The previous chapter concluded that we cannot out-car the car, and that survival depends on reusing existing industrial ecosystems rather than inventing our own.

This chapter is about what happens when you try to do exactly that, and discover that for this class of vehicle, the ecosystem is not open to anyone buying fewer than a few hundred units and that even when it lets you in, what you get still needs a great deal of work before it is usable.

I - Two Mature Ecosystems, and a Hole Between Them

If you are building a bicycle, you can buy almost everything. Hub standards, disc mounts, bottom brackets, headset dimensions, and thread pitches are documented and largely interchangeable. Parts ship in quantities of one, usually with a datasheet, and any bike shop in Europe can service them twenty years later.

If you are building a car, the ecosystem is larger still: tier-one suppliers, a vast aftermarket, and decades of standardization.

Both are mature. Both let a small team assemble a vehicle out of parts that other people manufacture, document, and support.

The intermediate vehicle sits between them, and it needs things neither ecosystem sells:

• a kW-class drive that can be commanded and instrumented, not just switched on
• a control bus that accepts more than a throttle and a brake cutoff
• steering that can be mounted on an arbitrary chassis
• a way to convert pedalling into electrical power without a chain line
• structure and enclosure for a vehicle that is neither a bicycle frame nor a full car.

None of these is impossible to obtain, and it would be dishonest to claim otherwise. Almost everything on that list exists somewhere, in production, made by someone competent. The barrier is never existence. It is one of four things, and they require completely different responses:

1. Minimum order quantity. The part is real, mature, and sold only business-to-business, in hundreds, with lead times in months. A team that needs one to find out whether the concept is even worth pursuing cannot get past the sales process. Nothing is missing except access, transparent pricing and freely accessible datasheet.

2. Sold, but not usable as delivered. The hardware is cheap and available in ones. It also arrives with a thin datasheet and firmware written for a different application, and turning it into something that drives your motor reliably is weeks of work. You bought a component; you did not buy a working subsystem.

3. Sold only as a closed system. You can buy it, but only as a matched bundle you may not modify, instrument, or extend. Adding a function the vendor did not anticipate is not an engineering problem but a licensing one.

4. Not sold as a product at all. Every constituent part is a commodity; the assembly, or the interface between the parts, is not something anyone sells at this scale.

The practical effect on a small team is the same in all four cases.

II - The Walled Garden in Propulsion

Electric propulsion for light vehicles looks, at first glance, like a solved commodity problem. It is not.

What is actually for sale is a bundle: motor, battery, display, and throttle, sold as a matched set, speaking a closed protocol. Inside the bundle everything works. Step outside it and you find:

• undocumented or deliberately obscured communication between components
• no supported way to add a function the vendor did not anticipate
• integration of anything custom gated behind OEM agreements and volume commitments

You are not buying components. You are buying a system you are not allowed to modify.

For a research project this is fatal, because the entire method depends on changing one variable at a time and measuring what happens.

But the deeper problem is not ours. It is the one described in Living Within Limits: these vehicles do not die of mechanical wear. They die when a vendor stops supporting a display, discontinues a protocol, or declines to sell a replacement controller. The mechanical parts of an e-bike routinely outlive the electronics that command them. Lock-in is not merely a commercial annoyance; it is a mechanism of premature obsolescence, and therefore a sustainability problem.

A vehicle designed to last twenty years cannot be built on a drivetrain whose lifespan is set by someone else’s product roadmap.

III - The B2B Wall in Pedal Generation

Chainless pedal generation, where the rider’s effort produces electricity rather than turning a chain, is not exotic. It exists, it works, and it is in production.

We found three suppliers. All three sell business-to-business only, with either minimum order quantities in the hundreds, lead times measured in months or devkit that cost more than a small car.

The part exists. It works. It is simply not for sale to anyone ordering fewer than a few hundred.

This is the ordinary condition of the category The component is technically mature and commercially unreachable at the same time, and the gap between those two states is precisely where small teams, research groups, and independent builders are stopped.

IV - Why Firmware Is the Highest-Leverage Gap

One gap is different from the others.

Field-oriented control hardware has become extremely cheap. Boards capable of driving kW-class motors are available for around 30 CHF. The silicon is not the bottleneck. What is missing is documentation and firmware: these boards are sold into adjacent markets, with thin datasheets and control software that does not anticipate the particular demands of an e-bike hub motor, which is high pole count, high inductance, low Kv, and hall-sensored. Sure we could also buy specialised controller but we fall back to the first issue or the price is a magniture order higher.

So the barrier to an open, instrumentable, kW-class drive is not cost, manufacturing, or supply chain. It is a few thousand lines of code that nobody has written down and published.

Firmware has a property that no mechanical part shares: it costs essentially nothing to replicate. One team writes it once; every team afterwards gets it for the price of a download.

We are not the first to see this. VESC has done exactly this for the electric skateboard and light EV world, and it remains the reference for what open motor control should look like. Our work is narrower: adapting this approach to the cheapest available FOC hardware and to e-bike hub motors specifically, a combination that VESC’s hardware ecosystem does not target and that its tuning defaults do not suit.

V - Building the Layer Instead of the Exception

There is an apparent contradiction between the last chapter and this one. Making It Viable argued that bespoke components destroy economies of scale and that custom parts are a disqualifier. This chapter argues that we had to build our own drive electronics, controls, steering, and pedal generator.

Both are true, and they resolve like this:

“Use commodity parts” is only an instruction if a commodity layer exists. Where it does not, the principle does not disappear. It moves down one level, and becomes: build the missing part our of commodity part so that it can also become commodity.

This distinction has consequences for how we work. A part built as a one-off for our prototype and a part built to become someone else’s commodity are not the same object, even if they do the same job. The second one has to be documented, has to use parts that are still purchasable in five years, has to be reproducible by a team that will never speak to us, and has to be published under a licence that permits all of that.

That is the standard every module in this project is held to. Each one exists because buying it turned out to be blocked by one of the four barriers above, and each one is documented so that the next team runs into the barrier once rather than repeatedly.

The parts catalogue for this category is not missing so much as closed, priced for volumes we do not have, or shipped in a state that still needs weeks of work. We are writing the open pages of it.

See the modules →

← Chapter 6: Making It Viable · All chapters · Build With Us →