MG1 rotor on right side (MG2 rotor instead?) Typo maybe ?
Whatever, very interesting.
GOBAO ECVT EMTB Mid-Drive Motor
- Thread starter slickrock
- Start date
Whatever, very interesting.
ArmyOfEvilRobots
New Member
- Mar 6, 2026
- 32
- 45
Good catch, I didn't notice when I posted the image. Yes, the MG1 rotor is on the left, MG2 rotor on the right, and whoever made the diagram did a bad copy-pasta.
Best diagrams, to my opinion, to be found in the e2drives patent applications.
Yes good find, this is a diagram of a coaxial drive like in the 1st to 3rd generation Toyota HSD transaxle.
The 4th and latest 5th generations are parallel layout, I think @Ndanger posted a diagram of one. The Gobao and Avinox MG are most probably parallel layout drives as those are newer and have a few advantages.
Coaxial vs. Parallel Axis Drive
Toyota shifted away from the strictly coaxial layout to save space, reduce weight, and improve mechanical efficiency:
Generations 1 through 3 (Coaxial): The gasoline engine, the smaller motor-generator (MG1), and the main traction motor (MG2) were all arranged inline on a single, long central axis. While elegant, this created a very wide transaxle that was difficult to package into smaller engine bays.
Generations 4 and 5 (Parallel Axis): Toyota completely redesigned the transaxle. They moved MG2 onto its own separate, parallel shaft adjacent to the main engine/MG1 shaft.
Reasons for the Parallel Axis Drive:
Reduced Width: Moving MG2 to a parallel axis allowed the transaxle to be significantly shorter end-to-end, making it much easier to fit. (Probably also better for ebikes to have a narrow crankset Q factor).
Less Friction: The parallel layout allowed Toyota to replace the heavy, multi-layered planetary reduction gears previously tied to MG2 with a simpler, lighter parallel reduction gear, cutting mechanical power losses.
In 2019 Toyota released approximately 23,740 patents awarded over more than 20 years of electrified vehicle technology development for free use(that is a lot of R&D). A boom of hybridization of cars in different forms in the automotive world started right after the Covid time passed.
I am pretty certain that these two ecvt drives and some of the existing ones floating on the market are scaled down ecvt drives based on fundamentals from Toyota's patents. The only main difference being is replacing the ICE with human leg input power.
That's a significant difference in power so the down scaling was the real R&D for Avinox.
When asking AI to scale down the Toyota e-cvt to emtb middrive level and substitute the ICE with human power this is what it threw out: If anyone is interested, there are some interesting other things to read...
Power Ratio Scaling:
In an eCVT car, the power ratio between MG1 and MG2 is roughly 1:2. In an eMTB, because human legs provide the entry torque and space is highly restricted, the ratio widens to about 1:4 in favor of MG2.
• MG2 (The Drive Motor): 250W to 500W rated with peaks up to 600W-900W). It provides the raw assistance power.
• MG1 (The Control Motor/Generator): 60W ta 150W rated. It does not propel the bike directly, instead it's main purpose is applying "reaction torque" to the planetary gearset to infinitely vary the gear ratio.
The Lever Analogy & Torque Split Math:
To understand how the eCVT controller commands MG1. we have to look at the torque split mathematics and the control loops governing the planetary gearset.
In a planetary gear system, torque cannot be applied to one component without a reaction torque balancing it on the others. The system behaves like a mechanical lever where the Planet Carrier (your legs) acts as the main fulcrum in the middle, balancing the Sun Gear (MG1) on one side and the Ring Gear (MG2/front chainring) on the other.
The Fundamental Torque Equation
The physical constraint of a planetary gearset dictates that torque splits based on the number of teeth on the Ring Gear(R) and the Sun Gear(S). In a compact bicycle gearbox, R should have about twice as many teeth as S. The static torque distribution should look something like this: MG1 always experiences cca 1/3 of your leg torque, while the Ring Gear (output) receives cca 2/3 of vour leg torque. This relationship is fixed by the metal gears(teeth number) and cannot be changed by software.
The Controller's Computational Logic:
The eMTBs controller processor checks sensor data (cadence, rider torque, wheel speed, and incline) to decide how to run MG1.
Scenario A: Simulating a Low Gear (Climbing a Steep Hill):
You hit a steep wall on a trail. Your leg cadence drops to a sluggish, knee-straining 40 RPM, but you want to spin comfortably at 80 RPM. To allow your legs to spin faster while the wheel is moving slower, the controller commands(this depends on the mode you are using Auto, Manual Shift,....) the internal inverter to spin MG1 forward in the same direction as your legs.
The Result: MG1 acts as a motor, consuming electricity. This relieves the resistance on the planet carrier, acting like a tiny mechanical lever that lets your legs spin up to an effortless 80 RPM while the bike crawls up the hill.
Scenario B: Simulating a High Gear (Decent speed on flat ground or flying down a fire road):
The wheel speed is high, and your legs are spinning out at 100 RPM and you want to feel more resistance so you can push the bike faster. To reduce the cadence relative to a very fast-spinning wheel the inverter applies an electrical load(regen) to slow MG1 down or actively spin it backward.
The Result: MG1 resists the movement and acts as a generator. It absorbs the excess kinetic energy from your legs, turns it into electricity and either shunts that power directly into MG2 to help push the wheel, or sends it back ta charge the battery. Your pedals stiffen up mimicking a hard, high-speed gear.
The "Dead Battery" Dilemma:
In Toyota's system, if MG1 has no power, the car cannot drive because the engine's power just spins the unresisted sun gear into infinity. On an eMTB, a dead battery means MG1 cannot provide reaction torque. Without a mechanical lock-out clutch to freeze MG1 when the battery dies, the rider would pedal completely in a void, unable to move the bike at all.
The Mechanical Safeguards for a Dead Battery situation:
To prevent riders from being stranded deep on a trail, engineers can use different mechanical failsafes:
The Electromagnetic Fail-Safe Lockout (Brake):
A small, spring-loaded electromagnetic brake is held open by the battery's residual power. The moment electrical power drops below a safety threshold, the magnet de-energizes. Strong mechanical springs slam a locking pin or clutch into MG1 (the sun gear), forcing it to freeze solid.
The "Default Fixed Gear" Mode:
With MG1 locked to the housing, the planetary gearset stops acting like a CVT and locks into a single, permanent mechanical gear ratio (usually a low, easy-to-pedal climbing ratio). Your feet are now directly, mechanically linked to the rear wheel, converting the eMTB into a heavy, single- speed analog mountain bike.
The BMS Buffer:
Modern Battery Management Systems (BMS) never let an e-bike drain to a literal 0%. When your display screen reads "0%", the system has shut down the power- hungry traction motor (MG2) to save a hidden 3-5% buffer. This remaining juice keeps the controller awake and keeps MG1 locked or positioned so you can limp back home safely.
Last edited:
Interesting. As for operation with a depleted battery, in reality that never happens; they can easily use a lockout system that engages below MG1's power threshold. The question might still arise in the event of an electrical failure, however.
So this tech is a scaled down hybrid car? Specifically the lowly Toyota Prius.? You better inform the marketing geniuses before this becomes public knowledge - maybe say F1 cars are hybrids these days etc
Interestingly F1 seamless shifting engages 2 gears at the same time for a fraction of a second. And here's one for you
E-Mountain Bike Battery
Energy Density: 140 to 180 Wh/kg
Formula One Battery (Energy Store)
Energy Density: 45 to 106 Wh/kg
(F1 priority is charging speed and discharge rate)
Remember the Zeroshift transmission 20 years ago? Was destined for road cars. Not quite the same tech as F1 uses, but both systems temporarily engages 2 gears meaning uninterrupted power. While the transmission worked, it created a big ask of the engine and driveshaft to match the new rpm. At around the same time, DCT came along, end of story (at least for road car usage).
Regarding the "Dead battery" dilemma, on the OWURU drive unit, there is a freewheel between the pedal axle and the chainring to allow a direct transmission of power when the motor(s) are off.
Source: US20200198727A1 - Power train - Google Patents
Source: US20200198727A1 - Power train - Google Patents
De$lo(
Member
So it becomes a single speed in the last gear you were in when the battery died?
Cant remember if I already added to this thread, but according to Gobao
"We stop giving power when the battery reaches 10%. The system then only shift. Without power the ratio goes to the lowest and you can theoretically pedal."
"We stop giving power when the battery reaches 10%. The system then only shift. Without power the ratio goes to the lowest and you can theoretically pedal."
No, you get a single speed with the ratio provided by the chainring and the gear at the rear wheel.
So nothing to climb steep uphills.
Yes you did (or maybe in another thread).
I've posted something similar about the city bike Owuru drive unit which is since 2023 on the market and lacking the freewheel.
I guess future Owuru drive units like the announced Owuru Ride for e-mtb and cargo bikes will have it.
Contrary to the Gobao solution, the purely mechanical solution provided by the freewheel still allows to pedal when the battery is fully drained. However without shifting.
And obviously it's still possible to combine the software and the mechanical solutions.
Red Peekay
Active member
"Theoretically" that is too funny.... so even Gobao question it
I would imagine it would't be a pleasant experience unless you were at the top and going down to get home!
Can I add to this, that whilst pedalling with no power in an E-CVT, the pedals will drive the smaller motor, MG2. This will generate some power so that you could technically vary the gear ratio, by slowing MG2.
Whether the MGU manufacturer allows you to use power generated by MG2 to vary the gear ratio, when the battery is flat, will be in the software design.
Remember. When MG2 is stationary. The gear ratio is about in the middle.
I like the concept of regen, has there been confirmed reports that the Gabao does indeed have the capabilities. I would like to suggest that regen may be a reach for a flat bat. First there would need to be enough current to power the logic boards and electronics.
Additionally concider an ecvt without electrical power, the system possibly lacks the necessary reaction torque to transfer your pedal power to the wheels, meaning your pedaling energy would simply spin the unconstrained motor instead of generating electricity or moving the bike forward...
I do hope that regen is a thing.
Additionally concider an ecvt without electrical power, the system possibly lacks the necessary reaction torque to transfer your pedal power to the wheels, meaning your pedaling energy would simply spin the unconstrained motor instead of generating electricity or moving the bike forward...
I do hope that regen is a thing.
I think it would be far more useful to focus on developing batteries with better wh/kg metrics than on regen. With better batterie and the superfast charging I don't really see much point in complication things trying to shoehorn regen in emtbs.
2 different teams of engineers (probably in different companies) that can't do each others job, so this will happen in parallel if we like it or not
I see regen as a net positive. We get some additional charging. Speed is controlled on decents (grandma could set it to come on as low as 20 kph). No freehub in the rear wheel will allow for reverse gear for cargo bikes. No tensioner. But the bike won't coast as well.
Why would anyone want that on a performance emtb? I don't get it. I don't want a motor to control my braking. All I want my motor to do is give me assistance.2 different teams of engineers (probably in different companies) that can't do each others job, so this will happen in parallel if we like it or not
I see regen as a net positive. We get some additional charging. Speed is controlled on decents (grandma could set it to come on as low as 20 kph). No freehub in the rear wheel will allow for reverse gear for cargo bikes. No tensioner. But the bike won't coast as well.
You can run a smaller battery. Drop some weight.
Embt is a niche in the large ebike space, and I agree that regen may be better suited for other applications. All the way from adventure/SUV type bikes to cargo.
I wish for occasional break, from braking on the long non technical sections on the way home. Secondly regen to help with thermal issues . Ultimately mild regen with controlled manual braking to suit the circumstances.
so the regen is in the mid drive ecvt motor ?
so one of the motors would turn the opposite direction - with a clutch to disengage the pedals?
once the (presumably electronic) brake levers are pulled?
how would that work in relation to actually slowing the bike down?
wouldn't there need to be a relationship to at least one of the wheels?
am I getting it wrong?
so one of the motors would turn the opposite direction - with a clutch to disengage the pedals?
once the (presumably electronic) brake levers are pulled?
how would that work in relation to actually slowing the bike down?
wouldn't there need to be a relationship to at least one of the wheels?
am I getting it wrong?
Last edited:
how you manage the variable length of the belt with this function of "brake" ? For the traction you have an belt tensioner. But if you use the motor for brake (or generator), you don't think that will have an problem
?In an eCVT drive there is no clutch, the pedals are most probably separated from the chainring. Pedals driving the planetary carrier inside the drive, the drive motor driving the chainring. They are so to say on separate axles.
thanks - Gotcha - what I'm not getting is how a rotational force is transmitted to one of the mid motors when braking
to effectively turn one of the mid motors into a dynamo and give the inductance required for charging
Braking Regen is normally handled by the large motor. The pedals can be held stationary just by the weight of your feet, whilst the output gear spins, just due to planetary gear configuration. There is no clutch required.
The output gear is directly coupled to the large electric motor through the ring gear. So when the output gear is being driven by your rear wheel under inertia, and you take the electrical output of the motor and connect it to a load, like a charging circuit. It will create a breaking force in the motor. This braking force transfers to the ring gear, to the output gear, to the rear wheel.
Where if you just leave the motors electrical output, open circuit. The motor will not brake, but free wheel, providing no braking force, so you'll just coast.
Last edited:
thanks - so the chain ring (ring gear?) would be driven via chain or belt from the rear wheel's inertia (whilst the pedals are disengaged) direct to the output gear - which in turn would force the motor to act as a dynamo
but at the same time the motor creates a braking force that is transmitted back to the rear wheel via the output gear via the belt.
I'm partly there... but maybe I need some diagrams
Similar Threads
-
Pinion pulls the plug on EMTB motors [mod edit- Fake News]- Started by BikeBert
- Replies: 19
-
Current Avinox MG Prototype Bikes Don't Support Battery Regen- Started by slickrock
- Replies: 94
-
-
-
