Most garage-builders who decide to assemble their own battery pack usually have a lot of experience. However, pack-building continues to be a frequent source of questions from new ebikers, so I decided to put together an article to cover the basics.
Cells in Parallel, the P-Count
[In this article, I will use a small rectangle-shaped pack as an example, which is the easiest style to understand when learning these principles. Once you have a firm grasp of this, you can easily scale your pack up to different shapes and sizes]
If you take several cells and connect their same-type electrodes in parallel, they all will act as though they are one large cell. Also, when you first connect them to each other, they MUST be at the same level of charge [The positive electrode is the Cathode, and the negative electrode is the Anode. In the pics below, the positive is a red plus sign, the negative is a black dash].
By that I mean…if you connect a fully charged cell “in parallel” to a cell that is low, the high cell will try to charge up the low cell in just a few seconds, since there is no built-in resistance between them across the connecting bus to slow things down. A rapid equalization will absolutely cause both of them to get VERY hot. Permanent damage is the least that will happen, and most likely one of them (or both) will go up in flames. However, if there is only a tenth of a volt in difference between them, that would not be a problem (for instance, from 4.1V to 4.2V would be OK, they would both equalize at 4.15V).
The P-count determines the capacity of the pack in Amp-hours (Ah), and it also determines the amount of current the pack will be able to produce, measured in amps. For this example, we will use my favorite ebike cell, the Samsung 30Q. It is factory-rated as having 3000-mAh (milli-Amp-hours), which is the same as 3-Amp-hours (3-Ah). If you have four in Parallel, the finished pack can be rated as 12-Ah’s worth of range (4P X 3-Ah = 12-Ah).
Now we come to the amount of power that the pack can safely put out. The 30Q is a hot-rod cell (along with the HG2 and the 25R), and it is factory rated at 15-Amps continuous. However, almost all cell models get hot if you actually run them continuously at their listed rating. I recommend that nobody should ever let their pack get hotter than 140F (60C) under any conditions. Doing that will lead to a very short pack life.
Fortunately, ebikes usually only draw their peak amps for a few seconds while accelerating. Once you reach a cruise phase, the continuous amps that the controller and motor will draw is MUCH lower when you are simply maintaining your speed. If we use the 15A figure as our designed max amps, the pack will just get a little warm, and that means it will last a very long time. Four 30Q cells in parallel that are rated for 15A means we can depend on getting 60A from this pack without damage (4P X 15A = 60A).
Cells in Series, the S-count
When you connect cells together in series, it doesn’t change the amps or the capacity, it only raises the voltage of the pack. Hooking them up in series means that you connect the positive end of one cell (or P-group) to the negative end of the other.
For the cylindrical 18650-format cells that are the most popular (18mm in diameter, 65mm long), they use the lithium-NCA or lithium-NCM chemistry (the cathode uses Nickel-Cobalt-Aluminum…or…Nickel-Cobalt-Manganese). Those chemistries have a nominal (average) voltage of 3.7V…and in order to get the longest possible life from the pack, use 3.3V per series-cell as the Low-Voltage-Cutoff (LVC), and 4.1V as the fully-charged target.
The common max charge is 4.2V per cell, but when cells rest (for any length of time) at that high of a voltage, it will significantly degrade their life. Charge the pack to 4.1V times the series number.
The BMS, how to connect it?
The BMS is the Battery Management System. It performs several functions. The two fat wires (red and black) from the charger will “bulk charge” the pack until it gets very close to being full, and then the charger will switch over to using a very low charge rate as it gets closer to being full. A 3A or 5A (continuous) charge rate is very common for the bulk charge.
This charging profile is called CC/CV, for Constant Current / Constant Voltage. It’s a simple and inexpensive way to accomplish a subtle goal.
We all want an affordable battery pack, so…we buy mass-produced cells. This means that there will always be very minor differences in the internal resistances of each cell. To use the example of our theoretical 7S/4P pack above…each 4P cell-group is “seen” by the charger and controller as one large cell. The parallel connecting metal strip ensures that they all constantly equalize to each other, so we must discuss them as if they were in fact, one large cell.
There are seven of these “in series” to get 24V. Now, we then set our dumb bulk DC power-supply as a charger to 3A (with NO BMS), and we use (7S X 4.1V =) 28.7V as our fully-charged goal. It works like a dream. However, only five of the P-Groups are actually at 4.1V. One P-group is at 3.9V due to high internal resistance, and another P-Group is at 4.3V due to low internal resistance. Since our dumb charger only reads the 28.7V of the assembled pack when it shuts off, it has no idea of the trouble that is brewing…
High-resistance cells run hotter than an “average-resistance” cell, but for this discussion, let’s just assume that it never gets “too hot” to cause trouble (over 140F / 60C). Also…cells located at the heart of a pack run hotter than cells at the edge, since the edge-located cells shed “some” heat to the outside shell of the pack.
That leaves the low-resistance cell to consider. It will dump amps faster than the other cells (when accelerating), and it will also gulp the charge faster, too. It will actually run cooler than the other “average resistance” cells, but…a bulk charger will overcharge it. If letting a pack sit overnight at 4.2V per cell will cut its life in half (compared to 4.1V per cell), then…what will letting one cell sit overnight at 4.3V do? It will lose capacity rapidly.
And that means that, the one bad cell will cause the entire P-group to experience voltage-sag near the end of a ride, and then…that one low P-group will cause the entire pack to experience voltage-sag. And this means that…for a split second on acceleration, the LVC will “think” that the entire pack is too low, and it will cut off ALL power in order to “save” the pack (one of its most important jobs).
The pack will still “work”, but…your accelerating days are over. You may have planned on buying a new pack in three years (or more), but because of one “slightly” bad cell, the entire pack is now useless to you after only a few months. This is why the bulk-charge phase (CC/Constant Current) only takes the pack to about 4.0V per cell. For the rest of the topping-charge, the Constant Voltage / CV phase is accomplished at low amps, with some sensitive electronics thrown in…
There is one crazy “voodoo electronics” feature of a BMS. The quantity of small balancing wires is going to be…the “series number plus one”. So, our theoretical 7S pack will use eight balancing wires. This just sounds nuts. Everyone who starts learning about batteries KNOWS that there are two poles (to each cell, or P-group). The red positive, and the black negative. You would think that to balance the charge at low amps during the final CV phase of the charge…you would need 14 wires instead of eight, when making minor adjustments to the seven P-groups (remember, each entire P-group will act like one large cell).
Look carefully at the pic above. The white plug connects the dumb harness to the smart BMS. You can see that for the BMS to access the positive and negative ends of the first cell (and the connected first P-group), it uses the number-1 and number-2 wire. However, when moving on the the second cell?…the BMS uses the #2 and #3 wires. The number two wire is sometimes used as the positive to the number-1 cell, and sometimes used as the negative to the number-2 cell (and “so on” down the line, until you reach all of the P-groups in the series string).
This eliminates 45% of the possible BMS harness wires, but..it also makes the proper connection locations confusing for new builders. Even if you are still a little confused, just follow the BMS instructions carefully.
Side note: If your battery pack dies suddenly for no obvious reason, it is usually from some component in the BMS failing, leading to a complete drain of the pack down to zero-volts, or…a gross overcharge above 4.2V. This is why they are sometimes called a Battery-Murder-Suspect.
Other chemistry options?
The common NCA/NCM chemistries provide the best balance between capacity in a small package, and a useful amount of amps. However, there are other chemistries that might be an option for certain applications. The NCA/NCM “nominal” voltage is widely regarded as being 3.7V per cell, and…for the max possible life, I’d use 3.3V per cell as the LVC and 4.1V as the full charge.
LiFePO4 / LFP is commonly called “Iron Phosphate”, and it has a nominal voltage of 3.2V per cell. That means that it takes 16 LiFePO4 cells to make a 48V pack, and NCA/NCM only require 13 cells for 48V. However, LiFePO4 is considered the most fire-safe (sometimes found as a starter battery on small aircraft), and they also typically last about twice as long as the common NCA/NCM 18650-cell packs.
A 4S pack of LFP is the most common replacement for a 12V Lead-Acid battery pack (4P X 3.2V = 12.8V nominal).
That being said, NCA/NCM in the 18650-format cells have a much better selection of choices, and provide high power and long range in a small package that is affordable, due to mass-production. LFP can be found in flat pouch cells, 26650’s (26mm X 65mm), and large cylindrical 38120 cells (huge! 38mm X 120mm, from Headway, see below). I have actually found LiFePO4 in the 18650-format, but…the capacity was VERY low.
One more obscure chemistry is Lithium-Titanate-Oxide (LTO). It has an even lower nominal voltage of 2.4V per cell. It may last ten times longer than NCA/NCM, and it also works well in very cold weather (along with providing high current). However, I have only found them in flat pouch cells, and they are hard to find, expensive, and have few size options.
The NCA/NCM chemistries can also be found in the slightly larger cylindrical 21700-format cells. These were developed by Panasonic in partnership with Tesla electric automobiles. However, they are currently expensive, and the majority of the factory production is going into the Tesla Model-3, and also the powerwalls (a home power back-up system).
There ARE options, but…after any depth of research, almost every ebiker comes back to the 18650-format cells in the NCA or NCM chemistries.
Honeycombs are Sweet
In the pack example above (7S / 4P), the cells are aligned in straight rows and columns, which I might refer to as “rank and file” (like soldiers marching). However, the next most popular arrangement is to nest the cells of one row into the valleys of the next, in what many have come to refer to as… a honeycomb layout. I like it, but there are benefits and drawbacks to every option.
Using a honeycomb layout is mathematically the most compact way to arrange cells. That being said, the common rank-and-file arrangement leaves “just enough” space between four adjacent cells to allow a thin bolt to pass through from one side plate to the other (as seen below), and a honeycomb does not.
The plastic end-cap cell holders are an important safety feature for a DIY pack. The common square layout usually positions the cells so that they are almost touching each other. That means there will be almost no air-circulation, and that leads to the cells in the center of the pack to gradually run warmer by the end of a ride.
If you have a large enough pack, and the cells can provide lots of current, then…a low-current ride will not even get them warm. At the other end of the scale, a small pack of low-current cells that are being hammered on a ride will definitely get hot. The hottest spot will always be the positive cathode tips on the cells in the center of the pack.
In the pic below, you can see that when using a honeycomb pattern, the side angle of the stacked cells is about 30-degrees away from the 90-degree corners in a common rectangular pack. This makes it a natural fit into the lower part of the frame triangle space. Mounting the weight of a pack low and in the center of the frame is the best place to put them, so the bike will handle well.
The Russian battery pack configurator program can be found here. Feel free to experiment with every possible setting to see what the setting widows do.
Triangle packs, and odd shapes
The easiest pack to design is a rectangle (as seen above). However, like the pic below, it’s sometimes useful to make your pack a triangle, or some other odd shape. In the pic below, the builder is trying out a dry-fit to see how the 100 cells shown would work, and also how to arrange the 5P paralleled groups which are now in odd-shaped clusters (instead of being in-line).
The huge battery pack shown above is from Mark’s Cromotor Phatrod, which can be found by clicking here.
What voltage to choose?
The actual precise voltages are approximate, but…here are the most common 18650 cell Series-counts, and the nominal voltages.
48V…13S____________88V…24S (100V when fully charged)
For high amps and good range, I like the 30Q which provides 3000-mAh, and 15A. If you want 20A per cell, check out the Sony VTC6, LG HG2, and Samsung 25R
If your pack is big enough that you don’t need the high amps of the 30Q, you can get more range by choosing one of the popular 10A cells, which are generally rated at around 3400-mAh. Three examples are the Samsung 35E, LG MJ1, Panasonic GA (a few months after any article is published, there may be even more options to compare, so do your homework before buying)
I like 48V and 52V for a variety of reasons. If you went to a higher voltage, it would require more parallel groups to be added to the series string to raise the volts. It would be easy to quickly end up with a design that is large, heavy, and expensive. Years ago, there were few options when it came to how many amps the 18650 cells could put out, so the only option for high performance was from high volts.
That being said…the more volts you use, the fewer amps you will have to pull from the pack to get the performance that your throttle is demanding.
Now…there are 18650-format cells that can put out 20A, but they are pricey, and the 30Q is popular for a reason. It can put out 15A, so a relatively small 4P pack can put out 60A. A 13S / 48V pack using 4P would be only 52 cells, and it would have 12-Ah of range. Efficient mid-drive systems can get up to 2 miles per Ah, so 12-Ah could result in over 24 miles of range.
As far as going to a lower voltage than 48V/52V, if you live where it’s fairly flat, then you might get acceptable performance from 36V (10S). A 4P pack of 10S is only 40 cells! (very easy to fit). Of course, even if you don’t need lots of volts, or lots of power, if you have the budget and the frame space to mount a larger battery, then the pack will run cooler. Helping the pack to run cooler will help it last as long as possible.
One last note, an ebike battery is one of the biggest battery packs you will likely ever buy in your life. If you can accomplish your goals with a 48V or 52V pack, either one of those can power an inverter in a disaster to provide 120V AC to your home. If you use 4P of common 10A cells (40A), and the pack is 52V, then…40A X 52V = 2100W. That’s enough watts to run your refrigerator and TV for quite a while. And the bigger the pack, the longer your appliances can run…
Why go to all this trouble?
In the pic below, ES member Kepler built a tiny 14S / 4P pack (52 cells) that perfectly fits his favorite ebike frame, which uses a Bafang BBS02 mid drive. In the interests of full disclosure, my favorite ebike uses a large 52V pack that I purchased from Luna Cycle.
Once you have decided on the voltage, the size, and the shape, I’ll be covering a lot of things in part-2 (click here), and among them are all the popular methods for connecting the cells to each other, such as spot-welding and wire-bonding (add link to part-2 here)…
If you liked this article, you might also like…
Send suggestions, corrections, and death threats to: Prisoner #41, Kansas state correctional facility for the criminally insane.
Written by Ron/spinningmagnets, February 2019