"Shave one kilogram off the reducer, and the helicopter can carry two extra kilograms of payload. Rotor engineers around the world all know the math. But for forty years, few have dared to turn in their answer."

You may never have thought about what the deadliest vulnerability of a helicopter is.

It's not the engine. It's not the rotor. It's something hidden deep inside the fuselage, something you never see—the main gearbox.

Engine speed is typically above 20,000 rpm. The rotor only needs two to three hundred rpm. That fifty-fold or even hundred-fold gap is borne entirely by a single gearbox. That gearbox is the main gearbox, often the single heaviest component in the entire helicopter, typically accounting for 8% to 10% of the total weight.

For a 10-ton class helicopter, it's not uncommon for the main gearbox to weigh nearly one ton.

More critically, this ton is not "dead weight." It must withstand the engine's full power output, reduce the speed, amplify the torque, and ensure the rotor turns smoothly and evenly. If any critical component inside the gearbox fails, the consequence could be rotor stall.

If the rotor stops, the helicopter falls from the sky.

The harsh reality of such a system is: in engineering, you can build in redundancy, isolation, and monitoring, but for the main transmission chain, the margin for "degraded operation" is extremely limited. There's no "try restarting" option.

That's why, over the decades, the overall configuration of helicopter main gearboxes has rarely seen radical overhauls—not because there are no new ideas, but because any "lighter" solution must simultaneously meet even more stringent requirements for reliability and manufacturability. What you have to convince is not just the review board, but the history built from accumulated flight hours.

Traditional solutions are well-established: spiral bevel gears handle the directional change, combined with planetary gears for step-by-step speed reduction. Three or four stages are common. Each stage means more gears, bearings, housings, lubrication systems, and structural margins – and therefore more weight. But the advantages are equally clear: abundant data, well-defined boundaries, and manageable risk – mature enough that even every bolt has decades of operational experience backing it up.

Would you dare to stuff a new configuration lacking the same "flight record" into a main transmission?

Most people wouldn't.

Face gears are precisely that candidate – everyone is watching them, but engineering demands extreme caution.

Yet what truly turned face gears into a star in the helicopter community is not their ability to turn a corner.

Face gears themselves are not new. They look like a flat disc, with teeth on the end face; when meshing with a standard cylindrical gear, the axes intersect. Simply put, they are naturally good at handling "turning" transmission paths: in some layouts, the directional change that would otherwise require additional bevel gears can be replaced or simplified by a more compact configuration.

But the real reason face gears keep coming up in helicopter circles isn't just "turning."

It's that they are well-suited for splitting power.

Splitting one power flow into two or even multiple paths, transmitting them through different routes, and then recombining them at the output. This is called split-torque transmission.

Most conventional gearboxes transmit power more like a "single-line series": A drives B, B drives C, C drives D. Each stage must handle the full torque, so each stage has to be built stout and heavy.

The split-torque approach is more like "parallel": after A, the power splits into two paths, each carrying only half the torque, then recombines before D. In theory, the gears and bearings on each path only need to withstand lower loads, so the parts can be smaller and lighter, and the housing and supports can also be more compact.

The principle is simple.

The difficulty is brutal.

But the beauty of face gears is that, in certain configurations, they have the potential to simultaneously accomplish directional change, transmission, and power splitting/combining in fewer stages. A conventional design might require three or four stages to achieve the reduction and power distribution that a face-gear split-torque system could compress into one or two stages under specific design targets, creating significant room for system-level weight and volume reduction.

In some publicly available research and engineering comparisons, results can be seen such as the following: on a specific platform and under specific design constraints, the introduction of face gears and reduction of stages has led to cases where the weight of the reducer system drops by as much as roughly 40%, while load capacity is also significantly improved.

It must be emphasized that such results represent configuration-comparison gains within the same platform, not universal conclusions that can be directly applied across different platforms.

But even if you treat this as just a "magnitude signal," it's still startling enough.

Because the main gearbox is already a nearly ton-class component. Achieving a system-level weight reduction on the order of "several hundred kilograms" is not a parameter optimization for a helicopter – it's a hard gain in mission capability, range, endurance, and payload. In many cases, it's the difference between being able to "carry an extra piece of equipment, extra fuel, or an extra person."

At this point, a seemingly obvious narrative emerges: Since the benefits are so substantial, why hasn't this become the default route for all types of helicopters?

Yet to this day, it hasn't seen large-scale adoption.

The answer lies not in the concept, but in the engineering details. Split-torque transmission has a fatal engineering problem: How do you guarantee that the torque split between the two paths is truly close to 50:50?

In theory, a symmetric design gives you 50:50. In reality, gears have manufacturing errors, shafts have assembly deviations, the housing elastically deforms, splines have backlash, and bearing and support stiffnesses are not perfectly uniform. Any deviation that "looks small" will push the load sharing off balance.

In 2002, NASA commissioned Boeing to build a 167 kW (approximately 224 horsepower, less than the power of a Honda Civic) proof-of-concept gearbox using face gears for split-torque. Test results: the split ratio of the two input pinions was 52:48. The two idler gears on the combining side showed a split of 57:43. At the experimental level, such deviations are "results," but in engineering terms, they are essentially "variables that must be designed and controlled."

52:48 – a difference of 4 percentage points. It doesn't sound like much.

But if you are the bearing taking 52%, your fatigue life will be shorter than that of the bearing on the other side.

If you are the idler gear carrying 57%, your tooth contact stress will be significantly higher than the design average – and in a system like a helicopter, where safety margins must be precisely calculated, a difference in life is a difference in safety margin.

On a helicopter, a difference in life is a difference in safety margin. And safety margin is never about the "average" – it is determined by the part that fails first.

The list of factors affecting load sharing is long: support stiffness, backlash, axial installation errors, spline backlash, structural elasticity… A team from the National Key Laboratory of Helicopter Transmission at Nanjing University of Aeronautics and Astronautics conducted systematic research and reached an interesting conclusion: the greater the support stiffness of the idler gears, the better, but for the input gears, the smaller the support stiffness, the better. Axial installation errors of the face gear cause the contact area to drift toward the tooth tip and inner end, directly worsening the meshing condition.

Translated into plain language:

It is not enough to simply manufacture each part. You have to tune the "softness/stiffness" and the "geometric positions" of the entire system into an extremely narrow window. A slight deviation, and load sharing goes off balance. Once load sharing goes off balance, some component will fail early.

Thus, what many public materials ultimately emphasize is not that "face gears can split torque" – everyone has long understood that principle. Rather, it's more like: Whether face-gear split-torque becomes an engineering reality depends on whether you can control shaft support stiffness, assembly errors, and structural elasticity to a level that keeps load sharing stable.

The concluding remark of that NASA report was not that "face gears can split torque" – everyone already knew that.

The conclusion was: the feasibility of face-gear split-torque depends on the precise selection of shaft support stiffness.

It is not a conceptual problem. It is a problem of precision.

Therefore, the application landscape of multi-path face-gear transmission is not as broad on paper as one might think. But where it truly fits, its value is overwhelming.

Tier one: helicopter main gearbox.

This is the starting point and final destination of all research. High power, extreme weight reduction requirements, naturally intersecting axes, and the need for multi-path redundancy—split-torque face gears are almost tailor-made. The validation work by NASA, Boeing, and DARPA all points here. Third-generation helicopter transmission systems are already seriously considering face-gear configurations.

Tier two: next generation rotorcraft.
Tiltrotors, coaxial rotors, compound thrust configurations – the transmission systems of these new architectures are more complex than those of conventional helicopters and have a stronger desire for compactness and light weight. Face gears offer a concentric layout: two face gears sleeved together face to face, with multiple pinions sandwiched between them. This "sandwich" structure is naturally suited for coaxial designs, enabling both power splitting and directional change in a very compact space.

Tier three: eVTOL and heavy UAVs.
The premise is that they ultimately still require high-power mechanical reduction rather than going fully distributed electric drive. If a central reducer is needed to distribute power from a turboshaft engine to multiple rotors, the logic of face-gear split-torque applies.

Exploration zone: marine and wind power.
Multi-path power splitting is not new in these fields, but "face-gear" split-torque is not a mainstream industrial route. In scenarios where you don't have to fight for every gram, the maturity advantage of conventional planetary gears is almost impossible for face gears to overturn.

A clear criterion: Is your application weight-sensitive? If the answer is "every kilogram matters" – for example, in aircraft – then face-gear split-torque is a serious candidate. If the answer is "a bit more weight doesn't matter" – for example, in mining reducers – stick with mature solutions and sleep well.

Stretching the timeline a bit longer makes it easier to understand why this has been "slow."

Looking back at this point:
The concept of face-gear split-torque has been seriously proposed for over thirty years, since the 1990s. The "tempting magnitude" of system-level weight reduction is not lacking.
Yet to this day, it has not replaced conventional solutions on a large scale.
Not because it is bad. It is because being "good" is a necessary condition in engineering, but never a sufficient one.

You also need to prove that, in mass production, the tooth geometry is consistent, load sharing is controllable after assembly, the state under thermal load coupling is predictable, wear and misalignment remain manageable after long term operation, a maintenance system can be established, and life data can be accumulated. In other words, you not only need it to perform beautifully on the test rig, but also to keep working “as you promised” after 10,000 hours.

From research paper to fielded model, from laboratory to flight envelope – the gap is not a few SCI papers.

It is the precision discipline of the entire manufacturing chain.

On the drawing board, all face gears are perfectly 50:50.

What truly gets them into the air is not the drawing, but an obsession with the third decimal place.

And to instill that obsession into complex systems such as face gears and multi path transmissions is, in essence, to turn “error” into “a variable that can be designed and controlled.” It is a contest of a complete face gear precision system.

With this focus, LASER GEAR is continuously advancing its related design and manufacturing capabilities – from tooth-profile machining to assembly control – gradually turning “feasible” into “reproducible.”