I am seeking some advice on the initial notch filtering setup (Step 19 of the Methodic Configurator) for a custom 4-EDF (70mm) build. I am following the AMC video tutorial and documentation carefully, but the high-frequency profile of this build is proving difficult to tune.
Motors/Fans: 2100KV with 12-blade Electric Channeled Fans.
FC: MATEKSYS H743-SLIM V4
ESC: HDZero Halo 4-in-1 70A (BLHeli_32.10).
The Challenge:
Given the 2100KV and 12-blade count, the primary noise frequency is sitting much higher than standard props. After following the AMC workflow to enable the notch, the vehicle experiences heavy high-frequency vibrations and significant RATE.Rout spikes.
I am attaching a screenshot of my pre-filter FFT. I suspect the current configuration is introducing excessive phase lag or failing to suppress the “whine” of the 12-blade fans effectively.
INS_GYRO_FILTER: With these specific ICM42688P sensors and high-RPM noise, should I be increasing this frequency to reduce phase lag, or is a lower value required to prevent aliasing?
Notch Configuration: Based on the attached FFT (showing peaks in the 300Hz+ range), do I need a wider INS_HNTCH_BW or a higher harmonic count to capture this 12-blade profile?
Geometry: Does the rectangular frame require an asymmetric approach to the notch harmonics to account for different noise levels on the Pitch vs. Roll axes?
Attached is the Methodic configurator zip file and here is the bin file of the log.
I want to ensure I’m not fighting phase lag created by unnecessary filtering. Any insight into how to properly center the notch for this high-blade-count hardware would be greatly appreciated.
The low thrust/weight is an issue. Previous attempts at using EDF’s have turned out poorly. Their inefficiency makes a great rapid battery discharger. And to @amilcarlucas point motor pole count is not fan blade count. It could be 12 I suppose but you need to confirm that.
I looked at the filter FFT and the nr of poles is definitely wrong!
ESC RPM telemetry is reporting incorrect RPM values because of this.
And the notch filters are tracking the incorrect frequencies because of this.
I totally miscaluclated the motor pole count, I’ve just double checked and they’re definitely 8 poles.
About the MOT_THST_EXPO: I actually threw the EDF on a test bench and logged the voltage, thrust, and current, then plugged everything into that Excel sheet from the Methodic Configurator to nail the curve. Is there another way to recalculate that with the 4 motors mounted? I would like to avoid a massive testing session if posible.
For the INS_HNTCH_ATT, I thought I was supposed to match the frequency on the graph, hence the (-), but I appreciate you clearing that up!
Hey dkemxr, yeah the pole count is definitely 8. About the thrust/weight ratio, it’s super inefficient for sure, but I’m just doing this as a fun test project to learn more about how EDFs perform and their characteristics in different scenarios.
Okay, I’ll change the motor poles and proceed with the new test/ noise filtering. For the number of poles, I counted the number of permanent magnets (8) in the rotor since it’s a brushless outrunner motor.
OK, but they are not at all suited to the dynamic requirements of a multirotor. It will be interesting to see what Auto Tune does to the Rate PID’s. That is if it runs at all w/o failure to level error messages. I would take some weight out if you can, that’s not helping matters. Even if it’s a smaller capacity battery. If it’s just a test project who cares how long it fly’s…
