Pack Design
Our primary power concerns and limitations are from our NGM-SCM150 motors and WaveSculptor 20 motor controllers. According to their respective datasheets, the NGM-SCM150 has a peak power consumption of 7.5kW and the WaveSculptor 20 has a continuous voltage maximum of 160V. Aiming for a maximum pack voltage of 150V gives us some buffer room and results in a peak current draw of around 50A per motor from the motor controllers.
...
Thus, with a 36s36p configuration, our total pack would require 1296 cells. At 48.5g each, we would have a total battery mass of 62.9kg. If we wanted to aim for a 60kg pack, we could build a 36s34p pack, resulting in 1224 cells at 59.4kg. By our target of 1C, this would still safely support 108.8A.
Note: To clarify the nomenclature we're using here, 36s denotes that there are 36 individual units (modules) connected in Series, while 36p denotes that there are 36 individual units (18650 cells) in parallel.
Module Design
We define each set of parallel cells as a module. Our goal is to design the modules such that they will be relatively easy to manufacture, modular, and easy to replace. We have two major concerns - airflow concerns—airflow and electrical connections.
Our current design takes advantage of 8s6p 18650 brackets available from China. To maximize airflow without compromising total volume required, we have decided on the cell arrangement shown below, where the x's mark the location of cells within a particular module. This allows for an air channel for each row of cells, while still packing the cells reasonably closely. Each module would then be approximately 7cm x 12cm x 16cm. With airflow from the side, this should adequately cool the pack. Modules would be arranged with the two air channels are running parallel to the box bed.
| x | x | x | x | x | x |
| x | x | x | x | x | x |
| x | x | x | x | x | x |
| x | x | x | x | x | x |
| x | x | x | x | x | x |
| x | x | x | x | x | x |
To connect the cells electrically, we plan We originally planned on spot-welding a grid of nickel strips to the cells. This is a standard procedure for building 18650-based packs, and is much safer , and more reliable , and easier than soldering or a purely mechanical solution. In order to carry the current, we plan on running 10 awg copper wire in the gaps between the blocks of cells which then run to blade connectors. To support our peak current draw of 100A, we should have . We would then solder 6 or more 10 awg wires connecting each module. Ideally, these should be evenly distributed for current sharing.
To reinforce the copper wire, we can punch holes in the nickel strip grid where the wire will be run and source thin copper discs to be placed where a cell would normally go. Then, we can solder the wire directly to the copper disc, providing some support from the 18650 brackets and reducing stress on the nickel strips. The purpose of punching holes in the nickel strip is to increase the copper to copper surface area. The goal is that during this, the nickel and copper would also be joined. This operation could be done before spot welding to reduce heat spread to our cells.
Note that this design is still in development and is subject to change. We have yet to determine how we'll be mounting these modules to the box itself and how it's supposed to withstand 20g's of force. We also need to determine which connector best meets our design requirements.AWG wires to the grid, terminating the wires in a blade connector. However, we have identified a number of concerns with this approach. The available connectors have a low lifespan of 25 mating cycles and require enough slack in the wire and clearance to actually connect and disconnect them. In addition, spot-welding all of our cells and soldering all the wire would be a labor-intensive task. Spot-welding requires trained operators and a calibrated machine, which are major bottlenecks considering around 3000 welds would need to be done.
To reduce labor, we are considering a purely mechanical pack. This approach uses a dimpled copper grid in place of the nickel grid, relying on compression through foam (and possibly magnets) to ensure an electrical connection. Even pressure will applied by running threaded rod through a number of modules in series with backing plates between each module. To achieve our series connection, we plan on extending the copper grid vertically and adding holes so that adjacent packs can be bolted together. We will likely use a copper block as a spacer, which can serve a dual purpose as our voltage sense tap. For safety, we plan on covering all exposed copper with insulated caps.
Inspiration
Original solderless pack design
Layout
An approximately 60kg pack will be difficult to lift. With additional electronics and the added weight of connectors, brackets, and the box itself, we expect the pack to be between 70~80kg 70-80kg total. To make this easier to manage, we are considering splitting the pack into two boxes. Note that we have approximately 40" x 47" x 9" of space to work with.
The AFEs that we are considering support up to 12 modules each, so with 36 modules, we will most likely put 24 modules in one box and 12 in the other. The pack with fewer cells will also contain the power distribution and BMS systems. If necessary, we can split the boxes differently, but this would require an additional AFE and processing of an incomplete AFE in the middle of our daisy-chain.
Both packs will also need an HV relay. The main box will also need to contain our auxiliary battery.
Follow-up Discussions
2017-03-01 Battery Pack Design Review
Decisions
Ultimately we decided to go with a Spot-Welded Pack Design for MSXII.