E-Flite mini-Convergence with ArduPilot

The E-Flite Mini Convergence is a smaller version of the original Convergence VTOL. It came out about a year ago: RCG thread

Recently the PNP version has been on sale for only $80, so I bought one to serve as an ArduPilot tri-tiltrotor testbed.

The stock flight controller (FC) is nicely integrated: it has a 5.25V regulator (EUP3484) rated at 3A which supplies the servo power rail and the 3.3V regulator for the MCU. The ESCs have only 3 wires; all motor current flows through the ESC power rails on the FC. ESC signal output is PWM at 400Hz, servo signals are 71Hz PWM. The FC PCB measures 32x46 mm and is hard mounted to the plastic frame which anchors the rear motor mount. Servos are Spektrum A360 3.7g (elevons) and A350 5g (tilt). The recommended battery is a 3S 800 mah 30C LiPo.


I chose to use the mRo X2.1 to replace the stock FC, since it is almost the same size (31x51 mm with vertical pins). I enlarged the opening in the foam to allow mounting the X2.1 with servo tape for vibration isolation, and cut a slot in the fuselage to provide access to the USB port and microSD card.

Since the X2.1 does not have an integrated BEC, I also installed an mRo ACSP7 power module and a 5V/2A BEC in the space provided for the stock RC receiver. An FrSky XSR receiver and the buzzer are mounted on the tail, and the Matek SAM-M8Q GPS is installed inside the canopy.

The X2.1, buzzer, ACSP7, GPS and cables weigh 31g and the stock FC weighs only 12g: weight will be increased by about 20g over stock.

The mini Convergence is flying well at a weight of about 300g with a Tattu 3S/850mah 75C battery (XT30 connectors) in both VTOL and FW modes.

Vibration levels are about double those of the full size Convergence, but still acceptable:


Forward and back transitions looked pretty good, with cruise speed in FW mode of about 20 m/sec and current draw less than half that for hovering. Hover is at about 43% throttle.

Plot of current and voltage in hover modes:

Extrapolated hover endurance (to 20% capacity) with 850mah battery is 6.3 minutes.

Log of QSTABILIZE/FBWA flight:

5 Likes

nice! That FFT shows energy at 500Hz. Any signs of aliasing in the logs?

Attitude estimation looks OK:

I guess the accelerometer sample rate is just 1KHz?

It looks like the IMU2 vibration level is due to a resonance in the rear motor mount which is also the mounting platform for the FC:

Flight testing of PR https://github.com/ArduPilot/ardupilot/pull/11763

This PR adds 2 new parameters, Q_FWD_THR_CH and Q_FWD_THR_MAX for quadplanes to manually control forward throttle in VTOL modes. This is useful to counter headwinds without pitching downward with the associated loss of lift from the wings.

This is a plot of groundspeed, battery current and pitch against the RC input with Q_FWD_THR_MAX =0.25

RealFlight 8 SITL demo of manual forward throttle control:

Add non-inverted S.port telemetry for YAAPU script on Taranis X9D+.
Oscar Liang post: https://oscarliang.com/uninverted-sbus-smart-port-frsky-receivers/
Photo:

Google Photos

Back on sale at $99

1 Like

I just flew my miniConvergence to test out a new PR, and I’m seeing some interesting coupling between roll and pitch. Can anyone offer advice on whether this is just bad tuning?

@tridge @andyp1per I’m playing with this miniConvergence again and wondering whether I can improve the PID tuning using the throttle-based notch filter.

It now has 3-blade 4x2.3 props on the 2500KV motors with 3S power and a flying weight of only 326g.
I think the frame may be too flexible since the motor nacelles seem to lack rigidity, and the logs seem to show some cross-coupling between roll and pitch.
Attitude control feels a bit loose but it flies reasonably well.

Could you take a look at the fft_prefilt and postnotch logs here? I’d appreciate any comments or advice you might have.
prefilt
postnotch
test_postnotch
BTW, this is a branch off a recent master with a new manual forward throttle option that works more like that in the stock Convergence.