VTOL -- freeman2100 vertical take-off and landing fixed wing +pixhawk

Sadly, it was a lot more than a hard landing… the drone is destroyed.

Thanks for the heads up on the ESCs. I don’t think that is what happend. The battery never got below 20v, so even at the highest setting the esc should not have shut down.

dave

Hi,

Really sorry that happened. I don’t think voltage or generally power is the problem. I run mine at voltages around 20V no problem. Also, it can’t activate the ‘landing detected’ in the air, I think thats impossible. Can you check in your parameters ‘BATT_FS_LOW_ACT’ if it has a value ‘5’ Which is a terminate command may caused your problem. You can check the link: https://ardupilot.org/copter/docs/failsafe-battery.html

Hi David,

Sorry for your crash. You and I have totally different philosophies when it comes to testing new aircraft or VTOLS. I will typically turn off all the alarms and automatic stuff so that I am in control. This comes from experience. By simplifying things, I believe that less than go wrong. Would you rather destroy just a battery pack or the entire plane? I also have never used any AUTOTUNE or QAUTOTUNE feature.

That being said, your initial problems started when trying to exit QAUTOTUNE mode and you had a critical battery alarm of 20.96v. Yes, this is below your BATT_CRT_VOLT setting of 21.0v.

When changing from QAUTOTUNE to QLAND, I do not see any change on the mode switch of Ch. 5. Was this done via another switch setting or the BATT_FS_CRT_ACT set to 4?

It does appear that Ch. 3 creeped up to (1900) to activate the MOTORS_EMERGENCY_STOPPED feature of Option 31. I would not recommend using this feature.

Some food for thought on your next VTOL maiden flight. Disable all the failsafes, test the copter modes first. They may perform fine using the default parameters or at least good enough to move onto a forward flight transition. Use your transmitter timer to set a hover/flight time that only uses about 50% charge. I often use two packs when testing and then change the transmitter timer based upon the charge left in the pack. These 10AH+ packs can fly for a long time!

Good luck on your repairs!

2 Likes

Greg,

Thanks for taking the time to review the log and for sharing your analysis and thoughts.

Funny you start out with our testing philosophies are completely different. I adopted the order in which I test Q modes from you, otherwise I probably would always start in qloiter. And I normally don’t set a battery failsafes.

I do autotunes because I don’t understand PIDs well enough to tune the drone myself.

The qland mode was not commanded but was triggered by the BATT_FS_CRT_ACT to 4. I also had q_options set to 32, use QRTL instead of qland. Not sure this worked as I expected. It never went to QRTL, instead it transitioned directly to qland. Not sure why? But it could be I was close enough to the homepoint.

Is a kind understatement. This, as you correctly identified, was the cause of the crash. I should have never set RC3 to 31. It was a poor, misguided decision that I will attribute to ignorance and sleep deprivation. It’s tough to be stupid

Tough way to learn a lesson, but I am glad that it was human error, which increases my faith in the firmware and hardware.

Thanks,

Dave

Dave,

You are definitely not stupid. It’s easy to rely on all these “bells and whistles” but try to simply add one at a time after the initial testing. There is a lot of work going in the wiring of these VTOLs so I hope that this saves you some rebuilding effort.

If you are not comfortable flying a copter, buy an inexpensive one and fly it until it becomes second nature and fun! In the “old” days, we built our own copters and flew them with KK2s or Multiwii controllers. Like anything else, when you learn from the ground up, it helps you understand what all the sensors and flight modes are doing. So step 1 is to hover test your Freeman in three modes; QSTABILE, QHOVER, and QLOITER…always landing under manual control. If you need tuning, you can create a video for us or simply tune it yourself using a couple of parameters.

Next, the transition from one of the modes above to FBWA needs to done with the Air Speed sensor disabled. It can be in calibration mode (ARSPD_AUTOCAL) but ARSPD_USE should be set to 0. You can compare the log graph of AS sensor vs. GPS ground speed after the flight.

Once you have several flights with transitions, you can test CRUISE and LOITER modes on a calm day. If these work well, then enable your AS sensor if the graphs look appropriate. I like to fly the plane in LOITER mode and see the graphed wind effect. The AS sensor will make your plane more efficient once enabled.

After that, you can start adding some automated flight mode tests or enable the battery monitor, failsafes, etc. By then, you will know just how long you can fly and be able to handle emergencies by taking over manually.

Good luck!

1 Like

Hi David,

Here are some interesting recent events that may have also contributed to your crash. It’s a warning on using the latest releases until they have been vetted over time.

WARNING! AutoTune bug in Copter-4.0.1/4.0.0

I use these guidelines. Many years ago, IBCrazy had a video on tuning with this:

P to low, slow oscillations, sluggish
P to high, stable but fast oscillations

I to low, very sloppy, hard to keep position
I to high, overshoot and oscillate, can be compensated by D up, but can go out of control
25% of P is good to start

D to low, overshoot only shortly
D to high, fast oscillations, but almost unnoticed

Everytime I do PID tuning, I refer to them. Works great for me. Hope it helps you and others as well.

2 Likes

Thanks Greg. Unless there is a compelling reason to update to the newest release, I wait for several months.

Dave

Thanks Anthony.

I love this forum.

Dave

1 Like

Will be building my Believer as a VTOL with PH later this summer when the possibility of getting
the wing kit shipped from China is a little more assured! This will be my go-to source for the build,
and thanks to all posters here!!

1 Like

Hi Mark,

Welcome to our build thread! It’s a bit slow right now due to the weather this time of year and COVID-19. My Freeman sits in quarantine waiting for some love! :mask:

1 Like

Hi,

I have a Freeman 2300-kit (the one without all the electronics).
I have all the electronics (Pixhawk 2.1, Mauch, …) and the load will be a Sony A6000.
But I do not have the motors.

I have only seen Sunnysky X3520 520KV with APC 15x10" for the front.
And Sunnysky 4112 450KV with 17x5.5" CF for the rear.
I cant find specs on these motors using the mentioned propellers.
The reason is that I am trying to use other motors instead (KDE Direct or Tiger Motors).
But I need more specs to find replacement-motors.
Do all the four motors have the same load in hover?

/MickeM

MikeM, Do you know about e-calc? It is a free motor / prop calclator web app, and you can select almost any type of motor, battery, prop, etc. For comparisons. I will look at the data for those motors and props, but suggest you could use the app to select a very similar motor.

Edit: I have the pro version. It has the full database. The free version is limited to the most popular motors. To me its well worth the cost. Also, I picked a 3S lipo and 150 ESC. I am sure the lipo was too small. If you tell me a cell count and ESC I will post the performance specs for those, and for any other motor and prop too.

They all have the same load if the plane is balanced properly, but they can have slightly different output settings, front to back, based upon prop size and motor Kv. Remember that forward flight is where you want to be efficient and that hovering is only a momentary portion of the flight.

See post #125 above for motor hover graphs.

Cheers!

Well, after a long delay I finally finished my Freeman rebuild. I have done several short test flights. I am pleased to report no crashes… I’d like to move forward with a Qautotune but I have a few concerns. The first one is stability. The plane flies ok in QStablize, QHover and Q loiter. So I started a QTUNE on the pitch axis. The first move was a significant yaw to the left, which surprised me. I let the tune continue for a few more minutes and didn’t like the way the airplane was responding so I abandoned the the tune and landed uneventfully. The parameters are set to comply with the tuning guide or as recommended in the QAUTOTUNE - VTOL autotune for quadplanes.

The second concern is amperage draw during the tune. In a hover with gently maneuvering, amperage is about 50 amps. During the tune you can see spikes to 150 amps. I am using a 200 AMP Mauch Power module and 60 amp ESCs so theoretically it should be fine. But the battery, XT90 connector and 10GA wire getting are warmer than I’d like. It’s not hot to the touch but definitely north of warm. To mitigate high amperage I am thinking about setting the current limit to 100 - 125 amps. What are the implications of setting a current limit.

Below, please find a link to the log from the tuning flight. You’ll see three flights; two quick test flights and an attempt at a tune.

Love to here if others did a QTUNE and the results.

Thanks,

Dave

1 Like

Hi David,

Congratulations on your re-build and successful hovers!

Your hover flights appear to range from about 2 minutes to almost 4 minutes for the QAUTOTUNE. These are significantly long hovers and would not be part of a normal flight so this is likely why your XT90 connector and wires get warm.

I don’t have any experience with QAUTOTUNE. I also don’t have any experience with the Q_M_BAT_CURR_MAX parameter but the THR_MAX parameter can be reduced from 100 to say 75 if you experience excessive throttle bursts during transition.

The Mauch PMs use hall-effect sensors so your current is not limited to the 200 amp rating. The sensor will simply clip the reading above 200 amps.

Can you describe what “ok” means? Are you not happy with some part of the performance?

Cheers!

1 Like

Greg,

Thanks for the reply. Yes, I understand the intent is to only be in copter for a limited time for landing and takeoff. Long term, I am not worried about the max current, it will be considerably less in plane.

My concern is during the tune. Each axis takes about 3-5 mins with spikes up to 150 amps. Limiting thrust and perhaps current to some value, may undermine the goal of the tune. The plan is to tune one axis at a time, land, let things cool down for 10-15 mins and move to the next axis.

I don’t have any experience with quads this big, so don’t really know how they are supposed to fly, which is one of the reasons I am interested in the tune. My quad experience is limited to DJI Phantoms and Inspire.They fly remarkable well. I don’t expect a 2m 8kg VTOL to fly similarly, but really have nothing to compare it to.

I know it may seem like I am stubbornly pursuing an AUTOTUNE. I believe it will lead to a much more stable copter, which is my goal.

Thanks,

Dave

Dave,

I still remember my first wind lesson in flying a VTOL. It was with a BEV FireFLY6. I had many years experience flying planes and then copters but never in hovering a plane in wind. It was a scary learning curve in just how much the wind and orientation can lift the tail or wings. No amount of tuning will overcome a hovering situation when the wind hits your plane at the wrong angle. Fortunately, a QuadPlane will outperform a tri-copter in this respect.

First, try to do your testing on a very calm day. You seem to be doing all the right things but be aware that flying a Phantom and Inspire will skew your expectations!

Perhaps you can ask your question right on the QAUTOTUNE blog. If I recall, the original testing was done by ARACE on a similar setup using a Crosswind VTOL.

Good luck!

1 Like

Im curious to know what power draw others are getting with their 2300 versions? I just want to know if its worth fiddling with motor angles, CG, etc.

My flight this AM was pretty poor with an average of 22A on the cruise with maximum payload. Takeoff pulled about 80A. Conditions were pretty much perfect with no wind and 18°C.