I am working with a DJI 10033/48kV (42 poles, 4kW) motor from their large agricultural drones and a flipsky fsesc 75100 pro v2.0. I know this ESC doesn't have an excellent reputation but it's what I have for now. I've read that low kV motors are notoriously difficult to tune, however my application requires it.
During auto-tuning the motor turns quite smoothly through all RPM. In use it has difficulty starting, makes horrible noises, stutters (at all speeds) and occasionally comes to a complete stop. I have tried different auto-tuning parameters, large/medium outrunner, generic/propeller, openloop erpm between 700-2500.
Beyond the autotuner I have mostly followed ChatGPT advice for various settings:
| Parameter (from FOC → Detection Results) | Comment |
|---|---|
| R = 30.4 mΩ | Right where a large, low‑Kv outrunner usually lands (20 – 40 mΩ). |
| L = 26.1 µH | Typical inductance for this size; well above the “too‑low‑to‑control” zone. |
| Lq – Ld = 3.22 µH | Saliency is present but modest → don’t bother with MTPA. |
| Flux‑linkage λ = 5.0 mWb | In line with other ~50 Kv motors (VESC numbers are in mWb, not Wb). |
| Observer Gain (×1 M) ≈ 40 | Auto‑calculated and within the usual 30–60 band. |
| Field | Your value | Why it matters / suggested change |
|---|---|---|
| Openloop ERPM | 2000 | Perfect starting point (≈ 95 mechanical rpm on 21 pole pairs). |
| Openloop Ramp Time | 0.10 s | Good for a prop that must spool quickly. If the spin‑up sounds harsh, bump to 0.20 s. |
| Openloop Time | 0.05 s | Leave; this is the dwell before observer takeover. |
| Openloop Current Boost | 0 A | This replaces the old “Startup Boost %” (GitHub). Give the rotor a shove: set 5–8 A (with a big prop 8 A is common). |
| Openloop Current Max | −1.00 A | “−1” means “unlimited.” Safer: set it to your phase‑current limit (e.g. 120 A) so a runaway blind‑drive can’t cook the motor. |
| Start Current Decrease / …ERPM | 100 % / 2500 | Fine; the boost tapers away as the observer locks. |
| Hysteresis / Lock Time | 0.10 s / 0 s | Add a little Lock Time (0.10–0.20 s) if you ever hear a quick stall just before lift‑off. |
When this didn't work the following was suggested:
| Field (FOC → Sensor‑less) | Set to | Effect |
|---|---|---|
| Openloop Current Boost | 7 A | Ensures the rotor accelerates steadily instead of dithering. |
| Openloop ERPM | 2 200 | Small bump helps if chatter starts ~90 rpm. |
| Openloop Ramp Time | 0.20 s | Softens the first ½‑turn so the observer lands on a clean sine wave. |
| Observer Gain @ Min Duty (Advanced tab) | 10 % | Extra gain keeps the estimator locked when duty is < 5 %. |
I have tried a lot of variation but can never achieve a noticeable change in the issues I described above. I am considering swapping the ESC for a trampa, but expecting that to fix my issues feels like wishful thinking.
Any advice?