Technical Progress Part III: Lithium Battery

As with our previous posts, you can skim the photos and captions for a quick read or dive into the text for more details.

The power capability of Boosted’s drivetrain would be useless without a battery that can match it.  From an iPhone to a Tesla Model S, the battery is often the unsung hero, usually quiet and hidden and only noticed when it stops working correctly.  Today we explain the performance and specs we needed from our battery, how we prototyped it, how it changed as we prepared for production, and how we tested our new design.

The final production battery and drivetrain.

Design Principles

To accelerate up hills, our drivetrain delivers an incredible amount of mechanical power.  To make this possible, the battery needs to deliver thousands of watts of electrical power, far more than most batteries are designed for.  Some batteries can deliver a lot of power for a brief amount of time, but since a long hill climb can take seconds or even minutes, we need sustained high power.  And while going down hills, the regenerative brakes need to be able to send substantial amounts of energy back into the battery, effectively requiring it to charge very rapidly.  This type of high power discharging and charging would damage most batteries, but since it’s normal operation for us, we need to reliably deliver this performance over hundreds or preferably thousands of cycles.

Riding uphill needs powerful motors AND a powerful battery.

It’s important to differentiate between power and energy, since they aren’t interchangeable words.  When we say energy (you pay your utility company per kilowatt-hour of energy), we’re talking about the range of the board.  Double the energy and you’ll double the range.  But power (which is usually measured in watts or horsepower) is how fast you use energy.  A high-power drivetrain is useless without a high-power battery, but with both you can go faster up steeper hills.

When we say powerful, we mean a different scale from most devices.  From left to right are AAA, AA, and 9V alkaline batteries, an Apple USB charger, a small and cheap AC-DC converter, a high-quality AC-DC laptop power supply, and the battery cell we use for our pack.  This cell can produce 230W continuously or 384W for a 10-second burst.

Secondary to delivering the necessary performance, the battery needs to be as light and compact as possible.  In our last technical update, we discussed why this is important to maintaining a great longboarding experience and keeping the vehicle as portable as possible.  The biggest differences in performance vs. weight come from different types of battery chemistries, and here are some of the most commonly used ones.

  • Sealed lead-acid, or SLA, is very cheap but very heavy.  It’s most commonly used for starting cars, and also found in low-cost electric scooters, bikes, and skateboards.
  • Alkaline is usually not rechargeable and is most commonly seen in small electronics in AA, AAA, C, D, and 9V formats.
  • Ni-Cd is rechargeable and cheap but hard to recycle.  They used to be common in electronics and remote-control toys, but aren’t nearly as weight-efficient as lithium batteries.
  • NiMH is rechargeable, moderately powerful, and commonly used for modern rechargeable AA-sized batteries.
  • Lithium-ion is the most expensive but it has the highest energy and power for its size.  This is why it’s the most common battery for modern phones, laptops, and electric vehicles, despite higher costs and greater engineering challenges.

From the beginning of this project, we’ve made decisions by asking ourselves “what would we want on our own personal longboards?” and then building it.  In this case, especially with our obsession with portability and handling, the answer was simple:  lithium-ion.

The battery must be compact enough not to touch the ground during hard carving, even with deck flex.

The most important design constraint was safety and reliability.  High-power lithium batteries need more care than the average car battery, with more complicated charging procedures and safety electronics between the battery and the rest of the system.  This was our number one priority, so it came first over performance and weight.  So our design parameters, in order, were:

  1. Safety and reliability during charging and operation
  2. Sustained high power over many charge/discharge cycles
  3. As light and compact as possible

Range

Most short trips around the home or office are only 1-2 miles (0.6-1.2 km).  About 1/4 of US commutes are under 5 miles (8 km).  And most public transit stations in urban areas are within similar distances.  We found by testing early prototypes that a 6 mile (10 km) range was more than enough for short trips and commutes, especially with a fast and portable charger.  For traveling more than that distance, a longboard usually isn’t the vehicle of choice.

Since we didn’t give the board any more range than needed, the battery remains incredibly light and compact.  It still handles like a great longboard, it’s easy to push, and it’s easy to carry.  Our test riders told us that the portability and handling of the board was way more useful than having a 10 or 20 mile (16-32 km) battery at the expense of added weight.

It actually turns out that a longer-range battery pack is easier to engineer, since you can use several regular, low-power batteries in parallel to get high power for the drivetrain.  The biggest challenge in designing our battery was getting this incredible amount of power without resorting to using a much larger and heavier pack.

We measure the range using a 175 lb (80 kg) rider on flat, smooth pavement at average riding speeds.  Riding very fast, on rougher roads, with a heavier rider, or up hills will decrease that range.

Prototype Battery

Our very first prototypes were built using the lightest and highest performing batteries available:  lithium polymer (or li-poly or LiPo).  These are most commonly used for expensive remote-control airplanes and cars, with a battery as small as a deck of cards able to deliver 1.3 kW of continuous power and 2.5 kW of burst power.  We mounted a 6 mile (10 km) battery to our drivetrain and it worked incredibly well.  So well, in fact, that every board we built between the first board in 2011 and our second beta run in June 2013 used LiPo packs.  So why change?

The lithium-polymer charger for our prototypes was cumbersome to use.  

There are several varieties of lithium-ion batteries, and most lithium-polymer batteries use lithium cobalt oxide (LCO).  The high power variants used in the RC world can easily be damaged or even catch on fire if they’re overheated, overcharged, or punctured.  So they need to be monitored while being charged with a special “balancing” charger.  These chargers, and the need to always be with the board when it was charging, was a frequent complaint from test riders.  We also sometimes had to replace batteries that became physically damaged, “swelled”, or had individual cell errors.  And it was difficult to calculate the battery’s state of charge without more advanced electronics.

High-power lithium-polymer batteries are easy to damage without proper care.  These are some of our test batteries that are no longer functional.

To hold the battery to the board, we designed a fabric pouch with button snaps.  This was a light design that was decoupled from the board flex and easy to fabricate in small quantities.  But it had no mechanical protection, didn’t look that great, and provided no water resistance.

LiPo batteries were housed in a fabric pouch with button snaps.  This was easy for protoyping but not durable for long-term use.

Production Battery

We first tried to find an existing battery pack that met our requirements.  We found small packs that couldn’t output enough power as well as high power packs that were large and heavy, but nothing met our needs and we knew a custom solution could work.  So we decided to pursue a completely custom battery that had never really been built before.  The design involved choosing the right battery cell, adding a battery management system (BMS) and enclosing the pack in a protective case.  And, of course, we would need to extensively test this new design.

Cells

After realizing the inherent risks in some lithium chemistries and finding it difficult to source low volumes of high-power cells, we decided to use a different lithium-ion chemistry known as lithium iron phosphate (LFP or LiFePO4).  This provides safety and high power, though at the expense of slightly more weight and bulk compared to LCO.  With a few suppliers to choose from, we optimized for reliability and performance with a very high-quality cell that we could easily obtain in the volumes we’re producing.  We use 12 cells in series, each with a nominal voltage of 3.2V, for a total pack voltage of 38.4V.

Some of the lithium cells we considered for our battery design, both in pouch and cylindrical formats.  

BMS

The battery management system, or BMS, is hardware and software that keeps the battery operating safely and effectively.  It protects against failures like
  • undervoltage, where one cell drops too low
  • overvoltage, where one cell is charged too high
  • overcurrent, where too much power is drawn from the pack
  • overtemperature, where the pack overheats
  • short circuit, where the positive and negative terminals of the pack are connected

To protect the cells, the BMS can turn the entire pack’s ability to charge and discharge on and off.  It also controls charge cut-off, balancing of the cells during charging, and state-of-charge (SOC) calculations to determine how much range is left on the longboard.  The BMS has its own processor and talks over CAN bus, a robust automotive-grade protocol, to the main motor control processor.

Surface-mount components were hand-assembled onto the BMS during early testing.  Production BMS circuit boards will be assembled by machine.

Enclosure

The enclosure design started with deciding how to arrange the 12 cells for maximum clearance from the ground and wheels and then adding space for a charging port and an on/off button.  To quickly mock up different cell configurations, we cut PVC pipe to the same dimensions as the cell and taped them together to check for fit.

PVC pipe mockups of battery cells, taped in different configurations, were used to prototype and test before building a single battery.

A flat rectangular block was the simplest layout that met our clearance requirements.

Once a cell layout was decided, we moved to CAD models and eventually 3D prints to test how different designs looked and felt in person.  We even created a CAD model of the BMS circuit board to make sure it fit correctly into the enclosure.

Mockups were made using foam, 3D-printed ABS plastic, and heat-shrunk plastic before the final part was injection molded.

Once we were happy with the enclosure, we paid for the injection mold tooling and got our first shots of the enclosure.  We use the same rugged glass-filled nylon as the electronics enclosure.

The battery pack integrates the cells, BMS, on/off button, and charge port.  Its processor communicates with the motor controller using CAN bus.

Charger

In addition, the balancing charger with buttons and a screen has been replaced with a simpler laptop-style charger.  The standard charger should complete a full charge in about 90 minutes, faster than most other electric bikes, scooters, and skateboards.

The new charger is similar to a conventional laptop charger with a simple barrel plug, with a huge improvement in the charging experience.  Here are some of the final test boards being charged.

Fixed, not Swappable

The LiPo battery we used for most of our testing was always swappable, and we planned to keep it that way in production.  But during the engineering phase of the project, with the heavy vibration and potential for a water splash onto the battery, we discovered that a connector was much less reliable, and therefore not as safe, as soldering the battery wires permanently.  And we also noticed that of the 40+ riders using 20+ boards for errands, commuting, and fun, only one ever asked for a second battery.  Finally, each battery also needs the BMS to be attached for safety reasons, so the cost of a second battery (along with enclosure and BMS) would be prohibitively high.  Since safety and reliability is paramount for us, and since our test riders have been happy without swapping batteries, we’ve decided to remove the connector and make the battery fixed and non-swappable.

Testing and Results

We tested the battery on the benchtop using high-power supplies and loads. 

A laptop and XBee radio were set up to download wireless telemetry from the electronics during road testing.

We tested the new battery design on the benchtop using our motor dyno and also solid-state loads, and of course we’ve spent hundreds of hours outside with test units.  Expect a future blog post about our testing equipment and the interesting things we’ve learned.

The end result of this testing and development is a battery that works incredibly well, with the ability to supply thousands of watts of continuous power from only 12 small cells, an easy on/off switch, easy and rapid charging, and a beautiful design.  Our cells are rated for thousands of cycles, so we expect each pack to last for years of daily use.  And most importantly, we have a safe, high performance, and compact battery that preserves the design vision we had for the lightest, most powerful electric longboard ever made.

Five test units with the production battery waiting for delivery.