The Carvalho Boom and the Stages of Quadcopter Flight

The results of a recent herbicide deposition study performed with the DJI T100 led us to observe that after ~13 m/s, swath width and drift were no longer directly related to travel speed; They appeared unaffected. This was completely unexpected as it was counter to several years of prior study with smaller drones. This led to a hypothesis that this new, faster generation of quadcopter-style drone was beginning to behave like a helicopter during high-speed flight, and that this was having an affect on spray deposition.

Let’s use the stages of quadcopter flight to set up the premise.

1. Hover

When a drone hovers, each rotor draws air from above and accelerates it downward in a high-velocity blast. The vertical component is called the “downwash” and the turbulent splash of air that hits the ground and spreads laterally is the “outwash”.

The initial strength of the downwash depends on the degree of “disc loading” which is the weight of the drone divided by the rotor area. The intensity of the downwash wanes with distance from the rotor, spreading out in three dimensions until it impacts the ground and becomes the outwash.

During hover, the drone recycles some of its downwash. This turbulence affects the stability of the drone, requiring a great deal of power, especially when it’s full.

2. Low-speed flight

A helicopter achieves forward thrust by changing the pitch of its rotor blades. Most drones have fixed-pitch rotors, so the entire drone must tilt forward to enter low-speed flight. This causes the column of downwash to tilt backward.

While the downwash is created by lift, “wake turbulence” is created at the tips of the rotors as high-pressure air beneath the rotor wraps around to the low-pressure area above. Given there are multiple rotors, downwashes and wakes interfere with one another. As the drone flies at low speed (~<3 m/s) the merged rotor wake is visualized by the pair of counter-rotating, cylindrical vortices that trail behind.

Observing spray droplet behaviour suggests an interaction between downwash and wake, the nature of which is likely a complicated continuum. There is an apparent separation of the spray, perhaps redistributing as a function of droplet size, with the finer spray drawn up into the vortices of the wake.

3. Effective Translational Lift (EFT)

As the drone accelerates it continues to angle forward to a maximum ~ 30°. At some point (~13 m/s?), it enters a state of “effective translational lift”, becoming more stable and more energy efficient. This is slower than is commonly reported for helicopter, and may have to do with disc loading, which is different for the two vehicles. At this point, the drone behaves more like a wing as it essentially outruns its downwash, moving undisturbed air over the rotors. This horizontal air provides some lift, making flight more energy efficient, at least until drag begins to pull on the drone.

The Possible Effect on Spray Behaviour

Droplets released beneath the drone at hover are captured by, and travel with, the column of downwash. They are driven into the ground and can even recirculate through the rotors. At low-speed flight, the downwash tips backwards. Even if some portion of the spray is released ahead of the downwash, most will be entrained and divided between the downwash and the wake. At some point, likely associated with EFT, more of the spray will escape ahead of the downwash as rotary atomizers fling droplets ahead of the extreme backward angle. This would only be likely if the atomizers were positioned at the fore of the aircraft.

Therefore, consider a horizontal boom positioned along the chord line ahead of the drone.  As the drone tips forward during flight, so too would the boom, bringing it closer to the ground and releasing droplets ahead of, and below, the leading edge of the downwash. And, if the droplets were large enough, perhaps most would fall, initially uninfluenced by either the downwash or wake, before being pushed down in a more uniform fashion.

Reception

In January 2026 I attended the 4th annual Drone End-User meeting in Kansas City. I presented this theory. Given that I am not an engineer, and I had limited data to support the idea, I appreciated their polite (albeit skeptical) response. I shopped the idea around the trade show floor where drone manufacturers suggested the obstacle avoidance would interfere, or the shifted centre of gravity would make the drone difficult to fly and to land. No one seemed interested. And then, Nino Carvalho introduced himself.

The Carvalho Boom

Nino Carvalho and his son, Emilio, own and operate NC Ag Spraying in the Central Valley of California, USA. Emilio was inspired by a mentor that flew a Huey helicopter in his business. In late 2025 they experimented with attaching a horizontal boom to a drone. With their permission, I’ve chosen to call this innovation “The Carvalho Boom”.

Their first attempt was with a DJI T50, but the boom mount interfered with the stacked rotors, and the atomizer cables were difficult to extend. The XAG P150 had fewer cables and only top-mounted rotors, so it was a better fit. After experimenting with various materials (PVC was too flimsy, steel too heavy) they mounted a length of ½ inch metal conduit directly under the drone.

In California, aircraft booms must be limited to 90% of the rotor width (because of rotor tip vortices). The greatest span of the rotors was 312 cm (122.8 in), so they made the boom 275 cm (~9 ft) long. They spaced the rotary atomizers evenly along the boom every 69 cm (~ 2 ft 3 in), extended the original 30.5 cm (12 in) nozzle cables to 305 cm (10 ft) to reach their respective electronic speed controllers, and plumbed them with 1.25 cm (0.5 in) diameter tubing.

They flew this first prototype over water sensitive papers. Dropping from a 3 m (10 ft) altitude to 2 m (6.6 ft) improved coverage uniformity and resulted in a 10.3 m (34 ft) effective swath width. They could see the downwash was interfering with deposition, and while increasing to a larger droplet size helped, it didn’t help enough. Then they made some design changes, extending the boom 30.5 cm (12 in) beyond the rotors, and they saw they had something. They reached out to Agri-Spray Consulting (Nebraska) and arranged to run a series of Operation S.A.F.E. fly-ins.

There were multiple flights that day, so we’ll focus on three specific load-outs. The critical parameters are listed in the following table in the order that they flew them. The first load-out (N7696-01) was deemed the best, and was the only one with the boom extended beyond the rotor tips. The other two are included here for interest. N7696-03 attempted to shift the boom back under the drone for cosmetic reasons, but also for ease of transportation. N7696-04 was the same configuration as the last, but with coarser droplets in an attempt to battle the downwash. The fly-in report from N7696-01 is shown below, but all three reports can be downloaded by clicking the links above.

There is great promise, here. Observers said it looked like the swath was rolled with a paintbrush and there were no observable vortices – just a sheet of spray. The following videos show some of the passes from that day. You can see vortices, but only when the boom is positioned beneath the rotors and not when it’s extended our front. A 10% CV is spectacular, and the profile of each pass (before averaging) was far flatter than any drone deposition I’ve seen previously. This design has not yet been used for custom application because there are still questions about how drone speed and pump flow will affect the swath. But, the Carvalhos are already discussing the next design, constructed with carbon fibre tubes.

Impacts and Musings

So, perhaps the theory of how the air is moving over the drone is correct, or perhaps it isn’t. I later heard that Dr. Fernando Kassis Carvalho (AgroEfetiva, Sao Paulo, Brazil) observed a similar phenomenon where swath width no longer changed after 13 or 14 m/s (personal communication). So, whatever the reason, the impact seems clear. Does this mean we’ll see a new generation of quadcopters with front mounted booms? That would be interesting, as it revisits the very early drone designs that employed conventional hydraulic nozzles on horizontal booms (albeit in the wrong location).

Should we also consider returning to hydraulic nozzles? Dr. Ulisses Antuniassi (Prof., Sao Paulo State University) studied a rotary atomizer from a DJI T40 and from a XAG P60 with WG fungicides, SL herbicides and either MSO or NIS adjuvants and found no trends in VMD, relative span or DV 0.1 Work by Dr. Steven Fredericks (Land O’Lakes) showed that the rotary atomizer from a DJI T40 created droplets roughly one ASABE category smaller than the software indicated. Common knowledge is that the XAG version produces a coarser spray quality than anticipated, and slow motion video produced by Mark Ledebuhr (Application Insight LLC) and Dr. Michael Reinke (Michigan State University) clearly showed the flooding issue reported by Dr. Andrew Hewitt (University of Queensland), where excessive flow to the nozzle interferes with its ability produce a uniform droplet size. And, I photographed no less than nine different rotary atomizer designs while at the End-User meeting. So, perhaps.

And what of kinematics? A drone’s acceleration time is calculated by dividing the change in velocity by the acceleration rate. We’ve seen at DJI T100 travel up to 100 m before it reaches target velocity. Admittedly, that’s when full and attempting to fly at high speed, but Kevin Falk (Corteva Agriscience) has related acceleration times between 25 and 50 m for 40 L drones. What happens to the spray from a quadcopter drone with a front mounted boom as it transitions through the stages of flight? Likely nothing good. Multiple headlands would be an impractical solution. Perhaps it’s sufficient reason for drones to start flying racetrack flight patterns like planes and helicopters, where they achieve high speed before passing over the target area. Current software does not allow that practice.

It will be fascinating to see the what the next generation of drone design will bring.

Acknowledgements

Thanks to Mark Ledebuhr (Application Insight LLC), Dr. Michael Reinke (Michigan State University), Kevin Falk (Corteva Agriscience), Dr. Tom Wolf (Application Research & Training), Adrian Rivard (Drone Spray Canada), Adam Pfeffer (Bayer Crop Science) for insightful discussion surrounding this hypothesis.

Special thanks to Nino and Emilio Carvalho (NC Ag Spraying) for sharing their experience and practical approach to improving drone spray deposition.