This third installment is mostly about the special considerations that are needed when building a high-amp battery pack, plus…I will list some of the products and tools that are sometimes used to build a pack so, let’s get started.
The spot-welders used by industry to make ebike battery packs are large and expensive, and it is unrealistic to even consider them for the home garage DIY builder. This is important because, there are many countries where it is still very difficult to buy a complete battery pack, since the most well-known pack sellers do not ship to many countries.
The rules for shipping a complete battery pack are complex and they are getting worse every year. Even if you do manage to get a pack builder to ship one to you, it is expensive and risky. Plus, even the most successful pack manufacturers do not make packs in every possible shape and size.
However, it is still actually fairly easy to buy the individual components to build a pack, no matter where you live (as listed in part-2, click here), and this means that resourceful garage DIY builders will continue to assemble their own packs. This is the main reason I wrote this series on pack-building.
The methods and materials that are used have been steadily improving, but… I hope to document some of the common “best practices” as they are currently known, in order to help new enthusiasts avoid the common expensive mistakes on their first build.
I simply cannot recommend any product from this brand (I apologize to Sunkko, but…I have to be honest). I only mention them at all because they are so well known, and they will show up immediately in any Google-search about “18650 spot welder”.
The biggest take-away I found from researching them is that…the 220V version is slightly more reliable than the 110V version (and the 110V version has a horrible record…Don’t trust me, research this yourself).
Many places like the EU and Australia are fortunate to already have 220V AC as the default system in all homes. Homes in the USA are typically 110V, but be aware it is not difficult to get your garage wired up for 220V for an electric clothes-dryer, and also many air-conditioning units in the USA also have a socket that is wired for 220V.
Since the 110V version and the 220V both draw roughly the same amount of watts, the 110V version draws twice the AMPS from the power socket of your home, which often trips the breaker (15A is common in the USA, 110V X 15A = 1650W). Even if you do get the 220V version (and either find a 220V socket somewhere in your home, or have one added), the electronics for both are a known weak spot. If you already have a Sunkko, here is a website with info on how to upgrade it, and fix it when it breaks (click here).
Also, Here is a video (click here) where a builder opens up a $240 (plus shipping) Sunkko 709AD model, and shows how to upgrade the guts so it will perform in an adequate manner (600-ish amps).
The Malectrics Spot Welder
This “pocket spot welder” from Germany seems to be working very well, and it has many satisfied customers. The programmable brain-chip is the popular “Arduino Nano”. The welding pulse time can be adjusted from 1-milli-second, up to 500ms.
Somewhere along the line, someone figured out that if you had a large battery that was capable of putting-out a lot of amps, you could add some type of on/off switch and timer to make a spot-welder out of it.
I have seen youtubes that did this where they added a car-starter solenoid switch as the on/off element. The problem with that design is that a car starter solenoid switch is a single physical switch that passes all of the full amps of the 12V car battery (12V X 200A = 2400W)…
The total power (with no resistance in-line) can include an energy as high as 800-CCA (Cold Cranking Amps). If there is an engine starter-motor in-line as a load (the “resistance” of the circuit), then a typical 4-cylinder engine would normally draw about 200A for a half second, and then it would settle into around 100A to continue spinning the engine.
However, it is a “bad design” to use a car solenoid starter-switch as the on/off element in a spot-welder that is powered by a car battery. This type of switch can occasionally permanently weld itself into the “on” position, since it would normally ONLY pass 200A peaks, but “now” you have removed the resistance of the cars’ starter-motor and the switch is passing the full 800A that the battery is capable of.
That’s the kind of amps that would produce enough heat to melt nickel strips as a spot-welder. I only mention this to point out that a common 12V car battery has proven to be adequate to power a spot-welding operation for 18650 cells.
Also…because a car-starter solenoid depends on two physical metal contacts to touch [and then be pulled away by a spring], doing this would cause a very inconsistent amount of welding pulse-time (measured in milli-seconds, which is one-thousandth of a second)
So…what can we use as an on/off switch that can handle high amps? and could also be controllable to adjust the spot-welding pulse in thousandths of a second? Perhaps the humble and common MOSFET semiconductor?
A “FET” (Field Effect Transistor) is a clever type of transistor that can be used as an signal amplifier, or a diode, or an on/off switch…depending on how it is configured in a circuit.
The Malectrics kit is a precision timer that uses a group of FETs [in parallel] to turn on-and-off a LOT of battery current as a spot-welder. You would also need to add a large [and fairly new] 12V car battery as the power source, and the most common mistake for a new builder is to use a “small and old” car battery that they already had. If you are using the longest weld-pulse setting, and yet you are still getting weak welds? it is likely that the battery you are using is weak.
12V X 800A = a pulse power of 9,600W
If the Malectrics device interests you, their website can be found by clicking here.
In the interests of full disclosure…when I first saw the kWeld, I became the first USA dealer for them, in order to speed up the quantity of US-based users. It looked very innovative and promising, and I wanted actual users to beat on them to see if it had any weaknesses, and…the results have been impressive. I was retailing them for about a year, and the company has grown rapidly, so they now have a “real” business partner to manage all US sales [18650 heat shrink & cell holders, inc].
At first glance, the kWeld appears to be very similar to the Malectrics. As far as function is concerned, one spot-welding cable simply passes its current through a large fuse from the car battery post to the spot-welding probe. The second battery post passes it’s current through a group of paralleled FETs that act as the on/off switch, on its way to the second spot-welding probe.
The kWeld costs more than the Malectrics, and here’s why. It uses a proprietary management system for the electronics, which measures the current passing through the device in real-time. This means that if the input voltage slowly drops while you are using the welder, the device automatically adjusts the pulse timing in order to ensure that the total amount of energy delivered is very consistent from one weld to the next.
It also has a battery-test mode to inform you if the battery you intend to use will be able to do the job. The original version of the kWeld was specifically designed to be used with either a lead-acid car battery, or a 3S “Lithium Polymer” pack (3S, 11.1V nominal LiPo, 12.6V when fully-charged). However, the acceptable selection of LiPo packs is limited to only a few specific models that are known to have very low resistance, and are capable of sudden high bursts of amps, without suffering any damage (look for the 3S battery designation of “130C, 5000-mAh”…or more).
Between the use of the readily-available car batteries (locally sourced, because they are heavy, and shipping would be expensive), and high-amp LiPo packs, the early kWeld devices developed a great reputation.
However, almost immediately…some customers requested the option of using super-capacitors (which usually operate at lower voltages). The second version of the kWeld had its voltage range widened so that the kWeld will work with an input from as low as 3V up to 30V (many nominal 24V / 7S lithium packs have a fully-charged voltage of approximately 28V).
The kWeld is made by Keenlab.de in Germany, and it is owned and run by an electrical engineer. Based on the success of a few enthusiasts who used super-capacitors, Keenlab has designed and began producing a supercapacitor-bank option which operates at 8.2V, plus a power supply that can take your 110-VAC house current and provide the 8.2V at very high currents, so that the power bank refills rapidly.
If you have access to a large quantity of used 18650 cells, using a 2P group that is charged to our recommended 4.1V per cell will provide the 8.2V that is needed for the kCap supercapacitor bank. Low-amp cells can slowly charge the kCap, and when you actuate the kWeld device, the supercapcitors will discharge the total energy very rapidly.
Although the factory warranty has limits, this welder has been used at 8.2V X 1400A for a pulse power of 11,500W
Global kWeld customers can contact Keenlab.de by clicking here.
For kWeld customers located in the USA, you can contact Keith at “18650 Heat Shrink and Cell Holders INC” by clicking here.
The kWeld North American Facebook support group can be found by clicking here.
Copper and Nickel Buses
If you are happy with a ‘common’ ebike kit, then you will also be happy with a common ebike battery pack [my default recommendation for a battery pack is a high-quality 48V-52V pack, like the Luna Cycle packs, as of the summer 2019].
That being said, maybe you are building an electric motorcycle, or perhaps a high-powered electric bike? You will NEED your pack to provide more amps than a ‘common’ pack. 18650-format cells exist that can provide 15A-30A (currently, Samsung 30Q and 25R, LG HB2, Sony VTC6). If your pack design requires a high-amp cell like this, the design bottleneck will be the cell inter-connects, and the two pack collectors (the positive and negative terminals).
Pure nickel as a bus material (and spot-welding as an assembly method) is the default choice for “common” cell interconnects (18650 / 21700 cylindrical cells). However, when your design needs HIGH pack-amps, the bus material becomes a critical design decision.
The common pure-nickel series bus-ribbon has a width of approximately 8mm (1/3rd of an inch wide), and it’s current-carrying capacity is measured by its’ thickness. Common battery packs (a 25A peak from 5P?) uses 0.15mm thickness for 5A per an 8A-rated cell. A 0.20mm thick nickel ribbon is common for the popular 10A rated “high-capacity” cells, like the LG MJ1, Samsung 35E, and the Panasonic GA.
Currently, the builders’ dilemma happens when using a high-amp cell (15A-30A). I have seen builders using double-layers of nickel ribbon for the series connections, but…I do not recommend that approach. At higher amps, using nickel as a bus-material ends up being more of a resistor than a conductor. By that I mean…it converts a LOT of battery watts into heat instead of accomplishing work. Also, the two terminal collectors will experience voltage sag, so the builder will not get the full amps that they paid for by using authentic high-amp cells.
This sounds like a joke, but I assure you, this really works. I wouldn’t use magnets on a nickel bus, because nickel is too stiff to conform to the surface of the cell tips. I feel that copper sheet is a very viable material for 15A-30A high-amp cells, such as the 30Q, 25R, HG2, and the VTC6.
If you want to experiment with copper bus materials, I would start with 30-ga, or possibly 32-ga (sheet-metal gauge is different than the wire-gauges we are already familiar with). One mil is one thousandth of an inch.
These thicknesses of copper sheet are thin enough that they can be cut with scissors. Copper is soft at these thicknesses, but an equal thickness of nickel is very stiff.
In the pic above, the soft copper sheet is being cut to form bus-plates that provide both the series and parallel connections. Neodymium “button” magnets are surprisingly powerful, and the electrodes of 18650-format cells are “nickel-plated steel”. Nickel has a very low amount of magnetic attraction, but the steel cores on the cell-electrode tips will strongly pull towards these magnets.
Since magnets are also very strongly drawn to each other, the builder above has glued the magnets onto the copper sheet with a gel “super glue”. You can also purchase button magnets with a chamfered hole in the center, which would allow you to insert a flat-head screw from the cell side through the magnet, and screwed into some type of side panel (perhaps 1/4-inch plywood, or plastic kitchen cutting board?).
I would also recommend firmly attaching the bus-plates by some method onto the plastic cell holders so the plates do not move around. I am told the magnets hold the plates firmly in place, but in a crash, I am concerned that they might shift around and possibly short out.
Spot-welding a Copper / Nickel sandwich
This is a new development. There have been many experiments with copper bus-material that is nickel-PLATED in the hopes that doing so would allow it to be easily spot-welded. However it is not readily available for purchase, and companies like Makita, Milwaukee, and DeWalt use an expensive laser to weld them onto cordless tool cells. This left home DIY nickel-plating as a subject of many experiments, which have not worked out so far.
Recently, an anonymous youtuber has posted that if you put nickel on top of a copper ribbon, It spot-welds easily with the common spot-welders listed above (kWeld, Malectrics, Boss-Level, Riba, etc). Here is the link to the youtube, fast forward to 8:38 to see the copper welding part.
The copper and nickel strips shown above are both 0.20mm thick, and 0.15mm thick nickel is reported to also work. Further experiments are continuing to determine all the other thicknesses that would work. After seeing this, my research revealed that this is an old metal-workers trick when trying to spot-weld aluminum sheets to each other. Here is a link to a youtube video that shows the metal sandwich method being used on aluminum.
There is some speculation as to why this works, but it has been verified to work, so this is an exciting development for builds when you want to use high-amp cells. An interesting side-note is that some builders have ordered pure nickel in the past as a bus material, and when it arrived…they discovered they had been sold nickel-plated steel.
It was bad enough to get ripped-off, but now they had to wait for authentic pure nickel to arrive in a second order. However, if you or a friend have any nickel-plated steel ribbon just sitting on a shelf, the copper in a nickel-copper sandwich would carry the current, so the nickel part of the sandwich could be the cheap steel-core ribbon.
Using high-amp flat pouch cells
Although I have occasionally seen large-format flat pouch cells on ebikes, and 18650 cells used on E-motorcycles, you will most often see it the other way around. For most ebikes, 18650 cells remain easy to configure into a custom pack that fits an unusual shape, and the performance and range of 18650 packs are often the best fit for most ebikers, with pouch cells being the style of choice for E-motorcycles. A cylindrical 18650 cell is basically just a pouch cell that has been rolled up and inserted into a metal cylinder.
However, even though using flat pouch cells will force the builder to use a fairly large rectangular shape as the building block of the pack, motorcycle conversions often need such high amps, that flat pouch cells are the only viable option.
Once you decide that you will be using this format of cells, I believe that the cell-connections are actually easier to put together, compared to making high-amp spot-welds.
If you add physical compression on the side-faces of flat pouch cells, it will lower the internal electrical-resistance. Doing that spreads out the current-flow more evenly across all of the chemically-reactive internal surfaces. If half the internal surfaces have a poor contact, they will flow less current and run cooler, but the surfaces with good contact will then be forced to flow much more current than they were designed for.
If this happens, then the internal “hot spots” will cause the electrolyte in those locations to “off gas”, which leads to a puffy cell appearance. The more gas that is generated, the worse this condition becomes (gasses forming inside will cause delamination of the internal layers).
Using only 20-PSI of compression force is much better than nothing, and I have seen published reports that pressures up to 100-PSI can be beneficial.
In the pic above, I have purchased some high-amp pouch cells, and I am tracing out the shape to cut cell separators, which will help to prevent a short between the tabs when I have them on my workbench while I’m connecting the cell tabs.
In the pic above, I am using pouch cells to assemble a 12V suitcase pack to charge my phone and laptop during a power outage, since I live in tornado country (and I formerly lived in earthquake country). The extra-large tabs are a clue that I have chosen high-amp cells, and I have done that so I can also use this pack to “jump start” my car, if needed.
I am using this opportunity to take pics of several parts of the pack assembly procedures that I feel are important for new builders. Please note that I am performing this task on top of a red plastic storage bin lid, and underneath this cell is a green plastic cell separator (seen in the pic above the previous paragraph).
[come back in one month to see additional pictures of this “pouch pack” build]
The tabs were marked as +/–, but it was in very small print, so I used felt markers to make large marks. I also used scissors to cut a tab-cover for the positive tab, using ziplock sandwich bags as the cover-material. I then attached the ziplock plastic tab-cover with a section of “low stickiness” blue painters tape, which is easy to remove and also does not leave any glue residue.
Take note that on this particular brand of cell, a small part of the top-right corner of the negative tab has been cut at an angle to make identification easier.
I found some aluminum flat-bars to use as tab-clamps, and these already had threaded holes. Be aware that the tab-clamps can be made of steel if you want, since they do not need to carry current (the two cell-tabs are pressed together, skin-to-skin). I am purposefully using aluminum clamps to act as a heat-sink. I previously believed that an aluminum plate that was located in-between the cells would draw internal heat out to the edge as a cooling measure, but doing that leads to the surfaces of the cell-sides to run cooler than the cores.
If the various regions inside the cell run hotter or cooler than each other, then that would also change the internal resistances of that cell. Extensive research has shown that cooling the tabs will lead to the most stable and even internal temperatures in the active materials, while still drawing-out some of the internal warmth in the active layers of the cell.
120F (49C) is very warm, almost too hot to touch. However, if that temperature is evenly spread out, then the cell should last as long as is possible. 140F (60C) is too hot, and no part of the cell should ever be allowed to reach that temperature. If one part of the cell is hot, and another part remains cool, then the hot part will die, which leads to that cell dying early, which leads to the type of voltage sag where the entire pack is retired.
Now, on to the construction of the tab-clamps…
It’s fairly easy to cut threads into a drilled hole on a 7075-alloy aluminum bar (click here for one option) because this alloy is more brittle, which allows the chips to break off. The 6061-alloy Aluminum will conduct electricity better, but it is very soft, and you can easily break a tap when using it as a bar-clamp material, when trying to cut threads into a drilled hole.
In this application, I like using threaded holes in the aluminum bar material (with a chamfered hole under the flathead screw-head) to allow the clamps to be as thick as possible, since using a common bolt-and-nut on a through-hole would require the clamps to be thinner. Since there is no lock-washer to to reduce any possible connector-loosening from vibration, I will use a thread-binding fluid, like Locktite.
This method allows me to use clamp-bars that are the same width/thickness as the cell. A series connection is made between two cell-tabs, and the clamp style shown has two identical pieces, so the material purchase calculations would be easy. Measure the thickness of the cell, and that thickness is the thickness of the clamping materiel, or thinner.
This popped up in a recent forum discussion comparing soldering buses to cells vs spot-welding. LFP (Luke) was the head battery pack engineer at Zero motorcycles.
“…This thread topic could be re-phrased as, is it better to hit yourself in the balls with a baseball bat, or a golf club?
Firstly, using FLIR on shiny nickel surfaces is laughable, because it’s something like 5-10% emissivity, so those temp values it’s reading are +or->100-degC of being useful for something.
Lastly, the temp on the top of the nickel is utterly irrelevant, as the top of the nickel isn’t the heat sensitive area.
When you solder a can, and then cut the cell apart, you will see that the separator layer is melted together and deformed on the ends, and the can has excessive gas pressure, and the thermal decomposition breakdown products of the carbonate esters in the electrolyte are going to cause it to age poorly and develop self-discharge from the impurities released by that thermal breakdown.
When you spot-weld a can, it’s not the outer-side that matters at all. Not even relevant to talk about, or look at, or consider. When you cut that cell apart, and you see the inside of the can, which was glowing red-hot steel (2500-degF) for the weld to be possible, you see that the internal corrosion-resistant surface lining of the can is vaporized away, and in it’s place is carbon debris from flash boiling electrolyte carbonates, and now that carbon will be polluting your cell to increase the self-discharge rates.
If you’re going to try to spot-weld, you don’t do it based on how the welds look externally, or how hard they are to [physically] pull off the can, those metrics are child’s play to nail, and just don’t matter with respect to making a pack that will last. You must tune the process by cutting open the bottom of each cell you bonded to and examining the damage to the end of the jellyroll, and examine the bottom of the can surface for what kinds of thermal decomposition products are now poisoning your otherwise very pure electrolyte.
Soldering packs is for applications where you don’t care if it even works once, and random self-discharge levels are acceptable, and random failure in a short time is desired.
Spot-welding (to the bottoms of cans) is for when you don’t care about leaking electrolyte out of the can from the stress-risers in the heat-affected zone near the weld fusion site (where it was 2500 degF), and you don’t care about having the corrosion resistance can-coatings remaining functional, and want to pollute your electrolyte with random carbon debris (think of setting your breakfast into a frying pan at 2500 degF)…”
So, what method should you use?
Well…only you can decide what risks you are willing to take, and what will fit into your budget. I only hoped to collect information about what materials and methods are currently being used, and to point out some of the dangerous pitfalls that new builders might be stumbling into.
As I pointed out, the positive tip of an 18650/21700 cell is fairly well isolated from the jelly-roll. This means that you can solder, spot-weld, maybe use fuse-wire, or use any other method that works for you. However, the negative end (especially the center spot) is sensitive.
For the negative, you can use magnets over a copper strip, a pressure contact over a copper strip (meaning a spring or Poron rubber disc), or perhaps quickly soldering a small fuse wire (located off to the side, never in the center of the negative end).
I hope this has been helpful in some way, and I will update this 3-part series as new information comes in.
Written by Ron/spinningmagnets, June 2019