MicroArduCopter, 3" props, Omnibus Nano, Success!

firenitin · May 10, 2020, 6:18am

Here’s my 3" setup using a Kakute F7 and T-Motor F20 II https://github.com/prgumd/PRGFlyt/wiki/PRGWhippet

MartyMcFly · May 10, 2020, 6:59am

Hi Nitin,that was a very interesting build report,really enjoyed it,im back at my 3" project so was very helpful,thank you for sharing

andyp1per · May 10, 2020, 9:17am

I’ve bought one of the $20 lidars that @MartyMcFly referenced. Configured and installed fine, get good readings. I don’t have optical flow on this copter - what EK2 changes do I need to make?

Also @wicked1 I had a really bad case of the leans on Friday with my KakuteF7Mini. Never seen it before. I wonder if its because it got much warmer. Might experiment a little.

anbello · May 10, 2020, 10:50am

To have rangefinder as primary alt source?

In this case:
EK2_ALT_SOURCE = 1 AND EK2_RNG_USE_HGT > 0

But beware there are some problems with this:

github.com/ArduPilot/ardupilot

EKF is not using RangeFinder when EK2_RNG_USE_HGT > 0, EK2_ALT_SOURCE = 0

opened 07:21PM - 05 Dec 19 UTC

fs007

BUG EKF

Copter 4.0 (actually this bug existed since copter 3.4) There is a coding err…or in AP_NavEKF2_PosVelFusion.cpp that prevents the rangefinder from being used as primary Alt-Source even if the EK2 parameters are set accordingly (EK2_RNG_USE_HGT > 0; EK2_ALT_SOURCE = 0) No matter, which value frontend ->_altSource has, all the code of “// determine if we are above or below the height switch region” is NEVER executed ! Discussion [here](https://discuss.ardupilot.org/t/bug-in-ap-navekf2-posvelfusion-cpp/49919) ``` // select height source if (extNavUsedForPos) { // always use external vision as the height source if using for position. activeHgtSource = HGT_SOURCE_EV; } else if (((frontend->_useRngSwHgt > 0) && (frontend->_altSource == 1)) && (imuSampleTime_ms - rngValidMeaTime_ms < 500)) { if (frontend->_altSource == 1) { // always use range finder activeHgtSource = HGT_SOURCE_RNG; } else { // determine if we are above or below the height switch region float rangeMaxUse = 1e-4f * (float)frontend->_rng.max_distance_cm_orient(ROTATION_PITCH_270) * (float)frontend->_useRngSwHgt; bool aboveUpperSwHgt = (terrainState - stateStruct.position.z) > rangeMaxUse; bool belowLowerSwHgt = (terrainState - stateStruct.position.z) < 0.7f * rangeMaxUse; // If the terrain height is consistent and we are moving slowly, then it can be // used as a height reference in combination with a range finder // apply a hysteresis to the speed check to prevent rapid switching float horizSpeed = norm(stateStruct.velocity.x, stateStruct.velocity.y); bool dontTrustTerrain = ((horizSpeed > frontend->_useRngSwSpd) && filterStatus.flags.horiz_vel) || !terrainHgtStable; float trust_spd_trigger = MAX((frontend->_useRngSwSpd - 1.0f),(frontend->_useRngSwSpd * 0.5f)); bool trustTerrain = (horizSpeed < trust_spd_trigger) && terrainHgtStable; /* * Switch between range finder and primary height source using height above ground and speed thresholds with * hysteresis to avoid rapid switching. Using range finder for height requires a consistent terrain height * which cannot be assumed if the vehicle is moving horizontally. */ if ((aboveUpperSwHgt || dontTrustTerrain) && (activeHgtSource == HGT_SOURCE_RNG)) { // cannot trust terrain or range finder so stop using range finder height if (frontend->_altSource == 0) { activeHgtSource = HGT_SOURCE_BARO; } else if (frontend->_altSource == 2) { activeHgtSource = HGT_SOURCE_GPS; } } else if (belowLowerSwHgt && trustTerrain && (activeHgtSource != HGT_SOURCE_RNG)) { // reliable terrain and range finder so start using range finder height activeHgtSource = HGT_SOURCE_RNG; } } } else if ((frontend->_altSource == 2) && ((imuSampleTime_ms - lastTimeGpsReceived_ms) < 500) && validOrigin && gpsAccuracyGood) { activeHgtSource = HGT_SOURCE_GPS; } else if ((frontend->_altSource == 3) && validOrigin && rngBcnGoodToAlign) { activeHgtSource = HGT_SOURCE_BCN; } else { activeHgtSource = HGT_SOURCE_BARO; } ```

andyp1per · May 10, 2020, 10:55am

Ah cool, thanks for the heads-up. Is this true in EKF3 also?

anbello · May 10, 2020, 11:01am

Yes, but I have to write some more otherwise comment too short.

andyp1per · May 10, 2020, 11:03am

Should be easy to fix? Or are there issues?

anbello · May 10, 2020, 12:39pm

I only tried to revert PR https://github.com/ArduPilot/ardupilot/pull/12250 because in the dead code there are condition to test if rangefinder go over max, but I saw same behavior.

wicked1 · May 12, 2020, 7:58pm

Let me know if you figure anything out w/ this.
I think you know most of my details… Basically, if I have GPS lock, it rarely does it. But if I take off before I have GPS, it leans. And, I’m often trying to fly w/out GPS… inside or in the forest, etc.
If mine leans when i do have GPS lock, then I have a real problem… Badly unbalanced prop or something.

The more throttle I give it, the more it leans… Whether that is from vibes or power noise, not sure.

Vibrations do seem to affect it. Better vibes = less leans, per copter. BUT, my omnibus copter has vibes spiking to the high 20’s, and it barely leans, while I’ve gotten my kakute mini vibes down to peaking at about 3 or 5, and it leans the most out of all of them. It’s better than when its vibes were up in the 20’s, but still does it.
I’ve tried adding capacitors to the battery, to the 5v rail, and directly across the IMU power pins w/out any change.
I’ve tried a lot of settings… Lowering Gbias_p_nse might have helped, but not positive. Switching to EKF3 might have helped, but not positive.
In the end, I just compensate for it. Level the copter on the bench. Take off and fly around gently so it starts leaning, then trim it level. Then that flight will be mostly OK. After power cycling and taking off again, it will pull to the right immediately upon takeoff, but then it develops the left lean, which corrects the right trim, and ends up being mostly level. It maintains level this way even after getting GPS lock…

So basically, I’ve learned to work around the issue. I’d like to solve it, though.
And FYI, I have not done any testing w/ your IMU filter selection firmware… My flight time has been very limited the past few months. I’m just happy to have a few minute to fly occasionally, and can’t really work on solving my issues at the moment. Hopefully work slows down, and preschools reopen, and I’ll have time again soon…

andyp1per · May 14, 2020, 1:22pm

My KakuteF7 Mini finally arrived from banggood - its v2

MartyMcFly · May 14, 2020, 1:50pm

thats great news Andy im trying to get ardupilot on to my Matek 4o5 flashes o/k with beatflight but wont connect to ardupilot but will keep trying have you a link for the board Andy please

andyp1per · May 14, 2020, 1:54pm

MartyMcFly · May 14, 2020, 1:56pm

It is just Ardupilot hex I flash please

wicked1 · May 14, 2020, 1:57pm

I’ve got an unused v1 here, if you really don’t want the v2. I’d trade. (Not sure what part of the world you live in, but I’m in the united states)

andyp1per · May 14, 2020, 1:59pm

Yes. More words because

andyp1per · May 14, 2020, 1:59pm

UK. It’s not going to be worth the shipping and it’s probably good for me to have both versions

andyp1per · May 16, 2020, 4:15pm

Finally got to do some actual flying with my 3" today. Very pleased with the results - nice and fast and stable. Only issue was I am getting a lot of video noise. I am having to power the runcam off the battery because of the current draw and I have a big cap and a DVS diode on the supply, but clearly a lot of noise is still getting through. Any suggestions to cure without adding much weight?

wicked1 · May 16, 2020, 4:22pm

I also have the runcam powered off the battery.

How is your VTX powered? Directly from the battery, or via the 5v rail from the FC?

On my omnibus copters, I use a vtx powered from the 5v rail, and it’s fine. But on the kakute, it has noise. I switched to a vtx w/ onboard power supply that connects to the battery, and it is fine on the kakute.

I have a 560uf low esr (oscon) cap on the esc/battery leads. And another one on the 5v rail. (5v one is a smaller 6.3v cap). Omnibus and Kakute are set up this way.
I did try doubling the capacitance on both the battery and 5v rail on the kakute, and it did not help the video interference. But like I said, a bigger VTX w/ built on regulator is working.

andyp1per · May 16, 2020, 4:26pm

I have a Unify Pro32 which only goes to 13v, so I need to use the 5v rail. That said its mounted in a 20x20 adapter, so I guess I could get the HV maybe. They are not cheap though - I would prefer to make what I have work.

andyp1per · May 16, 2020, 10:09pm

I’ve just reviewed the footage from the runcam - no noise at all so I concur that it must be either the VTX or the cabling from the camera to the VTX