karla

  • ****
  • 295
Re: GPS assist and Heli
« Reply #60 on: December 08, 2017, 01:57:45 am »
Thanks guys,
Yes, power cables twisted as fas it allows, even to servos, can't be done everywhere.
Yes, I did the simple compass check, north is where the fc think it is.

karla

  • ****
  • 295
Re: GPS assist and Heli
« Reply #61 on: December 08, 2017, 02:47:54 am »
Remove the Roll/Pitch in Auxmag orientation and redo a test.

Well, I removed the trims and put the values back to 0,0 before doing mag calibration. Then I simply left them at that.

f5soh

  • *****
  • 4117
    • LibrePilot
Re: GPS assist and Heli
« Reply #62 on: December 08, 2017, 07:17:57 am »
Auxmag orientation has no effect while calibrating Mag.

Remember you can test the Complementary+Mag+GPS fusion algorithm assuming you are using Next branch.
Maybe post some log, while flying at same heading may help.

karla

  • ****
  • 295
Re: GPS assist and Heli
« Reply #63 on: December 08, 2017, 07:38:05 am »
Thanks. Yes this heli is now using Next branch.
If using Complementary+Mag+GPS instead of INS13, what could be expected? Does it rule out some sensors?
Yes, I can do some logs while flying in same direction, once I got the heli re-configured with the mag/gps mounted further out on the tail boom. I have another trex450 with the gps installed at that location and its doing very stable position hold with a DJI Naza-H FC and GPS.

f5soh

  • *****
  • 4117
    • LibrePilot
Re: GPS assist and Heli
« Reply #64 on: December 08, 2017, 09:22:52 pm »
If you're happy with Basic (Complementary) in Attitude stabilization, Complementary+Mag+GPS will work fine for all GPS modes.

karla

  • ****
  • 295
Re: GPS assist and Heli
« Reply #65 on: December 09, 2017, 01:29:28 pm »
okay, lets see.
Have completed the hardware readjustments today.
tomorrow if weather allows will try it
K
« Last Edit: December 09, 2017, 01:35:46 pm by karla »

karla

  • ****
  • 295
Re: GPS assist and Heli
« Reply #66 on: December 12, 2017, 03:27:02 am »
Too windy this weekend but did fly a bit anyway, ended up with a bad soldering on the power cable and had to abort testing. Next time...

karla

  • ****
  • 295
Re: GPS assist and Heli
« Reply #67 on: December 22, 2017, 05:50:58 am »
Finally, yesterday I got to fly some 3 packs and try the improved location of the mag/gps unit and the Position Hold.
It holds position better.
Still have bowling motions, but not always and not as much as before, but still more than acceptable. So I conclude I still have remaining disturbances with the compass.
I only fly in Attitude mode and using Next.
It flies good in Basic, INS13 and 'Basic+Mag+Gps'. Tried PS in both INS13 and B+M+G, with no noticeable difference what I can see.

There might be additional issues with my Heli, though.
At the end of the last flight there was a sudden drop of engine power. Then motor spin back up again but enough late to almost end on ground.
I would need some help to determine (if possible) Was this because of a radio link failure (failsafe) or was the failsafe due to a short power outage from the LiPo's, a cable disconnect ?
Where to look in the log-file?

It happens during the last 20 seconds of the log file.
During this flight I move in and out of PH several times.

Thanks a lot

Re: GPS assist and Heli
« Reply #68 on: December 22, 2017, 10:55:20 am »
I forget:  Are you flying 16.09 or recent "next" or some other version?

There is a thing called "bumpless transition" in the thrust/throttle management PID for PH.  A safety fix was put in to "next" that broke this "bumpless transition" and I don't know if that breakage itself got fixed.  "Bumpless transition" is supposed to make sure the thrust actuator has the same value when you switch into PH, rather than some other value and the aircraft say descends before determining that it needs to add thrust.

If I am correct though, this would be for the thrust channel, so if your motor power dropped (to zero) but you are configured to use collective for thrust control, this is probably not your issue.  If you are configured to use throttle for thrust control then this could be the cause.

You could replay the log file and watch the bottom status line of the PFD to see if it went into failsafe.  There are also some alarms you could check by configuring and watching a scope on the chance that it was so brief that it did not show up on the PFD long enough to see easily.

If you use DJI/Naza GPS, you might try the patch 16.09 firmware I did or patch the next source with the same changes from that thread.

karla

  • ****
  • 295
Re: GPS assist and Heli
« Reply #69 on: December 30, 2017, 01:55:05 pm »
Hello and thank you cliff!
Christmas came in between :)

I forget:  Are you flying 16.09 or recent "next" or some other version?
Next

There is a thing called "bumpless transition" in the thrust/throttle management PID for PH.  A safety fix was put in to "next" that broke this "bumpless transition" and I don't know if that breakage itself got fixed.  "Bumpless transition" is supposed to make sure the thrust actuator has the same value when you switch into PH, rather than some other value and the aircraft say descends before determining that it needs to add thrust.
Right now its pretty Bumpy when switching from Manual to AltitudeVario thrust mode. However, much less so, after I adjusted the Thrust limit for Neutral, it was way off.

If I am correct though, this would be for the thrust channel, so if your motor power dropped (to zero) but you are configured to use collective for thrust control, this is probably not your issue.  If you are configured to use throttle for thrust control then this could be the cause.
Now, this is a really interesting idea you have. Really.
However, I don't think it works like this at all.
There is no way or place you can select if 'Thrust' is managed via throttle or collective.
Please prove me wrong, it would be fantastic :)

You could replay the log file and watch the bottom status line of the PFD to see if it went into failsafe.  There are also some alarms you could check by configuring and watching a scope on the chance that it was so brief that it did not show up on the PFD long enough to see easily.
Did replay it and looking for useful variables to check, but was unable to determine if the motor outage was due to a power outage from LiPo or a radio glitch causing the failsafe (and shutting motor down). Anyways, last soldering of power cables seems good. The OPLM unit on the heli did have a bad connector of the antenna and have swopped for another unit now. Also changed the 'Air Data Rate' up to 100.000 baud from 64.000 as suggested by Laurent.
Now it seems a good stable connection and I dont expect any motor or failsafe issues going forward.

If you use DJI/Naza GPS, you might try the patch 16.09 firmware I did or patch the next source with the same changes from that thread.
I use a OP gps unit now. However, I consider a DJI unit if this do not show improvement.
I have one on another Trex450 mounted at the same place and it holds position just fine.


The heli is set up fine now and I plan to head for the field tomorrow to test position hold :)

Re: GPS assist and Heli
« Reply #70 on: December 30, 2017, 06:20:03 pm »
It has been a long time since I looked at heli thrust, but back then, you use your transmitter throttle stick to control something called Thrust (called thrust for all vehicle types), then in the setup you tell it (SystemSetting->ThrustControl) whether Thrust controls the throttle channel (coming out of the FC) or the collective channel.

It should be possible to use a governor mode ESC (same RPM at any load) connected to an aux channel, then use Thrust=Collective for good heli setup ??
« Last Edit: December 30, 2017, 08:17:23 pm by TheOtherCliff »

karla

  • ****
  • 295
Re: GPS assist and Heli
« Reply #71 on: December 31, 2017, 05:47:24 am »
OMG, you are right, I found it.



This gives me new hope, will explore it.
But after I get the 'bowling' to a complete stop
Thanks

f5soh

  • *****
  • 4117
    • LibrePilot
Re: GPS assist and Heli
« Reply #72 on: December 31, 2017, 01:02:24 pm »

karla

  • ****
  • 295
Re: GPS assist and Heli
« Reply #73 on: January 01, 2018, 05:21:48 am »
Yes, I missed that.
It was not my focus at the time, sorry laurent.

Meanwhile, I flew some more yesterday.
Its still 'bowling' while in position hold more than I think is acceptable.
I compared the direction of north to my iPhone compass and noticed it points to the same direction but during a 60 sec period the direction is moving around +10 to -10 degrees.
This is when standing on ground and mags are different there, however it can't be good.

I think already did my best to twist cables and find best possible location to mount the mag/gps on heli. Now will try another unit.
If that will not work I think need to find a drastically different mounting.
Or maybe its the material on the whole tail boom...
I tried to have another unit with mag standing still and then move the heli over and around it, it is showing some more error % but its difficult to draw any firm conclusions.

 

Re: GPS assist and Heli
« Reply #74 on: January 01, 2018, 09:49:46 am »
Once you start calibrating mag, it is always getting data until procedure is complete.

You should not set model on the ground or set it on anything metal or even calibrate close to big metal things (car, metal tanks, etc.) at any time once you start the procedure until it is complete.

You can click the GCS mag cal "next position" button 5 times before you even start, and leave it on the 6th position, and then do all the positions, then click the button the 6th time.

You can just do the "Naza dance" instead of the 6 positions.  I do a different 6 position dance where I simply point (and wiggle a little) each of the 6 directions of the model (nose, tail, left, right, top, bottom) at the north mag field which for me is north and 60 degrees down.

World magnetic inclination map
https://en.wikipedia.org/wiki/File:World_Magnetic_Inclination_2015.pdf

LiPo alarms have magnets in them.

You can find the limits of the mag fields of things by setting a spare FC on the table.  Connecting it to GCS.  Watching mag scope in GCS while you move the model close to it.  When you see the scope change you have found a mag field in the model and have found how large it is.