Servers by jDrones

FireFLY6 VTOL with APM Plane

(David) #101


Looks great, thanks for posting. Are you using the same parameters you previously posted? Not sure why yours is flying so well and mine struggles with the transition and flies poorly in plane. Doesn’t seem like it should change much with a 6s power supply

(Greg Covey) #102


Yes, I using the GregsFF6wAPM_Hover4.param file from post 1.

One way around your brick wall may be to re-install the AvA firmware and re-test. It will be easy and quick as you only need to change the tilt servo cable.

(David) #103

Really frustrated.

Went through the entire set up process again: reset elevrons to factory default 5/8in up, recalibrated compass. recalibrated accelerometers and compared my parameters to Tridges and Gregs. Everything checks. Went to the field. No luck. 4 transition attempts, 4 failures. I don’t know what’s different about my set up. Probably a bit heavier with the 6s batteries but other than that should be really similar to Tridges and Gregs FireFly. Not sure where to go from here.

Here is the latest log file.

Thank you,

(Greg Covey) #104


Are you using two 6s packs in parallel or two 3s packs in series?

Has this setup previously worked using AvA?

(David) #105

Two 6s packs in parallel. Yes, this set up worked in AVA. Flew is at least a dozens times with AvA. The AvA transitions were beautiful. And it flew well otherwise. This was a fully assembled FF6 Pro.

We had repreated issues with their camera trigger and want to use VP-Systems CAMRemote. Howver, I can’t do that with with AvA. They either removed or hid the ability to remap ports. And they are triggering with I2C instead of with an aux port. The original plan was to continue to use AvA with CAMRemote, but that is not an option.



(Greg Covey) #106

It seems that one option could be to use AvA with just a simple camera timer setting; like an auto trigger every 3-5 seconds.

The FF6 Planner doesn’t give you the options as Mission Planner does to set the output. They use a standard relay output on Aux6 to trigger the camera. I never had an issue with my S100 cameras running CHDK with the MobileXcopter trigger for Cannon and Pixhawk. You can, however, set the duration in the parameters via Cam_Duration (normally set to 10 or 1 second duration). Most cameras need to use the relay trigger board that I used.

(David) #107

Thanks Greg. They are no longer using AUX6 for the trigger. For reasons, I don’t understand they are using I2C for the trigger and RCTM, which is another heartburn with AVA. I want to use my own GNSS card, not their L1 only solution.

If I could manipulate AvA to control trigger, camera feedback and RCTM, I’d probably stick wiith it… but I want the control afforded with Aurduplane and full up mission planner.

I am hopeful we can get this working with APM.


(tridge) #108

Hi David,
I spent some time reviewing your most recent logs this morning. My apologies that it has taken me this long to get to it.
The most useful log was 2017-10-12 14-04-04.bin as that is the only one that has an extended period of fixed wing flight as far as I can see (it is in FBWB mode).
For that particular flight can you tell me what the elevator neutral trim was? Was it 5/8 inch up? The reason I ask is that in that flight the average elevator was up about 50% from neutral. That would normally imply a very nose heavy plane. Can you also remember what CoG position you had for that particular flight?
So if I was looking at this log in isolation, without knowing that the FF6 already flies well, I’d be telling the user the following:

  • its quite nose heavy. Move the batteries to shift the CoG by an inch.
  • the fixed wing gains are far too low.
    What doesn’t make sense for me is that Greg and I don’t see what you’re seeing. Has the airframe really changed that much between models?
    On the gains, my normal advice based on this log would be:
  • increase PTCH2SRV_P from 1.1 to 2.0
  • increase PTCH2SRV_D from 0.02 to 0.1
  • increase PTCH2SRV_I from 0.5 to 1.0
  • increase RLL2SRV_P from 1.0 to 1.5
  • increase RLL2SRV_I from 0.3 to 0.6
  • increase RLL2SRV_D from 0.02 to 0.1
    these are really large gain increases, but the log shows lots of control lag, so clearly the gains are far too low.
    The good news is that you have good recovery, so you could try these updated gains with relatively low risk.
    The other part of the log that is puzzling is the transition. It is really horrible. I think you should slow down the forward transitions by reducing Q_TILT_RATE_DN from 30 to 15. You’ll still be able to recover just as fast as Q_TILT_RATE_UP will still be 40, but it should allow the system to remain more stable in transition.
    I’d also recommend reducing Q_TILT_MAX from 45 to 30. That will keep the rotors more upright until you have gained some more speed. I think that 30 will be enough for it to get to transition speed, but if not then try 35.
    I’d also again recommend using FBWA for your test transitions.
    I must admit however that I am really puzzled by your results. If we don’t succeed with this approach then what I think we’d have to do next is fly AvA on this exact airframe and get a log of it flying well so I can do a comparison.
    Cheers, Tridge

(Greg Covey) #109

David revealed that he is using the Pro model which is very similar to the DIY25 model except that it uses two 6s 5200mAh packs or twice the weigh of our battery packs. Other than that, the main power system difference from our working DIY15 models is the motor Kv for 6s.

(David) #110

Well it all came to a dramtic end today. Ironically, I went to the field confident that it was going to fly well. I found serveral problems in my set up that were likely the source of the problems. The vertical tilt was off about 3 degrees. Not a big deal but was causing the FF6 to move forward when it should be stationary. The horizontal tilt was about 7 degrees over rotated, past the intended thrust line. I believe this was causing the apparent heavy nose. There was a 5 degree difference between the left and right forward motor mounts. This was exacerbated by the tail engine assembly being canted 3 degrees right. Combined, I think this is why in airplane mode it only turned right.

I fixed all of these and confidently drove to the field, thinking it would surely fly.

With multiple attempts, I still couldn’t get it to transition. Slower was definitly better, but it wasn’t getting to the transition speed at q_tilt_max 40 (this was after I tried Tridge’s 30 and 35 degree recommendations). I think I needed to set q_tilt_max to 50. However, I had a dramatic and catastrophic crash prior to continued expermentation.

After reviewing the damage and impact area my working hypothsis is a propeller came loose. I double checked the writing was up on all propellers and tightened them, at least I thoguht I did. My habit is to grab the propeller and the rotor on the engine and spin until tight. I swear I did this on all six motors, but the crash site says otherwise. The tell tale sign was finding an unbroken propeller about 30 ft from the crash site. The threads on the engine and in the prop appear intact, with no indications of being torn off.

Continuing to speculate, I think the loss of the propeller induced a violent roll, ejecting one of the batteries and then the other, leading to a powerless and uncontrollable airplane accelerating toward the deck. The FF6 impacted the ground nearly vertical, nose first, crumbling the fuselage. It was hard to watch, knowing I couldn’t do anythign to prevent the inevitiable.

I am reminded of the old ABC Sport’s slogan: “The Thrill of Victory or The Agony of Defeat.”

I am grateful to Tridge for all the time and effort he contributed to my endevor. I am dissappointed I crashed it. I really thought I was close and I think I was.

This expereince has reaffirmed our commitment to Ardupilot and this fantastic community. Thank you for the great work.

Lastly, we are in the market for a durable, Pixhawk controlled VTOL airplane that is well-tested and ready to go. lt needs to cover at least 500 acres in one flight and be able to carry a sony A6000 or equvialent weight payload. Suggestions?


(tridge) #111

I’m so sorry you lost the airframe!
I am still puzzled by the flight results from earlier flights. The motor tilt could explain part of it (especially apparent “nose heavy”, I really should have thought about thrust alignment to explain that).
I’m curios about the motor mounts - you describe a screw thread? Mine is a pure friction fit. If it is a screw thread, is it self-tighening?
You may remember that I lost a prop in flight too - the top rear prop. The plane did turn almost upside down when I lost it, then recovered and continued to fly to land normally. After that I used some locktite on the shafts to prevent it happening again. Strangely enough, another club member found my lost prop on the weekend.

that does make prop-loss in flight pretty likely, yes. If you’d like me to look at the log as well I’d be happy to. Only if you want to.
Condolences on your loss!

(Greg Covey) #112


Sorry to hear about the crash. I know how if feels twice now. If the main parts and wings are still good, you may have the option to buy just a new body. I have done this on my DIY25 twice now. However, my t-style props on the DIY25 have never come off.

As for the prop coming off, my understanding was that the Pro model used custom hubs for the Aeronaut props that were counter-threaded so they cannot come off. Did you have some other setup?

(David) #113


The wings are fairly intact. And, upon initial inspection, the tilt mechanism doesn’t appear to be bent although the servo is ruined. The rest of the electronics all look good. I don’t think I’ll rebuild the firefly6. I am not sure a flying wing is the best design for our applications. It sure was a pretty plane though.

Those are the props I have. I am pretty meticulous when installing my props. Like most things in avaition, the human is the weak link, so I guess I missed something.

Thanks for all the tips as I worked through this process.


(David) #114

Nah, that is my bad. You assumed the fundamentals were correct. My big mistake was checking tilt angles and calibrating the accelerometers with the gear down, instead of with the belly of the airplane being horizontal to the deck. A 6 degree difference is significant.

No mine are friction mounted too, I was refering to threads on the motor.

I haven’t had the heart to look at the logs yet.

Thanks, That’s why they call it the bleeding edge.

(Rifki Hasyim) #115

Hi everyone,
I’m success maiden my FireFLY6 DIY15 using arduplane 3.8.2
QStabilize and FBWA mode is good, the problem only during transition seem yaw/roll-ing to left and right about 10 sec then fly nicely into FBWA mode.
Q_TILT_RATE_DN = 20 and not using airspeed sensor.

maybe someone can analyze my attached logfile.

thank you so much

(Greg Covey) #116

Congratulations, Rifki! Tridge has already commented on this below when I noticed the same thing with slower transitions.


(Rifki Hasyim) #117

Thanks Greg,

after change Q_TILT_RATE_DN to 40, the problem yaw/roll-ing during transition still happen and still too slow

this my second flight log

(Greg Covey) #118

Interesting. Are you pointed into the wind or breeze? Also, make sure the throttle is 50%-75% during the transition. I know that the AvA firmware had a “turbo boost” during the transition to help the plane move forward. I don’t recall having an issue with my DIY15 using the APM Plane firmware. Maybe @tridge will have some suggestions.

(Rifki Hasyim) #119

Yes, i always point to the wind during transition, throttle about 50%

(***) #120

We also have the same props and have had many issues with the hubs actually stripping out of the propeller, I now rip them apart and re-glue myself then re-balance, be very weary of those automatic tightening props the hubs are not glued in very well.