[sscott]

In theory, you can connect 127 I2C devices in parallel to the same pins, provided they each have a unique address. So I would go ahead and try connect them to the same pins.

[sscott]

tldr – Balance your motors and do in-flight accelerometer trim.

I think I’ve managed to resolve the drift problem. In case anyone else runs into this same problem, I thought I’d give some feedback on what steps I took:

1.) Motor vibrations

Apparently motor vibrations are a big cause of roll drift in angle mode. There is an entire discussion on this on the Cleanflight github page. It seems the consensus is to mount the FC on a shock-absorbing gel material. This is clearly not possible for the F3FC tri. The other option is to mount the motors on gel padding, but this would probably require a redesign of the motor mounts.

Armed with this knowledge, I spun up each motor in turn with the props off and looked at the accelerometer graph in the motors tab of cleanflight. I noticed that at *exactly* 1573 throttle, the right motor would cause a large positive Y accelerometer value. It was not like a normal oscillation that goes up and down, but just a large positive value (like 0.3), so clearly some kind of aliasing was going on. At one click up or down throttle, this vibration would disappear.

Using small squares of adhesive vinyl, I managed to balance the motor to remove the vibrations at this throttle value. Both motors now showed constant vibrations of 0.0015 over the entire. Further balancing got these vibrations down to 0.001, but I’m not certain if this final step was actually necessary.

2.) Accelerometer calibration

I placed the tricopter on a glass jar and repeated the accelerometer calibration, but this time I ensured that the frame itself was level, and not just the table, in case the jar was skew. I did this by placing a bubble level meter (spirit level) on top of the tricopter frame and adding shims until it was perfectly level.

3.) Bench tailtune and ESC calibration

I reran both of these just to be safe.

4.) In-flight accelerometer trim

At the field, I took off in angle mode and immediately noticed that the drift to the right was significantly less. Previously, I required 50% stick movement to keep it still, now it was more like 5-10%. I removed this final drift use stick commands to trim the accelerometer (process is described on the Triflight Checklist, under Accelerometer Hover Calibration). Around 10 movements were required.

It now hovers perfectly still… well, provided the wind stops blowing, which is only did for a few seconds at a time yesterday.

Thanks everyone for all the tips.

[sscott]

NI2O, to keep things simple, you can just keep the tri in angle mode all the time until you are comfortable flying it.

Assign one switch to arm.
Assign the other switch to tailtune.
Assign angle mode to any switch, but move the “sliders” in the GUI so that angle mode is always selected, regardless of switch position.

Once you get the tricopter tamed, you can then use the tailtune switch to change modes instead.

You can also use the rate mixing that Bengt suggests, but I would recommend keeping things simple at the start.

[sscott]

Ok thanks. So which version did you upgrade to?

[sscott]

@BranBepic, I have the exact same problem with the mini tri drifting to the right in angle mode. Did you ever find a viable solution to this?

[sscott]

Did you ever solve this problem?

[sscott]

Ok. Just to clarify, if you move the aileron/roll stick to the left, the right motor should increase in speed, assuming you have the tail pointing towards you.

[sscott]

Digikey stocks that part for $0.62

[sscott]

Hi NI2O

I also had to reverse the yaw and aileron direction on my Dx6i radio to get the bars in the receiver tab to move in the correct direction.

What do you mean by ” but copter itself follows the right direction of the aileron stick” and “but copter itself reacts in the right way”? Can you please explain more.

After reversing aileron and rudder, my tri seems to roll and yaw in the correct directions while in the air. It just seems to drift to the one side a lot, as I mentioned in my other post.

[sscott]

Out of interest, has anyone used the accelerometer trim functions? Apparently you can trim the accelerometer using stick movements (while disarmed) to correct for auto-levelling offsets. Perhaps this will help me if I can’t find any motor vibration issues.

[sscott]

Ok thanks Lauka, I’ll try that tonight. I did balance the props, but I didn’t check the motors themselves.

[sscott]

The tail servo leaning over is normal. It always has to compensate for torque since there is an odd number of motors.

Thanks, I thought the same thing, but this is good to know for a fact now.

If you calibrated your accelerometer and your ESCs, then maybe you have a bad motor? Loose prop? Bad prop? Prop upside down (aka, a reverse prop mounted upside-down on a motor that is not reversed, it will turn in the right direction and produce thrust but not nearly as efficiently as if it were a non-reverse prop on a non-reverse motor).

Props all firmly mounted.

Left motor: EMAX CCW motor wired to spin CCW. 6×4.5 prop, text facing upwards.
Right motor: EMAX CCW motor wired to spin CW. 6×4.5R prop, text facing upwards.
Tail motor: EMAX CCW motor wired to spin CCW. 6×4.5 prop, text facing upwards.
By “text”, I mean the writing with the prop size that is molded into the propeller.

If something electrical was connected incorrectly or you didn’t put 180-degree pitch correction in the configuration, it would have just flipped over, and should not fly, so that’s probably not it. A bad connection between motor and ESC would make the motor behave erratically, which it doesn’t sound like is the case. Possible a bad connection between the ESC and the PDB? Nothing is weird with your subtrim or travel limits in the radio?

Yes, definitely nothing completely reversed. I even checked beforehand by placing my hand beneath each prop, and all motors were definitely blowing air in a downward direction. I am pretty confident in my power connections (I design and assemble PCBs as part of my job). And as I mentioned, the aileron is centered at 1500.

[sscott]

The current sensor on my F3FC board works well. Cleanflight reports 0.16A while unarmed on the bench. I did not need to adjust any settings to get this reading.

If you’re seeing 10 amps in cleanflight, then you have a hardware problem. Either an incorrect current sensor (perhaps the wrong value current sensing resistor was installed), or you have a short circuit somewhere which is drawing an actual 10amps and the current sensor is freaking out.

I don’t think that tweaking settings in cleanflight will solve the problem.

I was running triflight 0.5 beta 3 firmware.

[sscott]

NI2O, I fly with a Dx6i. Just assign tailtune to either the Flap switch or the Gear switch.

[sscott]

Wow, ok thanks everyone. swissfreak, your build looks so clean with all the wires inside the boom. I think I might just try that even though extending all the wires will be a pain.

[Author]

Posts