 18650 GA 3500 mAh . bought in bulk.

1000watt bbshd

 

[Overclocker]

sanyo GA's already get warm at 7A continuous. so if you think the spot weld is the limiting factor then you should be using 30A-rated cells (HE4, HG6, etc). coz 7A is perfectly fine w/ spot-welded single-layer 7mm x 0.15mm nickel

here i used Samsung 25R. good solid soldered connections to avoid the spot weld bottleneck. i used my 1-second soldering technique. it's on youtube just search for it

 

[Robocog]

Here is the blurb Luna states.

•GA Sanyo cells, guaranteed Authentic
•3400 mAh huge internal capacity
•Ideal cell for most ebike builds
•10 amp discharge...double the amperage of the NCRB 3400

This is the most exciting 18650 cell new for 2016. Sanyo was bought out by Panasonic so its known as both the Sanyo 3500 GA and the Panasonic 3500 GA. This cell offers both high capacity and high amperage for the ultimate in power dense performance...almost ideal for ebikes and we feel the best cell available for most applications 40 amps and under.

 

[999zip999]

10a cell is a good cell . There are 25 and 30amp cells. It all depends on need and price. Which is also getting better.
You can drill or punch a hole in the strip then pretin and heat the heat transfer hole.

 

[999zip999]

What motor and controller are you running. As how hard are you going run the pack ?

 

[Robocog]

I will use it to commute, and light dirt. I don't hotrod as it busts stock drivetrain. I will USUALLY go max 25mph, with SCH marathon+ 29er 32s.

I will USUALLY always pedal along. For long trips, I will install the aux pack, of 7ah, to add parallel. I will NEVER take off under full power, with no pedaling. I will use the gears.

So, the average amp draw will likely be under 20.

Motor is BBSHD, controller is stock, but programmed for 1000 watt. Mostly flat terrain. No steep dirt hills etc.

So, 4p will net a true 40amp continuous amp draw. When 6p, it is 60 amp.

 

[Hillhater]

Whilst the GA cell can give 10A discharge, that is not without voltage sag levels higher than some other cells. (Eg lipo)
And during the later part of the capacity drain, it will be operating well below 3.0 v , so the wattage will be lower even at 10amps.

 

[redilast]

There are several problems with soldering. The biggest issue is heat obviously, but it is especially bad when welding cells with a PTC device which acts as a polymer resettable fuse to shut the cell down in an overcurrent or short circuit situation. The soldering process can cause the PTC to change phases and activate and in some cases even though when it cools down the cell will work again, the cell will have increased internal resistance lasting for possibly months or longer.

Now as far as I know, higher current cells (generally those above 10A) do not have a PTC, and so therefore you don't run into that kind of a problem, but there are other damages o the cell from just the short heat exposure. Laptop grade 18650 cells almost certainly have a PTC device which can be damaged by soldiering.

The other big problem is if the cells were to short out and they are not fused, well that soldier is going to get hot enough to flow and you'd have molten solder dripping down the pack. I rebuilt a powertool battery a while ago where I re-used the original fuse. I spotwelded to the end of the fuse then added some solder just to keep the resistance as low as possible in that area. Well somehow the customer managed to blow the fuse (30-40A blow point) and there was molten solder in the case of the battery. Not good. But the welds to the fuse held up fine in that short high temperature exposure where the fuse blew.

 

[liveforphysics]

Solder conducts electricity like crap. Nickel conducts electricity like crap.

If you really care about total pack impedance, using solder and nickel is how to ensure high resistance and add some extent of damage to the cells electrolyte (which is likely to be very minimal damage if it's a fast process but still damage.)

Likewise, use caution sanding modern cell cans at all with and grit paper, part of getting 3.5Ah into that tiny can is using very thin can walls (all modern 18650 can walls are a little spooky thin.)

 

[Robocog]

Good point, but, a little HAND sanding, JUST to roughen, is NILL. And, the THIn nickel, IS A BOON, as it will not transfer much REAL heat. My mouth blow would stop heat path cold. My mashing the parts together, for the solidification, will guarantee metal to metal contact. It will also START the taking away of heat.

My starting the heating on the nickel, and then mashing also sequesters the heat till it is mashed, in large part. The solder, was VERY thin, as the nickel is FLAT. Thus, it did not ball up, which would transfer heat. Directly after my procedure, I could apply my finger to the solder joint, and not only not burn myself, but it was lukewarm.

I get your point about fusing. I MUST, as any REAL heat would desolder the joints.

I also think using the batts gingerly, with no REAL hotrodding, is a good practice. I also think that my batt case, open all around, and with a scoop forward, cooling the batts, will help.

Of course, this means no RAIn riding. But, I will not do that anyway.

 

[liveforphysics]

I hope it all works great for you. It sounds like good potential for a cold solder joint and a recipe for a not so good under high current pack to me vs something like the Snath no-weld design, but for a low power application you may have good luck with it.

 

[Hillhater]

......use caution sanding modern cell cans at all with and grit paper, part of getting 3.5Ah into that tiny can is using very thin can walls (all modern 18650 can walls are a little spooky thin.)

..yes, the can walls are only a few thousands of an inch thick, but the base of the can is at least 3 times thicker (due to the manufacturing process) and of course the top end cap is not part of the main can it is much thicker also.
A good flux is much more effective than sanding anyway !

 

[Robocog]

I did try it as the nickel and batt cells were new, and I thought ir would be easy, as they said it would be. BUT, the flux and scratches worked a treat.

By the way, four days, and they all check out FULL and equal. 3.7.

 

[liveforphysics]

I did try it as the nickel and batt cells were new, and I thought ir would be easy, as they said it would be. BUT, the flux and scratches worked a treat.

By the way, four days, and they all check out FULL and equal. 3.7.

That's great to see.

Watching self discharge is one of the unique situations where getting to use a 6.5-7.5 digit DMM comes in handy. If you have access to a near by EE school/lab that may have a high precision meter, leave your battery in a styrofoam cooler to let it reach a uniform temp equilibrium between cell banks for a day or two in the same room with the meter. Then setup your phone to take video of the meter screen, open the box and take a video of the meter's screen for each cell-tap voltage for 1 minute. Make audio markers in your video where you call out each cell number as you're videoing it.

 

[Hillhater]

here i used Samsung 25R. good solid soldered connections to avoid the spot weld bottleneck. i used my 1-second soldering technique. it's on youtube just search for it

neat work. and im sure you minimized the heat transfer...
but the one thing all cell manufacturers/pack assemblers agree on is...keep welds or solder away from the center of the cell base (-ve end)
That is the most heat sensitive part of the cell.

 

[999zip999]

A light swipe of flux. Just a wisp. Just tinned very hot big fat tip iron. A well damded sponge. With thre e quick hands. Or so

 

[Robocog]

yup or four

 

[Overclocker]

neat work. and im sure you minimized the heat transfer...
but the one thing all cell manufacturers/pack assemblers agree on is...keep welds or solder away from the center of the cell base (-ve end)
That is the most heat sensitive part of the cell.

this is the NEG end of an old sanyo. interesting to see that green insulating disc so i don't think this cell would be harmed by soldering especially if done quickly

i can see why they recommend not to weld to the center because that's where they attach the negative to the metal can using ONE spot weld in this case. so if a pack builder were to weld at exactly the same spot then it might affect the integrity of the inside weld. however i don't see any problem w/ soldering affecting the internal weld

 

[Robocog]

VERY good to hear. I cant imagine I did anything but heat that disk, or like structure. Also, When I mashed it, the solder flattened out to SUPERthin. Very broad area. So, metal to metal contact. No heat pooling. NOT likely to increase internal resistance much if at all.

BUT, I must protect against overamping, to ELIMINATE the possibility of solder FUKOshimaING.

 

[Overclocker]

BTW not all cells may have the insulator disc. Will open up some more cells tomorrow...