Rates: what each number actually does to the stick

Rates are not tuning

Get this straight before anything else, because almost every rates argument on the internet is really two people talking about two different things.

PIDs decide how well the aircraft obeys. Rates decide what you are asking for.

The PID controller's entire job is to take a commanded rotation rate — say, 300 degrees per second of roll — and make the aircraft actually rotate at 300 degrees per second, right now, without overshooting, wobbling or lagging. That is a control problem, and it is what your P, I and D terms are for.

Rates never touch that loop. Rates sit upstream of it and answer a completely different question: when your stick is 40% of the way over, what number should the PID controller be handed? That is it. That is the whole job. Rates are a curve, nothing more.

This is why "my quad oscillates so I lowered my rates" is a nonsense sentence, and yet you hear it constantly. Lowering rates makes you ask for less, so the aircraft has an easier time obeying, so the oscillation gets smaller and you conclude you fixed it. You did not. You turned down the demands on a broken control loop. Fly aggressively again and it will be right there waiting for you.

If the aircraft does not do what you ask, that is PIDs. If it does exactly what you ask and you do not like what you asked for, that is rates.

The three numbers

Every rate system in mainline Betaflight is shaping the same thing: stick deflection in, commanded rotation rate out. One input axis, one output axis, one curve between them. The systems differ in how they let you describe that curve, but the curve is the only thing that flies.

Conceptually there are always three knobs:

1. Centre sensitivity — how steeply the curve rises as you leave centre stick.
2. Maximum rate — where the curve ends. The rate you get at full deflection.
3. Expo / curvature — how the curve bends between those two points.

Two of those set the endpoints. The third sets the shape in between. Once you understand that, the rate systems stop being mysterious and become what they are: three different ways of typing in the same picture.

One important caveat before the details. All of this describes Acro (rate) mode, where the stick commands an angular rate in degrees per second and the aircraft keeps rotating for as long as you hold it. In Angle mode, the stick commands an angle instead — 50% stick means "hold half of the maximum lean angle," and the maximum is set by its own separate angle limit, not by your rates. Your rate curve still shapes stick feel there in some firmware paths, but the thing at the end of the curve is not a rate any more. If you tune rates in Angle mode and then flip to Acro, expect a surprise.

Centre sensitivity — how twitchy it is around the middle

This is the slope of the curve at centre stick, and it is the number that decides whether the aircraft feels nervous or dead in your hands.

Almost all real flying happens within a small distance of centre. Holding a line, correcting for wind, easing into a turn, adjusting a hover — those are small stick movements. So centre sensitivity, not maximum rate, is the number you feel for 95% of a flight.

Too high and the aircraft twitches at inputs you did not know you were making. You will find yourself gripping the sticks harder, which makes it worse. Too low and it feels numb: you move the stick, nothing much happens, so you move it further, and then it happens all at once.

Maximum rate — how fast it can spin at full stick

This is the end of the curve. Slam the stick to the corner and this is what you asked the PID controller to deliver.

It is also the number that gets people into trouble, because it is the one that sounds impressive. Nobody brags about their centre sensitivity. But maximum rate only matters at the very edge of stick travel — flips, rolls, hard yaw spins, the fast half of a race line. If you never touch the endstops, it is a number that describes flying you are not doing.

Raise it if your flips feel slow and lazy and you are already hitting full stick to get them. Lower it if full stick makes the aircraft do something violent and disorienting that you did not intend, or — and read the last section — if your airframe physically cannot deliver it.

Expo / curvature — the softness in the middle without giving up the top end

Expo is what lets you have both. It bends the curve so that it is shallow near centre and steep near the ends, which means fine, precise control around the middle and the same big number when you slam the stick.

This is the correct way to fix twitchiness while keeping your flips. Rather than lowering maximum rate — which costs you top-end authority you may actually want — add curvature and the aircraft calms down where your thumbs live while the endpoints stay where they were.

The cost is linearity. With a lot of curvature, the relationship between stick position and rate stops being intuitive, and muscle memory built on one expo value does not transfer to another. Most people end up with a modest amount and leave it alone for years.

The rate systems

Betaflight ships several selectable rate types — among them Betaflight (the original), Raceflight, KISS, Actual and QuickRates. They are separate parameterisations, not aliases for one another. iNAV has its own arrangement.

Here is the honest summary: they are different parameterisations of the same curve. All of them produce a stick-in, rate-out mapping. None of them can do something the others fundamentally cannot. What makes them confusing is not the maths — it is that the same word means different things between them, so a number that means "maximum rate in degrees per second" in one system means something else entirely in another. That is why pasting numbers from one rate type into another produces an aircraft that is either unflyable or asleep.

Do not try to reason about the numbers. The Configurator draws you the curve. Look at the curve. That picture — how steep it is at centre, where it ends, how it bends — is the rates. The numbers are just a coordinate system for describing it, and it is a different coordinate system in each rate type.

Pick a rate type, learn to read its picture, and stop worrying about which one is "better." None of them is better. One of them may suit how you like to think.

What good rates feel like

Symptom	Move this	Which way
Twitchy, nervous, hard to hold a line	Centre sensitivity	Down
Twitchy at centre but flips are already good	Expo / curvature	Up
Numb, vague, you keep over-correcting	Centre sensitivity	Up
Flips and rolls feel slow and lazy at full stick	Maximum rate	Up
Full stick is violent and disorienting	Maximum rate	Down
Fine around centre, unpredictable near the ends	Expo / curvature	Down
Yaw feels sluggish compared to roll and pitch	Maximum rate (yaw only)	Up
Aircraft never reaches the rate you commanded	See the heavy-aircraft section	—

Note that last row. It is not a rates problem, and turning rate numbers up will make it worse.

How to actually set them

The method is dull and it works:

1. Start from a known-sane default. The stock rates in current Betaflight are perfectly flyable. They are a starting point, not an insult.
2. Fly something repeatable. A line you know, a turn you know, a flip you always do the same way. If you cannot repeat the flight, you cannot evaluate the change.
3. Change exactly one number.
4. Fly the same thing again. Not a different thing. The same thing.
5. Repeat, and stop when you stop noticing an improvement.

And now the blunt part: do not copy a famous pilot's rates. They are not a tune, they are not a setting, they are not transferable, and downloading them tells you nothing except how somebody else's hands work.

Rates are a property of your thumbs. They depend on your grip — pinch or thumbs — on your gimbal tension, on how long your stick ends are (a longer stick means more physical travel for the same electrical deflection, which changes everything about centre feel), on how your radio's own curves are set, and on what you fly. A rate curve that is sublime for a freestyle pilot on short sticks with a pinch grip is genuinely bad for you on long sticks with a thumbs grip flying long cinematic lines. Both of them are correct. Neither of them is portable.

The one useful thing about somebody else's rates is as a starting point in the right neighbourhood — and even then, only from someone flying a similar aircraft in a similar style.

Rates on a heavy aircraft

A 5 kg platform cannot achieve the rates a 250 g racer can. Not "should not." Cannot.

Rotational acceleration is torque divided by moment of inertia, and a heavy airframe has vastly more inertia, on longer arms, with heavier props that themselves take time to change speed. Its motors can produce more torque, but nowhere near enough more to compensate. The physics simply does not permit a big machine to rotate like a small one, and no number you type into the Configurator will change that.

So what happens if you ask for it anyway?

The PID controller sees a commanded rate the aircraft is not achieving, and does what it is designed to do: it demands more. It pushes some motors harder and others softer to generate the rotation. But motor output is bounded — it cannot go above 100% or below the idle floor. Once the controller's demand hits that ceiling, it has saturated, and this is where things get genuinely bad:

• The controller loses authority. It is already commanding everything it has. Any additional disturbance — a gust, a stick input, a correction it needs to make — has nowhere to go. The aircraft is momentarily uncontrolled in that axis.
• Mixing breaks down. With one motor pinned at maximum, the differential between motors that produces the rotation gets clipped, so the aircraft rolls, pitches or yaws differently than the mixer intended. Axes bleed into each other.
• The I term winds up. It is integrating an error it cannot fix, and it will unwind at the worst possible moment.
• Everything downstream inherits the mess. The oscillation you spend the evening chasing may simply be a controller repeatedly hitting the wall — which is why you check the motor outputs in a log before you interpret anything else.

The aircraft that results feels mushy and laggy and never quite arrives where you pointed it — which is exactly what people describe when they say a heavy quad "flies like a barge," and then go and raise the rates trying to fix it.

Lower maximum rates on a heavy machine are not a compromise. They are honesty about what the airframe can do. Set rates the aircraft can actually deliver, and the controller stays inside its authority, the mixer stays linear, and the aircraft becomes crisp, predictable and precise — genuinely nice to fly, just not fast to spin. That is not a lesser aircraft. That is the aircraft you have, flown properly.