Why there are H flying well and other not, after firmware upgrade?

[chiloschista]

Hello ensemble,
after some thinking why my H does no more fly as it was at first and after THoff note on another thread I checked for level the floor where I do accelerometers calibration.
The floor is level. Then what?

Well the landing gear is level, what about the controller board?
I checked with a precision gauge the motor levels and discovered that the motors are off on height each other by amounts between 3mm and 5mm.
Not enough! The motor axis are not leveled, nor are then the propellers.
A quick raw test was to put the copter reversed on ground. The motors are definitely not at the same height and the propellers are not leveled.

So I start to think that accelerometers calibration is done of thumb in any case, unless there is a way to figure out the offset to place the copter when doing that, to have the accelerometers board really leveled.
Someone has the chance to guess it right, someone not.
Any thought about that? Any experience?
Would you like to check your copter for body and motors alignment?
Thanks and best regards,
Ric

 

[QuadBart]

Just my personal experience; I used to be totally anal about this and totally ensure the surface I was doing the accelerometer calibration was completely level. I then did a little test cause I thought there is NO way you had to have so level a surface that you could place a marble on the surface and it wouldn't move.

I took my Q500 and did accelerometer calibrations on surfaces I thought looked totally flat then flew the bird. Afterwards I'd take my level one that I used for ensuring my scopes on my long range rifles are level, and checked the surfaces. Most of these surfaces were tilted. The bubble was nowhere near the center. BUT the Q500 flew perfectly fine, no drifting, no tilt, no altitude drifting. I even did it on top of a trashcan lid and shimmed it with folded paper till it looked level and it flew fine. I even did the calibration on my sidewalk that appears level but water does run in one direction, no issues...

Point is, I don't think these things are super sensitive and need to be absolutely measured and verified level. I think they only need "eye-ball" accuracy for level. I would think if it had to be absolutely level they would state that and that would be too much for the avg. user.

Now if my bird, like my H experiences issues, I'll take out the marble tile and measure it all out and ensure its completely level. I've not found doing that made any difference for me. If it was to fix a drift problem it still drifted...

Again, just my .02

 

[Steve Carr]

Just my personal experience; I used to be totally anal about this and totally ensure the surface I was doing the accelerometer calibration was completely level. I then did a little test cause I thought there is NO way you had to have so level a surface that you could place a marble on the surface and it wouldn't move.

I took my Q500 and did accelerometer calibrations on surfaces I thought looked totally flat then flew the bird. Afterwards I'd take my level one that I used for ensuring my scopes on my long range rifles are level, and checked the surfaces. Most of these surfaces were tilted. The bubble was nowhere near the center. BUT the Q500 flew perfectly fine, no drifting, no tilt, no altitude drifting. I even did it on top of a trashcan lid and shimmed it with folded paper till it looked level and it flew fine. I even did the calibration on my sidewalk that appears level but water does run in one direction, no issues...

Point is, I don't think these things are super sensitive and need to be absolutely measured and verified level. I think they only need "eye-ball" accuracy for level. I would think if it had to be absolutely level they would state that and that would be too much for the avg. user.

Now if my bird, like my H experiences issues, I'll take out the marble tile and measure it all out and ensure its completely level. I've not found doing that made any difference for me. If it was to fix a drift problem it still drifted...

Again, just my .02

I tend to agree with all of that. The exception would be for those having a problem. It's still a good practice to ensure the calibration is not causing the problem. Just a matter of elimination.

 

[Brian Mackey]

I tend to agree with all of that. The exception would be for those having a problem. It's still a good practice to ensure the calibration is not causing the problem. Just a matter of elimination.

does it matter if you do this indoors or outdoors?

 

[chiloschista]

Well, thanks both for your answers.
I'm doing autopilots tests from ten years now, especially for MatrixPilot team. On that one, with fixed wings, the accelerometers and gyros offset is recorded before each flight, so we had to keep the plane almost level and almost still. Now we have a feature that allows those offsets to be checked and recorded once forever.
I did a lot of success flying with those "almost", and the board itself is mounted "almost" leveled, which means maybe a couple of degrees off, but I noticed it was definitely not necessary to be more finicky than that.

But now I feel that all the flying I did with drones and RC planes in the past decades, can be thrown down a ... well toilet.
I definitely can't sort it out and can't understand which really could be the problem. Considering also that I'm doing calibration the same way from beginning and at beginning it was flying rocksteady.
Maybe I will try to calibrate it in the garden at a random place.

 

[Tuna]

The accelerometer needs to be able to measure if the H is moving, so it judges the position of the central unit. The motors don't have to be level at all - the controller will adjust power to each motor until it 'feels' that the unit is level, or moving in the direction requested. There are a few videos of people hanging odd things from the landing gear that show that the controller doesn't really need a perfectly balanced machine to be able to hold steady flight.

A lot of things go on during a firmware upgrade, and if you're calibrating as well, there are a bunch of ways you can get your H into a bad state. People have reported that phones and watches can cause a bad calibration, so I suspect that the upgrade itself isn't causing the small number of problems we've seen reported - but that people are giving their H's a full shake down during the firmware upgrade and that's what's throwing up trouble. This last week has also seen some fairly intense solar weather, which may be contributing to the issues being reported.

 

[chiloschista]

does it matter if you do this indoors or outdoors?

Should not as they are mainly sensitive to heat and vibrations. Better not doing it after having left the copter hours in the car under the sun and has to stay very still. Moving it during calibration means great problems.

 

[chiloschista]

The accelerometer needs to be able to measure if the H is moving, so it judges the position of the central unit. The motors don't have to be level at all - the controller will adjust power to each motor until it 'feels' that the unit is level, or moving in the direction requested. There are a few videos of people hanging odd things from the landing gear that show that the controller doesn't really need a perfectly balanced machine to be able to hold steady flight.

A lot of things go on during a firmware upgrade, and if you're calibrating as well, there are a bunch of ways you can get your H into a bad state. People have reported that phones and watches can cause a bad calibration, so I suspect that the upgrade itself isn't causing the small number of problems we've seen reported - but that people are giving their H's a full shake down during the firmware upgrade and that's what's throwing up trouble. This last week has also seen some fairly intense solar weather, which may be contributing to the issues being reported.

I can agree with you, especially having a DIY multirotor, which is definitely out of balance, but flies very well (Pixhawk), but if the motor axis are not aligned, better said vertical, well there could be problems.
But those should have been present from beginning.
I was used to check for solar activity and GPS predictor, but have seen that has not the terrible effects you can read around, unless you should do high level autonomous flying. Not our case and that should cause some drift or some difficult getting a GPS lock, not stability issues. Unless the H works in its very own way.
By the way I meant if the motors are not leveled, neither is the board.

 

[Brian Mackey]

Should not as they are mainly sensitive to heat and vibrations. Better not doing it after having left the copter hours in the car under the sun and has to stay very still. Moving it during calibration means great problems.

thank you, I was avoiding doing it inside but tough to find level, outside

 

[Merlin]

I use a 2 foot square sheet of 3/8 inch plywood, four thin chocks and a small spirit level. It's easy to carry and setup anywhere.

 

[Easy12]

May be off topic.....Since I have my H I never calibrated and she flies just fine. Even after SW updates!

 

[Brian Mackey]

May be off topic.....Since I have my H I never calibrated and she flies just fine. Even after SW updates!

Cool, life on the edge of a razor blade is always more fun

 

[Steve Carr]

I would suggest turning off the GPS while in flight to see if it performs better. Also look at the flight logs to see if you can spot any errors. Several people who have had erratic behavior turned off the GPS and found the bird suddenly flew normally.

I'm of the opinion that sometimes during the update some of the flight systems are not updated correctly resulting in the bad flights. You can also check with Yuneec to see if they can talk you through a reinstall of the update. I have heard in the past this could only be done by tech support at their facility, but one person reported a few days ago that tech support did it over the phone and it fixed the problem.

 

[Brian Mackey]

I would suggest turning off the GPS while in flight to see if it performs better. Also look at the flight logs to see if you can spot any errors. Several people who have had erratic behavior turned off the GPS and found the bird suddenly flew normally.

I'm of the opinion that sometimes during the update some of the flight systems are not updated correctly resulting in the bad flights. You can also check with Yuneec to see if they can talk you through a reinstall of the update. I have heard in the past this could only be done by tech support at their facility, but one person reported a few days ago that tech support did it over the phone and it fixed the problem.

yep, just redo it, but, that was on the ST16 I can;t speak for the H

 

[chiloschista]

Cool, life on the edge of a razor blade is always more fun

I know a guy who put in flight a new RC plane without checking the centre of gravity. The plane flew well. He did say "well, checking the CG is not necessary, I will never do it anymore". Good luck

 

[chiloschista]

Report of a test flight.

Before flight an accelerometers calibration was done again, but on another leveled floor.
I got in flight and had a very stable copter. I did climb, dive, going straight forward and backwards. It did everything well straight, stopping as expected and drifting no more that 20-50cm, probably due to low GPS accuracy (longtime to get a GPS lock and "only" 10-11 sats). I thought it was fixed.

But I thinked a bit more at what could get the IMU confused.
So I started to climb and diving yawing all the time, going around in circle. I switched in smart mode, which I rarely use, and pushed the copter forward and backwards while yawing.
Everything went well. If accelerometers or compass are not set well, this way of flying should have real issues. I thought again it was well done.

But not enough convinced, I continued to fly, trying to hover very close to a tree, trying obstacle avoidance etc. ...
AND: at a certain point the copter started to fly no more straight. This is the beginning of the toilet bowl effect in what I experienced so far.
The exact behaviour was if I push the copter forward it does an arc to the right, drifting about 5m in a ~30m straight path. When I stop pushing, the copter does a "half bowl" clockwise, drifting another ~5m, then it almost stops. Coming back it does the opposite, but always in a clockwise motion.

Now the guesses.
The copter at beginning was stable: I guess the offsets are set well.
The copter stabilization drifts with time.
They fixed an issue about high temperatures usage.
I guess this copter uses gyros as well. The gyros offset can be set once, so this has not to be recorded at bootup.
Accelerometers offset has to be set leveled. That could be the reason to have an accelerometers calibration.
Acceleremoters should not drift, while gyros drift.
Compass should not drift. It can only be set wrong, but should not lead to other effects than false heading related navigation.
Do the Yuneec developers change any setting in the IMU PID to better match for high temperature drift?
Is that related to Integrative value of the PID (guessing they use a full PID*), which is now working wrong and accumulates drift with time, instead of correcting it?

I bet there are hidden functions to set PID and to set gyros offset.

What I could do next is try to recalibrate accelerometers again, taking account of the body being leveled. But I'm almost sure this has nothing to do with the drifts I'm experiencing.
Maybe I'm figuring out what's happening? Frustrating is that we really don't have any way to fix that ourselves.

Comments and experiences welcome!

Ric

* there is a smart way to correct gyros drift using GPS position. That's used in MatrixPilot firmware. So a PD is enough, not needing the Integrative.

 

[Tuna]

Have you given the H a period long enough to get a full GPS almanac? It's possible your GPS configuration changed during the flight.

You should also take a look at the logs to see - GPS accuracy, number of satellites, any compass warnings and so on. If you don't have other software to do that, my app can help - the link is in my signature.

 

[chiloschista]

Tuna, thanks for your input. I'm an enthusiastic user of your app already = : D
Yes I checked in the telemetry and GPS accuracy is between 0.6 and 0.4.
Anyway I really am not convinced that this is related in any way with GPS. I already described why above.
Maybe I do another flight without GPS.

 

[Tuna]

Tuna, thanks for your input. I'm an enthusiastic user of your app already = : D
Yes I checked in the telemetry and GPS accuracy is between 0.6 and 0.4.
Anyway I really am not convinced that this is related in any way with GPS. I already described why above.
Maybe I do another flight without GPS.

I'll be honest, I'm just guessing at your problems - it's not clear to me what the difference is between one machine and the next and I don't like not knowing . It's frustrating to me that I can have a great experience with my drone, yet others have so much trouble.

 

[Rayray]

Now the guesses.
The copter at beginning was stable: I guess the offsets are set well.
The copter stabilization drifts with time.
They fixed an issue about high temperatures usage.
I guess this copter uses gyros as well. The gyros offset can be set once, so this has not to be recorded at bootup.
Accelerometers offset has to be set leveled. That could be the reason to have an accelerometers calibration.
Acceleremoters should not drift, while gyros drift.
Compass should not drift. It can only be set wrong, but should not lead to other effects than false heading related navigation.
Do the Yuneec developers change any setting in the IMU PID to better match for high temperature drift?
Is that related to Integrative value of the PID (guessing they use a full PID*), which is now working wrong and accumulates drift with time, instead of correcting it?

I bet there are hidden functions to set PID and to set gyros offset.

What I could do next is try to recalibrate accelerometers again, taking account of the body being leveled. But I'm almost sure this has nothing to do with the drifts I'm experiencing.
Maybe I'm figuring out what's happening? Frustrating is that we really don't have any way to fix that ourselves.

Comments and experiences welcome!

Ric

* there is a smart way to correct gyros drift using GPS position. That's used in MatrixPilot firmware. So a PD is enough, not needing the Integrative.

Wow, I got totally confused about the 3rd guess. I do not really understand what you are doing or why you are doing it. Perhaps you are trying to reverse-engineer the Typhoon H, or re-engineer it. I wish you luck, obviously you are spending a lot of time doing whatever you are doing.