Wingman AI
Wingman AI Settings
Free — no credit card. Get a free key →
PID Calculator

Compute starting PID values from Ziegler-Nichols ultimate gain method, or start from firmware presets and adjust with sliders.

Method
Quad Class Preset
Adjust Gains
P (Proportional)45
I (Integral)80
D (Derivative)30
F (Feedforward)120
Symptom Diagnostic

Select the symptom you're experiencing. Wingman will diagnose the most likely cause and provide the fix.

CLI Command Generator

Generate Betaflight CLI commands from your tuning values. Paste directly into the CLI tab in Betaflight Configurator.

PID Values
Advanced
Generated CLI
Noise Frequency Zones

Visualize where different noise sources live in the frequency spectrum and how filters interact with them.

V
%
Blackbox Spectral Analysis

Upload a Blackbox log or CSV to run real FFT analysis. Identifies noise peaks, motor harmonics, and frame resonances — then recommends filter and PID settings.

📂
Drop Blackbox file here or click to browse
Accepts CSV export from Blackbox Explorer · .bbl/.bfl raw logs · or plain text gyro values
Or paste raw gyro values manually…
Filter Advisor

Determine recommended filter settings based on your hardware and noise profile.

Tuning Reference by Drone Class

Quick reference for recommended PID ranges, filter settings, and special considerations per quad class.

ClassPIDFFNotes
Tiny Whoop30-5060-10015-3060-10048KHz PWM; thrust_linear=25; low authority
3" Cinewhoop35-5570-11020-4080-130Prop guards add drag; reduced authority
5" Freestyle40-6570-12025-4580-140Standard target; wide I range
5" Racing50-7570-10025-40120-200High FF; low latency; iterm_relax=30
7" Long Range35-5550-9020-3560-120Conservative; iterm_relax_cutoff=10
8-10" Cinelifter25-4540-8015-3040-100Heavy; low D to avoid motor heat
Fixed Wing (BF 4.6)5-505-505-30—S-term primary; SPA I_FREEZE; TPA=airspeed
Tuning Algorithms Reference

Overview of classical and modern PID tuning algorithms applicable to drone flight controllers.

Classical Methods
Ziegler-Nichols (1942)

Increase P until sustained oscillation → record ultimate gain Ku and period Tu → apply formulas. Creates quarter-wave decay. Often too aggressive for drones — use as starting point only.

TypeKpTiTd
P0.50·Ku——
PI0.45·KuTu/1.2—
PID0.60·KuTu/2Tu/8
Tyreus-Luyben (1992)

Less aggressive alternative. Better suited for systems where quarter-wave decay causes problems.

TypeKpTiTd
PI0.31·Ku2.2·Tu—
PID0.45·Ku2.2·TuTu/6.3
Åström-Hägglund Relay (1984)

Replaces PID with an ON/OFF relay to produce controlled limit cycle. Safer than direct Z-N cycling — output stays close to setpoint, no risk of runaway instability. Basis for ArduPilot's AutoTune which performs controlled "twitches" per axis.

Modern / ML Methods
Neural Network Blackbox Analysis

FPVtune: Trained on thousands of real blackbox logs. Analyzes gyro noise spectrum, step response, D-term resonance → generates optimized PIDF + filter CLI commands. Supports BF 4.3-4.5+. Accounts for RPM filter presence and flying style.

Reinforcement Learning

Neuroflight — First NN drone controller (2018). Replaces PID entirely. Trained in simulation to adapt to changing conditions.
RAPTOR — Single NN policy across 32g-2.4kg quadrotors (2025).
TD3 Curriculum — 94% success in dynamic mass scenarios, 60%+ error reduction vs static PID.
MRAC — Adaptive control for unknown payloads, outperforms static PID on PX4.

AI Tune Advisor

Describe your quad, symptoms, or paste Blackbox data. Wingman AI will analyze and recommend PID/filter settings — all sessions are logged to the database.

Current Context (auto-captured)
//Wingman
I can see your current PID settings from the calculator. Describe your quad (frame, motors, props, firmware) and what symptoms you're experiencing. I'll analyze and recommend tuned values.

You can also paste Blackbox frequency data, motor temps, or describe flight behavior — the more detail, the better the recommendation.
Tuning Session Log

Every tuning interaction is logged locally. Export sessions to the Forge troubleshooting database for community knowledge.