Why does my quad flip on takeoff?

The symptom

You raise the throttle. The aircraft does not lift. It snaps over onto its back, or onto an arm, in the time it takes you to think "no" — and it does it fast, faster than any tuning oscillation, faster than you can react. Props are usually destroyed. Sometimes arms are too.

This is the flip of death, and the important thing to understand is what it is not. It is not a tune. It is not a filter. It is not a gain you can lower. A control loop that is merely badly tuned wobbles, oscillates, or wanders — it does not invert the aircraft in 200 milliseconds. A flip that instant means the flight controller is applying correction in the wrong direction, so its own correction drives the error harder, and the loop runs away in the only direction it can. Something in the chain from "which way am I tilting" to "which motor do I speed up" is wired backwards.

The good news is that the list of things that can be backwards is very short.

Check this in 60 seconds

Take the props off. All four. Do it now, before you read the rest of this, because everything below involves spinning motors and you are about to deliberately provoke the fault.

Then, in order — and the order matters:

1. Which way did it flip? Write it down before you forget. Consistently backwards? Consistently to the left? Diagonally over one arm? This is your single best piece of evidence and it is the one everybody throws away.
2. Spin each motor individually from the Configurator's Motors tab. Not with the sticks — from the tab, one at a time. It is the same bench check that belongs in everything you do before the first power-up. Confirm that the motor the firmware calls 1 is the motor you think is 1, and that it turns the direction the layout diagram says it should.
3. Look at the props you just took off. Are they diagonally paired — the two on one diagonal one handedness, the two on the other diagonal the opposite? Was every one of them mounted the right way up?

If step 2 or step 3 surprises you, stop. You have found it, and nothing further in this article applies.

The ranked causes

Flip pattern	Most likely cause
Instant, violent, always the same direction	Motor output swapped or motor direction reversed
Aircraft strains, gets light, then tips over slowly	A prop mounted upside down, or the wrong prop on a corner
Flip is instant and clean about one axis, aircraft is otherwise healthy	Board orientation never told to the FC
Aircraft levers over an arm; one corner never came up	A dead or stalled ESC / motor
Aircraft tips only when you creep the throttle up	Not a fault — see the last section

1. Motor order or motor direction is wrong

This is the overwhelming favourite and you should assume it until you have proved otherwise.

The flight controller has a mixer: an internal table that says "to pitch forward, slow the front two and speed the rear two." That table is written in terms of motor numbers. If the ESC signal wire you soldered into pad 1 goes to the motor that the mixer thinks is number 3, the FC will correct a forward tilt by speeding up a motor that makes the tilt worse. The loop then diverges instantly. Same aircraft, same tune, same firmware — one wire in the wrong pad.

A reversed motor direction produces the identical failure. A motor that turns the wrong way pushes air the wrong way — it makes downforce instead of thrust, and its reaction torque on the yaw axis is inverted too. To the mixer it looks exactly like a swapped output, because the corner responds to a throttle increase with less lift instead of more.

This is why the two must be checked separately and in sequence:

• Numbering first. In the Betaflight Configurator's Motors tab, spin motors one at a time and confirm each number is physically where the firmware's layout diagram puts it. If the numbering is wrong, fix it properly: re-solder, or remap the motor outputs in firmware. Do not "compensate" for a wrong map anywhere else.
• Direction second. With numbering confirmed, check each motor's rotation against the layout. A standard quad has the two motors on one diagonal turning one way and the two on the other diagonal turning the other. If a direction is wrong and you are running DShot, you can reverse that motor in software from the Configurator rather than swapping two of its three phase wires — that is what the direction control in the Motors tab is for. On analogue protocols, swap any two of the three motor wires.
• Prop handedness third. Only once numbering and direction are both confirmed do you decide which prop goes on which corner, because prop handedness is determined by the motor direction, not the other way round. Get this backwards and you will happily "fix" a good motor to match a bad prop.

Now use the direction of the flip. A quad that pitches nose-down every time is telling you the fault is on the front pair; one that rolls left every time is telling you it is on the left pair. It flips towards the corner that is not producing the lift it should be. That halves the search space before you touch anything.

2. A prop mounted upside down, or the wrong prop on the corner

A propeller of the correct handedness, mounted inverted, does not produce reduced thrust. It produces thrust downwards. One corner actively pulling the airframe into the ground, while three others push it up, is a lever, and the aircraft will roll over it.

The signature is slightly different from a wrong motor map: the aircraft usually gets light first, strains, and then tips, because three motors are doing real work while the fourth fights them. It is less instantaneous and more agonising, which is exactly why people misread it as "underpowered" and give it more throttle.

Check the leading edge. On every prop, the leading edge is the thicker, more rounded one, and it must face the direction of rotation, with the printed face of the blade up. Two props from the same set on the same diagonal, one flipped, is a classic — especially with props that are not clearly marked, or a mixed bag of R and non-R props from three different orders.

3. Board orientation was never configured

If the flight controller is mounted rotated 90° or 180° from the arrow-forward default, and you never told the firmware, then the gyro's idea of "pitch" is the aircraft's roll — or its negative. Every correction goes to the wrong axis, or the right axis with the wrong sign.

The tell is that the flip is instant, clean, and axis-consistent, and the aircraft is otherwise entirely healthy: it arms, the motors all spin, the numbering is right, the directions are right, the props are right. Betaflight, iNAV and ArduPilot all have a board-alignment setting for exactly this, and the Configurator will show you the model on screen — tilt the aircraft by hand and confirm the on-screen model tilts the same way. If the on-screen model rolls when you pitch the aircraft, you have found your fault without leaving the bench.

A 180° error is the nastier one, because the aircraft still looks correct sitting level, and only reveals itself the instant it needs to correct anything.

4. A reversed axis after a custom mount

The cousin of the above: an unusual mounting angle, a stacked FC with a rotated gyro, or a custom board-alignment value that overshot. Same diagnostic, same test — tilt the aircraft by hand, watch the model. Trust the model, not the numbers you typed.

5. A dead or stalled ESC or motor

If one corner never spins up at all, or spins up and stalls under load, the aircraft levers over that arm. It will look like an instant flip from a distance, but with props off it is trivially obvious: one motor does nothing in the Motors tab, or it stutters and desyncs when you push it. A corner that is not receiving a valid signal will usually have been beeping at you since power-up.

Causes are physical, not configurable — a cold solder joint, a burnt FET, a bad bullet connector, a bent bell rubbing. Check the corner it flipped towards first, because it is telling you where to look.

6. Tune, ground I-term, and creeping the throttle

Last, and least — but real. Two things belong here.

An aggressively tuned aircraft sitting on the ground with the motors spinning will accumulate integral term while it cannot move: the FC asks for a correction, the ground refuses to let the aircraft respond, and I keeps winding up. Then you lift off and all that stored correction dumps into the motors at once. Betaflight's I-term relax and related anti-windup machinery mitigate this, but do not mistake airmode for a cure — airmode does the opposite, keeping full PID authority alive below throttle idle, which is exactly why an aircraft pinned to the ground with airmode on can wind up. An aggressive tune on a heavy airframe with soft landing gear can still bite.

The other is procedural, and it is worth saying plainly: punch it, don't creep it. A multirotor has no control authority while it is still touching the ground, because the ground is taking the load. If you feed in throttle slowly, the aircraft spends a long second in a state where it can tilt but not correct — one leg unloads, it leans, and it tips. Commit. Get the aircraft off the ground quickly and cleanly, and let it have the authority to hold itself level. This is doubly true on a heavy platform with tall legs.

How to know you actually fixed it

Not "it took off once." Do this:

1. Props off, motors spun one by one from the Configurator. Each number is in the right place, each direction is correct. Say it out loud as you go.
2. Tilt test on the bench. Roll the aircraft by hand and watch the FC's model on screen follow it. Pitch it and watch the model pitch.
3. Props on, correct handedness, correct side up, leading edges forward. Check all four again after you have fitted them. This is the step people skip because they just checked.
4. Then, and only then, a hover. A short, decisive one, over grass, with nobody standing where it will go.

If it comes off the ground level and holds attitude, you fixed it. If it flips again, you did not — and it will flip the same way, which is your next clue.

Rehearsing it

A simulator will not find your cold solder joint, and it will not tell you which pad a wire is in. That is a bench job with a multimeter, and no amount of physics fidelity substitutes for it.

What it can do is give you the experience without the wreckage. Run real firmware against a physics model of a heavy airframe, deliberately swap a motor in the map, and watch the flip happen — the direction, the speed, the way the loop diverges. Do it once and you will recognise it forever, in a log, in a bystander's phone video, in the half-second before it happens. On a 5 kg build, the first takeoff is the flight where a wrong motor map costs you the aircraft. That is a very expensive place to learn what one looks like.