LibrePilot Forum
Users => Vehicles - MultiRotors => Topic started by: FeuerVogel666 on March 26, 2016, 11:01:21 pm
-
Hi everybody,
First I am sorry for my bad english.
I have set up my 650 Quad with a Revo and LibrePilot.
It flys really well and smooth and dont mind about windy conditions or fast flying.
But sometimes it suddenly begins to oszilate, then i have to land, disarm and rearm, so it flys still again.
I have flown around 30 till 40 times now.
And have crashed it now for the 3rd time, because it suddenly tilts to the left.
If that happens the right motors make a short fullpower burst, so it turns around 90° to the left and crashes down.
Can it be a fault of the cc or of my setup from the software?
-
So I have tested a few things and this on two copters with different setups.
But let me explain this.
At first I had Problems in my 650 Copter. Sometimes it tilts to one side and I have crashed it 3 times or so.
Then I build it up with a second Revolution but got the same behavior.
To make it secure it is nothing from the other Hardware, I took my Hexacopter, and changed it from APM to Revo.
Took it in my Backyard and tested it. But it made the same (see my Video under this text).
So i took another Transmitter, but always the same.
In the meantime i had got a 3rd Revo for warranty, cause I thought the first one was broken.
So I tested this one and got still the same behavior.
BUT when I change to Basic (Complementary) it flys smoother and dont show anything from the tilting.
Does this mean ALL 3 Revos are broken or is there still an error in the Firmware or maybe an mistake made by me with the setup?
In the Video I got all Green in the GCS and flew in Attitude with Throttle on Manual and Yaw Axislock.
https://www.youtube.com/watch?v=_MxLFD6o8U8
-
Maybe due to vibrations and Accelerometers go crazy.
Maintain firmly the frame on ground and monitor the horizon behavior while apply throttle. Be safe !
Solutions:
- Add filtering, go to Attitude tab
- Add adhesive foam on FC case to isolate from vibrations
- Balance props.
-
So, i tested a little around with higher filtering and so on.
I watched the Horizon while flying and even a while, when the Copter was standing still on the ground.
I made a log of the behavior.
Sometimes, while Copter is standing on the Ground and Motors are disarmed, the Horizon jumps around for an second and goes back on normal, but I dont know enough to figure out what is happening.
I put the Log down here, it happens to the end.
Logged while the copter stands on the Ground an is disarmed.
Sometimes this kick is much stronger, sometimes it is a little tick.
I dont know what to do now.
Where can I buy an Revo wich isn´t a clone. Best would be in the EU, but no need to be.
-
Please edit your title: you have a Revo, not a CC3D.
All Revolution boards are clones, the last genuine board is maybe 2 years old now.
-
Ok, I have edited the title.
Ok, but where should i buy to be sure to get Boards wich are build within the specifications...
-
There's nowhere you can order from and be sure it will be a good board. You order one and check it when it comes in to make sure it's good. The main problem is the regulator and it's capacitor creating noise on the barometer. The thread https://forum.librepilot.org/index.php?topic=311.0 talks about how to check the barometer. If it's bad you either complain to the seller or fix it yourself.
I've had good luck with the one we all hate the name of, "Revolution Mini". It's a revolution clone in a smaller form factor with the oplink removed. You can add what they call an "oplink air" to it and you effectively have a full revolution but in two smaller boards instead of one bigger one.
-
Hi,
I'm new to the forum and a beginner in the rc multicopter world. I'd like to share my observations on a similar issue.
I've experienced similar behaviour (sudden tilts) with my copter build. Using 'next' build with I2C mag support. ZMR250 with Revo and Beitian BN-880 GPS & Mag, Micro MinimOSD and OPLM telemetry, PIDs tuned well as far as I can judge. Generally the copter flies very well in all manual and autonomous modes. But sometimes when using INS13GPS and flying in self-levelling mode (no matter manual or auto-pilot) the copter suddenly jerks, mostly to the right and jumps up the same time. The artificial horizon in OSD goes crazy and it takes several seconds to show correctly again. Most of the times that happened (the crazy horizon) I was in RATE mode so no problems with the control of the aircraft but once it happened when landing in ATTITIDE (again INS13) and it was a nasty surprise. Luckily managed to keep the copter in the air till the attitide estimation recovered.
I've checked the telemetry records and noticed that when this happens, there is always a glitch in datastream from GPS (just missing data for a single second), this also causes the 'NO GPS FIX' message on OSD. These glitches (missing 'packets' in the continuous dataflow from the GPS) can also be observed in uBlox Control Center when on the ground. They happen very, very rare, but they do. And then the EKF estimation just goes crazy and it takes some time to recover.
It would be great to have some kind of validity filter and/or dead-reckoning for the GPS data in the Revo firmware, or in case the GPS fails, the EKF to be switched momentarily to Complementary just as a failsafe solution, till the EKF estimation is good again.
P.S. Sorry for my not-so-good English, it's not my native language and I know it's poor but I hope - understandable.
Greetings to all the developers and members of this forum, I hope the LibrePilot will continue to improve, it's a great project!
-
Hi Jdl,
The artificial horizon in OSD goes crazy and it takes several seconds to show correctly again.
Vibrations.
Maybe due to vibrations and Accelerometers go crazy.
Maintain firmly the frame on ground and monitor the horizon behavior while apply throttle. Be safe !
Solutions:
- Add filtering, go to Attitude tab
- Add adhesive foam on FC case to isolate from vibrations
- Balance props.
-
I've seen the same to happen when sitting on the ground, motors off, no vibrations at all. Never observed crazy horizon in Complementary mode, only in INS13. No matter in the air or in the ground. I've no record of OPLM telemetry showing the problem when sitting still on the ground, I'll try to make one.
P.S. Props and motors are balanced, revo is mounted on a 3-4mm foam pad. I'll try some filtering... Still suspect the loss of GPS data (and maybe momentary jump of reported GPS coordinates to some distant point).
The problem is not reproducable intentionally, playing with throttle doesn't affect the horizon at all.
-
Horizon moving a little while steady is normal.
See https://librepilot.atlassian.net/wiki/display/LPDOC/INS13+-+GPS+Navigation
-
Sometimes, while Copter is standing on the Ground and Motors are disarmed, the Horizon jumps around for an second and goes back on normal, but I dont know enough to figure out what is happening.
I put the Log down here, it happens to the end.
I replay your log, at end your GPS do a big position jump around 4meters maybe that cause the spike in attitude estimation.
There is two traces in screenshot, the raw GPS trace and estimated position from INS13
-
I managed to dig an example in my telemetry records that shows how momentary 'No GPS Fix' causes horizon jerk. Here is portion of GPS Input data extracted to CSV:
date, Time, Sec since start, Connected, Data changed, .Latitude, .Longitude, .PDOP, .Satellites, .Status
2016-04-04, 14:39:59.422, 126.976, 1, 1, 425112365, 274762658, 1.5, 14, Fix3D
2016-04-04, 14:39:59.649, 127.203, 1, 1, 425112269, 274762389, 1.5, 14, Fix3D
2016-04-04, 14:39:59.799, 127.353, 1, 1, 425112177, 274762121, 1.5, 14, Fix3D
2016-04-04, 14:40:00.12, 127.566, 1, 1, 425112086, 274761853, 1.5, 14, Fix3D
2016-04-04, 14:40:00.199, 127.753, 1, 1, 425111996, 274761583, 1.5, 14, Fix3D
2016-04-04, 14:40:00.367, 127.921, 1, 0, 425111996, 274761583, 1.5, 14, Fix3D
2016-04-04, 14:40:00.429, 127.983, 1, 1, 425111907, 274761313, 99.98999786, 0, NoFix
2016-04-04, 14:40:00.812, 128.366, 1, 1, 425108864, 274763301, 1.709999919, 10, Fix3D
2016-04-04, 14:40:00.884, 128.438, 1, 1, 425108864, 274763301, 1.709999919, 10, Fix3D
2016-04-04, 14:40:01.12, 128.566, 1, 1, 425108802, 274762996, 1.649999976, 11, Fix3D
2016-04-04, 14:40:01.209, 128.763, 1, 1, 425108738, 274762696, 1.709999919, 10, Fix3D
2016-04-04, 14:40:01.357, 128.911, 1, 0, 425108738, 274762696, 1.709999919, 10, Fix3D
2016-04-04, 14:40:01.409, 128.963, 1, 1, 425108676, 274762414, 1.649999976, 11, Fix3D
2016-04-04, 14:40:01.579, 129.133, 1, 1, 425108616, 274762138, 1.649999976, 11, Fix3D
2016-04-04, 14:40:01.799, 129.353, 1, 1, 425108557, 274761859, 1.649999976, 11, Fix3D
And a screenshot of Scopes page while replaying the log in GCS.
Flying on gyros only at this moment so nothing unusual in AccelState scope but AttitudeState shows clearly the wrong estimation.
-
... and another screenshot of that exact moment, with "Raw GyroState" and "Raw magnetometers" scopes, if this can help.
-
That clearly a GPS glitch.
14sats, no fix and next only 10sats.
-
I had a look at your GPS unit
http://www.banggood.com/UBLOX-NEO-M8N-BN-880-Flight-Control-GPS-Module-Dual-Module-Compass-p-971082.html
and it says
Receiving Format:GPS,GLONASS,Galileo,BeiDou,QZSS and SBAS
Just out of curiosity, can you try disabling all other sats and using only GPS ? Additional check, does it happen when flying in the clear area (without any walls/buildigs in proximity of ~200 meters).
My colleague experience similar jumps on M8N that uses GPS and GLONASS, but he didn't have the problem with M6N which uses only GPS. Also I was told walls can confuse GPS by reflecting signal and creating contradicting trajectories with different timings, so you end up with jumps.
-
Yes. But I think it'd be a great improvement to address such GPS issues (detected by NoFix or PDOP > xxx) in the Revo firmware, if possible.
When in autonomous flight, I cannot guess something that can help. But when flying in Manual Auto-levelling mode, and using INS13GPS, and when such GPS glitch is detected, isn't it possible to switch for some seconds to Complementary and use simple attitude estimation based on accelerometers and gyros only?
-
I had a look at your GPS unit
http://www.banggood.com/UBLOX-NEO-M8N-BN-880-Flight-Control-GPS-Module-Dual-Module-Compass-p-971082.html
and it says
Receiving Format:GPS,GLONASS,Galileo,BeiDou,QZSS and SBAS
Just out of curiosity, can you try disabling all other sats and using only GPS ? Additional check, does it happen when flying in the clear area (without any walls/buildigs in proximity of ~200 meters).
My colleague experience similar jumps on M8N that uses GPS and GLONASS, but he didn't have the problem with M6N which uses only GPS. Also I was told walls can confuse GPS by reflecting signal and creating contradicting trajectories with different timings, so you end up with jumps.
The issue with NoFix happens regardless of the flight area. Last time is was on the open field with no buildings or trees or anything kilometers around.
I definitely suspect this is a M8N firmware problem. Not sure if M6N did the same, because I was flying then with classic CC3D and used GPS only for OSD telemetry. But I have some memories that NO GPS FIX have also occured during flights in open areas with M6N.
I'll test different settings with M8N: GPS, GPS+GLONASS, EGNOS on or off, SBAS used for ranging, integrity, corrections and combinations between them. Of course, when SBAS is used for integrity, GLONASS cannot be used.
Still, regardless if this is M8N speciffic issue or not, it would be nice to have a failsafe NoFix workaround for attitude estimation in INS13GPS and manual flight modes.
-
I found my old notes on M8N GPS I tested some time ago. I have 3 of those units, tested only two.
Got lock immediately after setting outside of window, 13 sats after 5-10 seconds in view and lock
- Scopes show 8-9 sats used
- No earlier training for 20min was done
- Green path on the map after enabling diagnostics shows small interval green path around window of my house, 44min no spikes
- I observed twice satellites dropped from 9 to zero on scopes, fix was lost and recovered immediately to 9 sats, but no spikes on the map
So either I haven't tested long enough for it to happen, or maybe it only happens in flight ? I haven’t flown those GPS units successfully yet for other reasons though. At least I didnt notice couple of meter jumps in a bench test.
Still, regardless if this is M8N speciffic issue or not, it would be nice to have a failsafe NoFix workaround for attitude estimation in INS13GPS and manual flight modes.
Might be a good idea to have some additional stronger filtering/smoothing before it enters EKF...
-
I had better luck with "GPS only", but it did not fix the issue completely.
There are some things that could be done when the jump is detected, and you would not even need to leave INS13 mode.
The jumps I have seen have been associated with NoFix, and the GPS position looked normal before the jump, including having plenty of satellites, then jumped up to 200-300 meters away on the next GPS update (with 0 sats during the jump update IIRC), then quickly move back toward the original (correct) location over the course of the next 5-10 updates. This is not an INS13 issue because "Show Diagnostics" also shows it. In flight, even in non-GPS flight modes such as ATitude are affected, the level goes bad for the several seconds it takes to recover.
I just now created a tracking to address this issue.
-
Hi, in the last months i was pretty busy.
Is there something new about this issue?
I really want to go flying my Copter, but i cant, because it crashes.
Do I have to buy me an other Controller or is there something going forward with Libre?
Im pretty sad about this.
-
I reread your posts earlier in this thread and downloaded the log you posted. As f5soh said, there is a big GPS position jump just before the end. There is an issue open ( https://librepilot.atlassian.net/browse/LP-279 ) to try to find a way around the bug in ublox's m8 gps firmware that causes this position jump but nothing has been done yet. I've only heard of this problem with the newer M8 series gps, the older M6 and M7 gps were more reliable. Both of the GPS I fly with are older M7 ones and I don't seem to have the problem.
Eventually someone will code something to protect against the error in the M8 gps units but my best short term suggestion would be to get a M6 or M7 gps.
-
I have a 450 quad with a M8 series GPS that I've flown about 40 - 50 flights of 15 min each. I haven't seen this problem.
I get a GPS lock first time of the day in less than 3 minutes.
Rest of the day it's less than 1 minute.
This is the one I have:
href"http://www.ebay.com/itm/301711214747?_trksid=p2057872.m2749.l2649&ssPageName=STRK%3AMEBIDX"
The only time I've seen a problem is when I first started using it.
I was using the offset of Roll=180, Yaw=180, Pitch=0 to get zeros between onboard mag and aux mag.
It flew OK but, occasionally it would loose Yaw control when flying sideways.
I changed the offset to Roll=0, Yaw=0, Pitch=180 and that stopped.
-
The GPS problems only affect some people. The only factor that's known to make it worse is flying near a building or house where the GPS signal bounces off the building and seems to confuse the GPS. In wide open spaces they seem to work more reliably. The M8 seems to be the only one that has the big jump in position, the M6 and M7 don't seem to have a problem.
-
Old thread, but I'd like to share the solution that works for me.
1. I disabled SBAS (not sure this helpd too much, but still seemed to decrease the rate "No Fix" occurences).
2. Set Dynamic Model to 4G. I suppose this is the key! In 4G - no or very little filtering performed by GPS itself. And no need this filtering to be done in the uBlox module, as the EKF that runs on the Revo board does the job better.
I've flown many hours since I changed this setting and I hadn't experienced the glitch anymore. No big jumps in position or missing GPS data. No sudden tilts or jerks. No "No GPS Fix" alarms.
-
Do you find that PositionHold "wanders" more with it set to 4G? I assumed that would be the case, so I have never tried it.
-
Old thread, but I'd like to share the solution that works for me.
2. Set Dynamic Model to 4G. I suppose this is the key! In 4G - no or very little filtering performed by GPS itself. And no need this filtering to be done in the uBlox module, as the EKF that runs on the Revo board does the job better.
I've flown many hours since I changed this setting and I hadn't experienced the glitch anymore. No big jumps in position or missing GPS data. No sudden tilts or jerks. No "No GPS Fix" alarms.
Hi Jdl and FeuerVogel666, Thanks a lot for bringing this issue to attention. I have been struggling some 6 months with the same thing.
The only change I did from the default was to change the GPS Dynamic model from 1G to 4G, the rest is the same. The result is profound. No crazy 360 degree rolls once in any 6 minutes of stable level anymore. I have only done this with test for one day but I feel its the solution and will report back here if I have other experiences these upcoming flights.
And for TheOtherCliff, no the position hold is rock steady! (will let you know if subsequent flights show different experience).
-
Do you find that PositionHold "wanders" more with it set to 4G? I assumed that would be the case, so I have never tried it.
Sorry for the late answer!
Yes, I think the PositionHold is more unstable now. It maybe depends on current conditions, sometimes no "wanderings", sometimes - a lot of, but I'm ok with this as it saves from large jumps in GPS position.
Yesterday I had an unexpected (and still unexplainable for me) crash. I opened a new topic for the problem, although similar to this. But I'm sure the GPS is not the cause for it. Here is the topic:
https://forum.librepilot.org/index.php?topic=2449.0