Is it propwash, or is my tune bad?

The symptom

The aircraft judders. Not a constant buzz, not a slow wallow — a burst of sharp, ugly, high-frequency shaking that arrives, lasts a moment, and goes. It shows up in the video as a shimmer or a tearing wobble. And you want to know one thing: is this normal, or have I got a bad tune?

This is probably the most-asked question in FPV, and it usually gets a shrug for an answer. It does not deserve one. There is a clean test.

The one question that separates them

When does it happen?

Not what it looks like. Not what frequency it is. When.

Propwash only appears when the aircraft is flying through air it has already churned up. That means: descending, hard braking, tight turns that carry you back across your own wake, or the moment you stop dead after a fast pass. In every one of those, the air arriving at your props is turbulent — and it is turbulent because you made it turbulent, a fraction of a second ago.

A bad tune oscillates in clean air too. Level forward flight, where the aircraft is constantly outrunning its own wake. A punch-out, climbing into undisturbed air. A flick and a stop, at altitude, with nothing around you. If the aircraft shakes when the air is clean, the air is not the problem. Your tune is.

That is the whole article. Everything below explains why it works.

What propwash actually is

A multirotor flies by throwing a column of air downwards. When you descend, you descend into that column — you are, quite literally, chasing your own exhaust. The air under the props is no longer still: it is moving, unevenly, in vortices, at speeds that change from millisecond to millisecond and from one side of the disc to the other.

The props bite into that mess unevenly. One blade passes through a slug of fast-moving air and makes less thrust; the next passes through a slower pocket and makes more. The result is a genuinely chaotic, rapidly changing disturbance torque on the airframe — and the flight controller, which is doing exactly its job, tries to correct it.

Here is the part people resist: the FC is not misbehaving. It is being disturbed by real, physically chaotic air, and no gain value makes chaotic air smooth. You cannot tune your way out of aerodynamics. Every multirotor ever built does this, from a 65 mm whoop to a cinema lifter. It is a property of the flight regime, not a setting you left wrong.

What a bad tune looks like instead

An oscillation from a tune is the opposite of chaotic. It is clean, repeatable and at a consistent frequency — a control loop ringing at its own natural frequency, which is a property of the airframe and the gains, and does not change just because the air did.

Three tells, and they are all about correlation:

• It correlates with stick input, not with airflow state. It appears when you demand something of the controller — a flick, a hard stop, a hold at high rate — regardless of what the air around you is doing.
• It is repeatable. Do the same thing, in the same place, in clean air, and you get the same wobble at the same point every single time. Propwash varies; it depends on how you entered the descent, how fast you were going, what the wind was doing.
• In the log, it looks periodic. A tune oscillation is a sine wave hiding in the gyro trace. Propwash is a burst of broadband noise with no clean period. Pull a Blackbox log and this distinction stops being a matter of opinion.

If what you have is a repeatable oscillation, then you are in PID territory, and the sibling question — which fault is it — is answered in Why does my quadcopter shake?

Triage table

Two rows deserve a note. If it happens in a hover, it is not propwash — you are in ground effect or in still air, not in your own descent wake — and it is very likely mechanical. And if it happens everywhere, that is not propwash either, no matter how much it looks like it in the goggles.

What actually reduces propwash

You cannot remove it. You can make the aircraft recover from it faster, which is what "good propwash handling" really means.

• A healthy D term. D is the damper: it is what stops the aircraft ringing after a disturbance, and propwash is a stream of disturbances. An under-damped aircraft wallows through the wake; a well-damped one absorbs it. Weak D is the most common reason propwash looks catastrophic.
• D_min. Betaflight's D_min runs a lower D at rest and raises it dynamically when the aircraft is being disturbed or you are demanding a fast change. That is exactly the situation propwash creates, so it lets you have strong damping when the wake hits, without paying for that D in noise and motor heat during ordinary cruising.
• Dynamic idle. This holds a minimum RPM rather than a minimum throttle. It matters here because in a descent the throttle is low, and a prop that has spun down close to nothing has almost no authority to correct anything. Keeping the motors turning means they can still bite when the wake arrives.
• Filtering that lets you run that D. You can only run useful D if the gyro signal is clean enough that D is not amplifying noise into your motors. A working RPM filter, balanced props and healthy bearings are what buy you the D you need. This is the compounding loop: cleaner aircraft, less filtering, more D, better propwash.
• Props and disc loading. Nobody says this, and it is often the biggest lever. A heavily loaded disc throws a faster, more energetic downwash and drops into it harder. A larger disc at lower loading, or a prop with a less aggressive pitch, changes the aerodynamics of the problem in a way no PID value can.

Now the hard truth. Propwash can be reduced, never eliminated, and chasing it to zero wrecks an otherwise good tune. The classic failure mode is winding D up chasing the last bit of judder in descents, and ending up with hot motors, a buzzing hover and a wrecked aircraft in the name of a manoeuvre you do for two seconds a flight. A quad that only shakes in descents does not have a problem. It has physics.

Propwash on a heavy platform

If you fly heavy, this section is the one that matters.

A 5 kg aircraft has a high disc loading and pushes down a much stronger, faster column of air than a 250 g racer does. It also has vastly more inertia, so it accelerates and decelerates slowly — including vertically. Put those together and you get the worst possible case: the aircraft sinks into a powerful downwash and then stays there, because it is too heavy to get out of it quickly. It is not passing through its own wake. It is sitting in it.

There is a real fix, and it is not a PID value. Descend with some forward motion. Give the aircraft a horizontal component and it flies out of the disturbed column instead of following it down. Pilots of large platforms do this instinctively, and it works so well that a vertical descent that judders badly becomes a diagonal descent that is nearly clean, with not one setting changed.

Straight-down descent on a heavy platform is a manoeuvre you should be sparing with in any case. It is also, at extremes, how you enter a vortex ring state — the far end of the same phenomenon, and a genuinely dangerous one on a heavy aircraft. Treat "descend with forward motion" as technique, not as a workaround.

How to know which one you fixed

Be strict, because this is precisely where you will lie to yourself.

• Test in clean air first. Fix the tune against level flight, punch-outs and flick-and-stops, where propwash cannot contaminate your evidence. If the aircraft is clean there, the tune is not the problem.
• Then fly the same descent, twice. The identical entry, the same speed, the same line, with exactly one thing changed between them. Anything less and the difference you see is the air, not your change.
• Compare logs, not feelings. If the gyro trace during the descent is no cleaner but you are convinced it "feels better", it does not feel better. It feels like you wanted the change to work.
• Check the cost. Did the change that improved the descent make the hover buzzier or the motors hotter? Then you did not fix propwash — you bought two seconds of smoother descent with a permanently worse aircraft. Put it back.

And accept the endpoint. Once the mechanics are sound, the filtering is honest and D is as high as your noise floor allows, whatever judder is left in a hard descent is the air, and it is staying there. Fly around it.