Tuning guide

Is the noise the frame or the props?

A blackbox spectrum tells you where the energy is. It does not tell you what made it. The difference between a fixed frame tone and a throttle-tracking prop ridge is visible in the log — but naming the cause still needs a hardware check.

1. A peak is not a diagnosis

The FFT takes the gyro signal, cuts it into windows, and shows how much energy sits at each frequency. That is a measurement. It becomes a diagnosis only when the peak's behaviour — fixed or moving, narrow or broad, one axis or several — matches a physical mechanism you can verify.

This is why "there is a peak at 180 Hz" is not a useful tuning instruction on its own. A motor harmonic, a frame mode, a camera mount and a loose screw can all put energy at the same frequency. The useful instruction is "this peak stays put while throttle changes, so check the frame hardware before you filter it".

2. The two signatures worth knowing

Signature	What it looks like	What it is consistent with
Fixed tone	A narrow peak whose centre stays at the same frequency from hover to full throttle. It may be strongest where a motor harmonic crosses it.	A frame, arm, standoff or mount resonance — a mechanical mode of the structure. Check screws, carbon and mounts before filtering.
Moving ridge	A ridge that climbs as RPM climbs, often with evenly spaced copies at higher frequencies. In a spectrogram it is diagonal.	Motor rotation and propeller blade-pass energy, plus their harmonics. Normal at some level; a problem when it is excessive or reaches the control loop.
Broad raised floor	No single dominant peak; the whole band is louder than it should be.	Many small sources at once: damaged props, dirty motors, a hard-mounted FC, wiring rattling, or simply a noisy build.

The fixed-versus-moving distinction is the one that changes the fix. A static notch is only meaningful for a tone that does not move. A motor harmonic that sweeps with throttle needs a filter that tracks RPM, not a fixed notch at one throttle setting.

3. Other things that look like frame noise

• Camera mount flex. A soft or cracked TPU mount can produce pitch/roll energy at low-to-mid frequencies. It often appears in video as jello before it looks like a tuning problem.
• A loose battery or strap. A pack that can shift acts like a pendulum and shows up on both pitch and roll at a low frequency.
• An antenna mount. A long flexible mount can put energy on yaw while roll and pitch stay quiet in the same band.
• Electrical noise. ESC or power noise can appear at higher frequencies and is often stronger in the D-term than in the gyro. Check the capacitor and wiring before blaming the frame.
• FC mounting. Crushed or hardened gummies raise the floor across all axes rather than producing one narrow tone.

4. The flight that separates them

1. Fly a slow throttle sweep in clean air. Climb gradually from hover into the high-throttle band and back down. The point is to make RPM change while the frame is otherwise undisturbed.
2. Compare hover with high throttle. A peak that is present in both and stays at the same frequency is a frame candidate. A peak that only appears at one RPM is a motor-harmonic candidate.
3. Compare axes. A tone that appears on both roll and pitch at the same frequency is more likely structural. A yaw-only tone suggests an antenna or yaw-specific source.
4. Change one mechanical thing and repeat. Swap to a known-good prop set, tighten the arm and motor screws, or remove the camera. If the peak moves or disappears, you have found something. If nothing changes, the cause is elsewhere.
5. Use the motors tab with props off only as a bench check. It can reveal a rough motor or a rattle, but it cannot reproduce the aerodynamic load or the closed loop. A quiet bench test does not clear a prop or frame.

5. Why "just lower the filter" is the wrong first move

Lowering a lowpass reduces the noise the controller sees, but it also adds phase delay. That delay makes the quad respond later, which can increase overshoot and make the tune feel disconnected. If the noise is a fixed frame tone, a correctly placed notch is more targeted than lowering every filter. If the noise is motor harmonics, RPM filtering is the mechanism designed for it. If the noise is a damaged prop, neither filter is the fix.

The safe order is mechanical first, then the most targeted filter that addresses the confirmed pattern, then re-log. Filtering below what the evidence requires buys nothing and costs phase.

6. What the spectrum cannot tell you

It cannot tell you whether a peak is "too much". A normal motor harmonic can be taller than a frame tone that is actually causing a problem. It cannot tell you the condition of a bearing from a single flight. It cannot separate two sources that happen to share a frequency. And it cannot prove a filter change helped: only a before/after log with the same props, battery and maneuvers can do that.

This is why a good analyser hedges. "Consistent with a frame resonance" is a testable statement. "Your frame resonates" is not, unless someone has checked the frame.

7. How Easyquadtune handles peaks

The analyser requires a peak to stand above the local noise floor, to have a believable width, and — for the mechanical labels — to be confirmed on more than one axis. It compares the high-throttle band with the cruise band instead of averaging the whole flight into one spectrum, because a short punchout disappears in a whole-log average. It reports the peak and the likely category, then recommends a hardware check rather than a filter value.

A peak that is consistent with damage is still only a flag. The next step is always "look at the prop, the screw, the mount", because that is the only way to turn a correlation into a cause.

Questions pilots actually ask

Does a peak at one frequency prove the frame is resonating?

No. It proves the gyro saw energy there. A motor harmonic or an electrical source can land at the same frequency. A throttle sweep, a prop swap or a bench check is what turns the peak into a diagnosis.

Why does motor noise move with throttle?

Rotation frequency rises with RPM, and blade-pass frequency is rotation frequency times blade count. Their harmonics move too, which is why they appear as diagonal ridges rather than fixed lines.

Should I just lower the gyro filter until the noise disappears?

That trades noise for phase delay and leaves the cause in place. Filter only as far as the evidence requires, after the mechanical cause has been checked.

Can a static notch fix any peak?

No. A static notch suits a tone that does not move. A motor harmonic sweeps with RPM, so a fixed notch covers it at one throttle setting only.

Where these questions come from

• Oscar Liang: How to tune FPV drone filters and PID with Blackbox — the throttle-sweep method and how to read raw gyro frequency content.
• Betaflight: Gyro and D-term filtering recommendations — ctzsnooze on frame resonance being independent of props and why notches must match a measured peak.
• libdrone: Resonance and filtering — the difference between a fixed frame mode and swept motor harmonics.
• BlackBox Mate: Noise analysis — reading a spectrogram: fixed spots, diagonal ridges and vertical artefacts.
• Betaflight: Dynamic D and DMin — how damping is varied and why motor heat limits it.

More guides: how to read a blackbox log, wobble on hard stops, overshoot vs droop, I-term relax, dips in fast turns, when a log is not good enough.

Easyquadtune is not affiliated with the Betaflight project. Always fly responsibly and perform an initial hover test in a safe environment after any tuning change. The worked example on this site is an illustrative render, not a customer log.