Regen is limited by the battery, not the motor
Almost every discussion of regenerative braking on light electric vehicles focuses on motors and controllers. In practice the binding constraint is what the pack will accept — and that constraint moves with state of charge and temperature.
Regenerative braking sounds like a motor feature. Electrically it is a charging feature: the motor becomes a generator and pushes current back toward the battery. Everything that normally governs charging a lithium pack — current limits, voltage ceilings, temperature windows — applies, except now it is happening while the vehicle is moving and the rider is asking for braking force. Five constraints stack up.
The motor has to be able to do it at all
Most geared hub motors contain a freewheel clutch so the motor does not spin when you coast. That clutch mechanically prevents the wheel from back-driving the motor, which means no regen is possible regardless of what the controller or battery would allow. Direct-drive hub motors — and geared motors specifically built without a freewheel — are the ones that can generate. This single mechanical detail is why regen is rare on light electric vehicles, and it is decided long before any of the electrical questions matter.
State of charge sets the ceiling
Regen pushes current into the pack, which is charging. A pack near full has almost no headroom to accept it: push current into a cell that is already at its upper voltage limit and you are overcharging it. So the allowable regen current is not a fixed number — it is a function of state of charge, falling toward zero as the pack fills. The practical consequence is the one riders notice first: regen is weakest at the top of a hill you just climbed with a full battery, which is exactly where you most want it.
Temperature is a harder limit than most people expect
Lithium-ion cells will deliver current well below freezing but must not be charged below roughly 0 °C — doing so plates metallic lithium onto the anode, which is permanent damage and a safety problem, not just wear. Any regen implementation therefore needs a temperature window that inhibits energy return when the cells are too cold, and the upper bound matters too. The window in which a pack may accept charge is meaningfully narrower than the window in which the vehicle can be ridden.
The BMS will act, and it is not subtle
If regen current exceeds what the pack BMS permits, the BMS protects the cells by opening the charge path. That is the correct behaviour, but from the vehicle side it appears as a protection event in the middle of braking. Designing regen means staying inside the pack’s limits by design rather than discovering them through cutouts — which requires knowing what those limits actually are, in writing, from the pack supplier.
With two packs, you also have to choose a destination
Everything above describes one pack. On a vehicle with two, the returning current has to go somewhere specific, and the two packs are rarely in the same state. A pack at 95 percent can accept very little; a pack at 40 percent can accept a lot. Sending braking energy indiscriminately at both means the limiting pack sets the ceiling for the whole system — or worse, that the packs are asked to equalise through a path never designed to carry charge current.
A note on how much range regen actually returns
Less than the marketing implies. A light electric vehicle and its rider carry a small fraction of a car’s kinetic energy, and a large share of any stop is absorbed by rolling resistance and aerodynamic drag before the brakes do anything. Recovery is real but modest, and it is highly dependent on terrain — a hilly commute recovers meaningfully more than a flat one.
The most credible published figures come from Grin Technologies, who sell regen hardware and still publish the conservative number: roughly 3–8% added range. Independent instrumented rides line up with that — around 5% on flat urban routes, closer to 1% on genuinely flat terrain with few stops, 8–12% on hilly city riding, and 12–18% on a sustained mountain descent.
You will also see much larger numbers quoted, and they are usually answering a different question. A single hard stop can return something like half of that stop’s kinetic energy. That is a real measurement, but it describes one braking event, not a trip — and quoting it as a range benefit is how a 5% feature becomes a “50%” claim. Recovery also scales with mass, so figures measured on an e-bike overstate a standing scooter by roughly three times.
The more honest arguments for regen on a light vehicle are brake-pad wear, controlled speed on long descents, and one-pedal-style ride feel. Riders running regen hubs report going from replacing pads every few months to barely touching them, which on a loaded cargo bike is a maintenance line item rather than a nicety. If a supplier quotes you a large range figure without naming the terrain and the state of charge it was measured at, treat the number as decoration.
One thing regen never is, anywhere, is a brake in the regulatory sense. The international rules define regenerative braking as explicitly not part of the service braking system. Mechanical brakes stay, always, and anyone implying otherwise is selling you a compliance problem.
Two packs and regen: the real tradeoff
There is a widely repeated claim that you cannot have two packs and regenerative braking at the same time. That is not true, and it is worth being precise about where it comes from.
The cheap aftermarket combiner modules genuinely do block regen, and their vendors say so — product pages carry warnings like “this does not work for bikes with regenerative braking” and “not compatible, do not use.” They are being straight with their customers. Those modules use a single switching element per pack that conducts one way only, so reverse current is blocked by construction. Our current shipping board shares that property.
But that is a property of one product class, not of paralleling. Plenty of production vehicles run two packs and recover energy under braking. They do it by connecting the packs directly, with no blocking element at all, and letting each pack’s own protection limit how much charge current it accepts. At least one manufacturer documents exactly this, describing recuperation current distributed across both packs and capped by the battery management system. The technique is standard in larger systems too, where packs are joined by contactors — which conduct in both directions once closed.
So the honest framing is a tradeoff rather than an impossibility:
- Connect the packs directly and regen works fine — but the packs must be closely matched, because nothing is arbitrating between them. Mismatch is now your problem, and the rider’s.
- Put a one-way element between them and mismatch stops mattering — but you have given up regen to get there.
Getting both at once is not a physics problem, but it is more than a component swap, and the reason is worth stating precisely. A one-way element blocks reverse current because that is its whole function. A switch that conducts both ways solves that — but a two-way switch left closed just hard-parallels the packs again, so you are back to cross-charging whenever their voltages differ. Neither part on its own is the answer.
What the job actually needs is a two-way path plus something deciding when it may be open — matching voltages before connecting, and separating the packs again when they diverge. Bidirectional switching parts exist at this voltage and are well documented; chip vendors even publish reference designs that name preventing reverse current during motor braking as a design goal. What none of them supply is the arbitration on top. That is the part we are building, and it is why this is a redesign rather than a firmware change: the return path, the reverse blocking between charger and regen source, and the rule for which pack may accept energy all have to be designed in deliberately.
What we are working on — in design review
Regen into a two-pack system
Our production power management board combines two packs onto one traction bus. The revision we are designing adds a controlled return path so braking energy goes back through the packs’ charge ports — reverse-blocked from the charger, current-limited per branch, with a temperature window and a high-state-of-charge foldback so a full pack is never pushed past its ceiling.
It is an engineering concept under active review, not a released product, and the first pilot deliberately caps total regen current well below what the packs could theoretically take while the protection scheme is validated. We would rather describe it accurately than sell it early. If you are working the same problem, we are interested in design partners.