EP90 & EP150 — Flight Operations

EP90 & EP150 — Flight Operations

RFD-OM-002 · v1.0 · released 23 August 2026. Owner and approver: Joni Sytsma, Chief Remote Pilot.

Evidence classes. [V] demonstrated in hardware or flight · [S] simulation · [E] estimate or calculation.

⚠️ This is not the tail-sitter manual, and almost none of it transfers. RFD-OM-001 covers the recovered VTOL tail-sitters — the EP250 and EP450 — which take off vertically, fly, and land vertically on the same sortie. The aircraft on this page are not recovered. Read this page before you touch one, even if you are current on the EP250.

---

1. What this covers

Aircraft	In scope?	Why
EP90	✅ Yes	90 mm tube, canards, grid surfaces, no recovery
EP150 canard variant	✅ Yes	150 mm tube, canards, no recovery
EP150 TVC variant	❌ No	Thrust-vectored and recovered — vertical landing, reusable. Follow the tail-sitter manual
EP250 / EP450	❌ No	Recovered tail-sitters — RFD-OM-001

The dividing line is recovery, not calibre. If the airframe lands vertically under its own control it is a tail-sitter and RFD-OM-001 applies. If it does not, this page applies.

---

2. The three things that make this class different

Read this section aloud to anyone flying one for the first time. Everything else in this document is procedure; this is expectation, and getting it wrong is how people get hurt or surprised.

2.1 One flight per airframe. It is not recovered

These aircraft have no vertical landing capability. There is no recovery mode, no parachute and no flare.

They are flown under control all the way to the ground — the aircraft arrives where you sent it, not wherever it happens to fall. What it does not do is arrive gently, or fly again afterwards.

That must be the expectation every single time you fly one. Not a risk to be managed down — the designed outcome. Plan the sortie as terminal, brief it as terminal, and place people and property accordingly. Expect to lose servos on arrival, which is why both aircraft are deliberately built on cheap servos: on an airframe you fly once, cost per flight beats component quality.

If a crew member is expecting to walk out and pick the aircraft up intact, they have not been briefed.

2.2 The launch is violent, and it will surprise you

Both aircraft have a great deal of thrust for their weight, so they take off vertically with no difficulty at all — and considerably more enthusiasm than anyone expects.

	Thrust-to-weight	Src
EP90	~1.5 – 1.6	[V]
EP150 at 750 g flown mass	2.4	[V]
EP150 at 500 g mass-optimised	3.6	[V]

At a T/W of 3.6 the aircraft will accelerate to somewhere frightening if the throttle is left where it started. Tell the operator what to expect before the first launch. It is a genuinely exciting procedure, and an operator who has been warned watches it; one who has not, flinches.

2.3 The operator is not in manual control, ever

This is the hard rule of the class, and it is a physical constraint rather than a policy.

The airframe is axisymmetric and has no wings. If it rolls, there is no visual reference that tells you which way up it is — no wing, no canopy, no asymmetry to read against the sky. Control inputs made without knowing the roll attitude are as likely to be wrong as right.

So:

• The aircraft flies under autopilot control, or it is lost. There is no third state.
• Manual recovery is not a skill that can be trained, and no operator should be assessed on it or expected to attempt it.
• Where an RC link is fitted (see §6) it exists for arming, mode selection and termination — not for steering.

Once it is under control, it is a very good aircraft: give it a target and it flies like a very fast rocket of an aeroplane. The whole design problem is getting it under control, not flying it once it is.

---

3. The flight profile

The mission is a target location set on the ground before launch, not a stick-flown sortie. The operator’s job is to set that target, verify the aircraft is healthy, and launch.

1. Launch vertical at full throttle, held until a velocity gate.
2. At the gate, pitch over while ramping the throttle back — at T/W 3.6 it otherwise accelerates well past where you want it.
3. At a pitch-angle gate, hand off to fixed wing and let NPFG fly the waypoints to the target.
4. Run to the target. There is no return leg and no recovery phase.

The tube-launch sequence

A canard aircraft launched from a tube cannot have live control surfaces on the way out — the canards would bind against the tube wall. The sequence flown is:

1. Fixed thrust, typically 100% for 2 to 5 seconds. [V]
2. Control system inactive throughout, so the surfaces stay stowed and unloaded.
3. Control activates after a configurable delay.

So the launch is deliberately unguided and uncontrolled until the aircraft has enough speed to have aerodynamic control at all. At that point the autonomy system takes over, stabilises the aircraft, and flies the rest of the profile.

⚠️ Control activation is the highest-risk moment of the sortie. It is the instant the aircraft goes from an unguided body to a controlled one, and the stand-off distance must assume it may depart in an arbitrary direction at that moment. Brief it, and stand where that is survivable.

---

4. Power-up

The procedure is the same as the rest of the fleet. There is no power switch — connecting the battery is turning it on.

1. Undo the bayonet.
2. Find where the battery lead goes into the ESC pack.
3. Plug it in.
4. The aircraft wakes on connection, and acquires GPS on its own.
5. Wait for GPS lock before doing anything else.

⚠️ It is live from the moment it has power. Stay clear of the prop disc, keep the aircraft pointed somewhere harmless, and treat the launch tube as the safe place for a powered airframe.

---

5. Batteries

Both aircraft run 4S, and both will fly on either chemistry:

	Mass	Range	Notes
LiPo	Lighter	Baseline	The default build
Li-ion	Heavier	Considerably further	Buys endurance at the cost of launch mass and T/W

🔴 You have about two minutes. Plan the sortie around the battery

The EP150’s real pack is 4S 450 mAh [V] — not the 1.3 Ah the older modelling assumed — and that one number sets the whole sortie:

		Src
Pack	4S 450 mAh	[V]
Full-throttle draw	~70 A, which is 155C	[V]
Boost duration	2 to 5 seconds total	[V]
Cost of the boost	~9% of the pack at 2 s, ~22% at 5 s	[E]
Cruise draw at 38 m/s	~14 A	[E]
Usable flying	≈ 2 minutes	[E]

Size the circuit to the battery, not to the airframe. This is a one-circuit aircraft. Any plan that assumes loiter, a second pass or a diversion is not a plan this pack can fly.

The barrel is 100 mm × Ø65, so a larger pack is physically possible and is the most accessible route to more endurance — but nothing currently flies on one.

---

6. Getting at the aircraft

Every RFD aircraft can be reached several ways. Which of these a given airframe has depends on the variant — check the build before you plan around any of them.

Method	Where it connects	What it is for
USB	A hole in the side of the fuselage	Direct autopilot access — configuration, parameters, log download. Ground only
Trailing umbilical	A wire tail led out the aft end	Serial telemetry while the aircraft sits in the tube
Wi-Fi (onboard Raspberry Pi)	Wireless access point	Browser GCS from any tablet or phone. Setup only
RFD900	Internal radio	Telemetry at range, in flight
RC receiver tag	Servo-extender tag on the autopilot	Arming, mode selection, termination

6.1 The trailing umbilical

A wire tail led down to the aft end of the aircraft, in one of two forms depending on the variant:

• Three wires — RX, TX and ground. Serial telemetry.
• Four wires — RX, TX, ground and VBAT. The VBAT line floats the battery, so the aircraft stays awake and alive in the tube indefinitely.

⚠️ Floating the battery is not free. Holding a lithium polymer at full charge degrades it exactly as it would on the bench. An aircraft left on the umbilical for weeks is a battery-life decision, and it should be a deliberate one rather than the default.

6.2 Wi-Fi and the browser GCS

Some variants carry a Raspberry Pi, which brings up a wireless access point. Join it and, instead of the captive portal you would get at an airport, you land straight on our ground control station — the browser GCS, on whatever device you joined with.

In the field this is usually easy: there is rarely any competing Wi-Fi where these aircraft fly, so the access point is simple to find.

🔴 Wi-Fi is a setup link, not a flight link. Once the aircraft arms and flies away it is out of Wi-Fi range very quickly. Never plan a sortie around Wi-Fi telemetry. If you need to see the aircraft in flight, it needs the RFD900.

Operating the GCS itself is a separate subject and is not covered here.

6.3 RFD900

Some variants carry an RFD900.

Operator-supplied radios are supported — talk to us first. Fitting your own radio is a normal thing to ask for and we will work it through with you. It is not something to plan around unilaterally, because these are small aircraft and tightly constrained on space, mass and volume. Any radio comes out of that budget, and on the EP90 in particular the budget is thin.

6.4 RC receiver tag

A servo-extender tag exposes the autopilot’s RC input so a receiver can be plugged straight in.

• RFD flies Team BlackSheep radios on SBUS. That is what is fitted, tested and supported.
• A DSM tag can be supplied for customers. RFD holds no DSM radios, so it is a build option we can expose but cannot exercise on the bench — a customer taking it should plan to prove the link themselves.

Remember §2.3: the RC link is for arming, mode and termination. It is not a steering link.

---

7. Speed, and why this class flies fast

These are the speeds this class flies at.

A centre-of-gravity range will be published here. The useful form of it for an operator is a range rather than a single point: fly whatever payload suits you, as long as the aircraft balances inside the published range. Ask us for the current figures until they appear here.

	EP150 at 500 g	EP150 at 750 g	Src
Level top speed	61 m/s (220 km/h)	61 m/s (219 km/h)	[E]
Minimum trimmed speed	~32 m/s	~40 m/s	[E]
Best-L/D speed	25 m/s	31 m/s	[E]
Turn radius at 38 m/s	~200 m	—	[E]

What limits this aircraft at the low end is trim, not stall. It runs out of canard authority long before it runs out of lift.

Fast is where this aircraft is comfortable. Slow is where it runs out of control authority. Somewhere upwards of 50 m/s it has all the lifting surface it will ever need.

It arrives pre-optimised and configured by us for fast. That is a deliberate choice rather than a compromise — and if slow is what your mission needs, we will give you a different aircraft rather than detune this one.

⚠️ The turn radius is an operational constraint, not a footnote. At 38 m/s the aircraft needs about 200 m to come round. A geofence, a survey area or a range boundary smaller than the vehicle’s own turn geometry is not a safety feature, it is a launch inhibit — the aircraft will refuse the mission outright rather than fly a circuit it cannot complete.

⚠️ Cruise speed costs range steeply. Best L/D is 2.51 and cannot be improved by adding area — induced drag depends on span, and span is fixed at one calibre by the tube.

That is the trade this class exists to make, not a shortcoming. Range is not what this aircraft is for. If range is the requirement, RFD builds other aircraft for it — see RFD-OM-001, and talk to us.

---

8. After the flight

There is no recovery procedure, because there is no recovery. What there is:

1. Wait. Do not approach until the aircraft has come to rest and any pack is confirmed undamaged. A disturbed lithium pack is the hazard, not the airframe.
2. Recover the pieces. Expect broken servos; expect the canards to be the first thing gone.
3. Get the log off it. The SD card and the flight log are the whole point of the sortie — they are the only evidence of what the aircraft actually did. See SD card formatting if the card will not read.
4. Sort the wreckage. RFD’s historical practice, and it is a good one: keep every part that is not broken and build it into the next airframe.
5. Write the sortie up on the day it flew, while you still remember it.

---

9. Other calibres

The design and manufacturing process is flexible, and the fuselage is shared: the EP90 and the EP150 are the same 65 mm body, 500 mm long, differing by the propulsion module and the tube they launch from. Another calibre is a module change rather than a new aircraft.

This applies upward as well as downward. Because thrust vectoring adds both weight and a little aerodynamic drag, an EP250 built this way — canard-steered, not vectored — actually does a little better than the TVC EP250 on the flight performance that matters here. What it gives up is recovery, which is the whole trade of this class.

If a customer wants a calibre RFD does not currently build, that is a conversation worth having rather than a refusal.

---

10. Related

• EP90 · EP150 — the aircraft pages, and the source of every figure here
• EP90 & EP150 — Training — RFD-TM-002
• Document control — document numbering and issue status
• Fleet Roadmap
• RFD-OM-001 — the recovered tail-sitters. Not this class