This is my third crash (repeating the 'exact' same scenario deliberately) in the past two days. I have found the core problem of the issue -> Using camera (RPi camera module) on-board the flight wipes out GPS signal 'entirely'.
Here is my setup:
As far as I could see, I could (luckily) not find interference on 2.4 GHz of Txm/Rx and the WiFi (802.11 b/g/n) that I was initially expecting.
But to my surprise, what happens is as soon as I start shooting the video, the 'satcount' goes to 0 for about a few seconds and then starts getting back to normal. Originally I have satcount of 9.. drops quickly to 0 as soon as I start the video shoot and then slowly increases to 7 until video footage is completed.
When the satcount reduces, copter goes CRAZY (yes, literally crazy), 'does not land' but flies away to some remote corner and crashes before I could take back control (Log attached shows the glitch clearly). I cross checked that this happens only in LOITER mode when taking the video; Loiter uses the GPS. Shouldn't it land, when copter loses GPS fix (for 5 seconds)??
I did ground tests without flying, just commanding the video to start, brings down satcount to 0, as seen in Mission Planner.
I used Aluminium shielding, it helps quite a bit, but may make copter drift during flight video unreliably. The cause may also be the camera + camera cable.
I wonder how can this be possible in the first place. uBlox GPS uses 1.575GHz carrier wave and such a popular RPi camera module should come with ISA interference certificate/clearances, I am completely bamboozled!!! Another post issuing similar problem but with HK camera here.
Below is my latest setup, still some glitches in satellite count, but yet not the perfect solution. Glitches in satcount is 1 to 3 during video shoot. Did anyone faced this? Looking for a neat solution.
Any suggestions are welcome.
Thanks and regards,
Maybe it's not a signal issue but a power one. Perhaps the uBlox is losing power when the cam draws more power to record? How is the cam powered?
I agree with UAVStuff as it might be a power problem and not an RF interference problem. I have my camera/RF transmitter at the other end of my rover to help mitigate RF interference with the GPS.
TCIII ArduRover2 Developer
The cam is powered via one of the ESC (BEC, 5V). The GPS is powered from APM and APM is powered from 3DR power module directly from the battery in my quad. I do not see a reason for power issue (although, I do get your point).
I ran a lot of tests (at home) [14 separate tests to confirm], I see the glitch is only for a few seconds before the GPS regains its satcount while shooting video for even a minute. The ribbon cable of the camera is also a culprit, different position of the cable leads to different scenarios.
Having Aluminium in the middle of camera (and its ribbon cable) and GPS helps more than enough (but still small unknown glitches) though quite better and consistent. So, I think can't be a power issue, right?
FYI, right now on board, I have:
- 2.4 GHz Receiver (8CH)
- 900 MHz 3DR Telemetry
- 1.575GHz GPS ublox receiver
- 2.4 GHZ (Channel 10) WiFi (on RPi) running in ad-hoc mode with various routing protocols and algorithms for my research purpose.
@@ A Discovery @@
Found the precise reason for this problem. This can happen to anyone so be aware. Tomorrow I will be going for an EMC chamber test for the amount of EMI interference.
The problem is that the camera cable acts as a 'Dipole antenna' and radiates harmonics of certain frequencies, some of which touches 1.575GHz in milli-watts for certain duration. The signal received by GPS is around the same frequency but in micro-watts. This wipes out the signal. Brings the 'satellite count' as seen by GPS to 0 (no power issue here). This makes the copter crash as this situation is not yet handled in code !!
The key is to build a baby faraday cage for this cable and keep it at an angle (Fleming's left-hand rule) to keep the magnetic interference on a different axes that can prevent the interference with the GPS which is situated near and on top of the multi-copter.
In the end, practically carried out at least 25 tests to confirm this happening, all with similar results. Also, I moved the cable gradually at different orientations, keeping the camera and GPS positions intact - Giving sat count fluctuations.
I will try to put more concrete information on this soon.
That's amazing, good find. Never heard of this happening before.
Hi Shyam, I have a post in another thread where I make a reference to this exact problem. My particular camera is the one I use for FPV and is a SecurityCamera2000 PZ042 board cam that has been put into a metal case also sold by SecCam2000. The cam works great and the ONLY issue is the GPS problem. I have found that as long as the GPS module is a foot or so away from the uBlox GPS it will lock 11 to 14 sats. With the camera at 6 inches the uBlox will only lock 6 to 8 sats. I've also read the Go-Pro camera is noisy to GPS as well. When I use my Go-Pro on the quad it does indeed drop sat count though not as bad as the FPV cam does. With my FPV cam out on the front of the quad and the GPS module mounted far out on one of my rear arms I normally get at least 11 sats which seems to be good enough as I don't have any kind of glitches.
So, slowly this problem is coming to light.
I am sure many of us out there do not realize that (many) cameras do indeed drop the GPS sat count. When it drops the satellite count by a few numbers, people don't (won't) realize. Though, it comes to light only when there is a drastic change.
People who claim to be a (hero) when flying other's work autonomously over people's head? Ok, how would one deal with external interference? I flash a wave of 1.5 GHz on this quadcopter, and the hero is now liable for any life taken by his quad. I am saying this as people have not deliberated on this issue too deeply (from the scientific literature I have read till date) and this is a potential cause of things going unexpectedly wrong. So, do not underestimate it!! Note that all civilian GPS will use 1.5 GHz as the GPS carrier frequency (even if your quad is 50,000$). There is always a risk involved, so always keep an eye on your flying baby ;)
Since, the state-of-the-art autonomous flight relies (only) on GPS at the moment and it is the only technology available, it may be a good idea to find either:
I agree with this issue. I have verified I have the same issue. I am using a Midland Helmet camera. I have used a gimbal too, but just switching on the camera leads to GPS problems (3DR). I am in the process of looking at a solution. I have noted the on/off switch uses a magnet. While simply flying with the camera does not seem to be an issue with the GPS or mag sensor, as soon as I switch the camera on, I start losing GPS lock. I plan on testing a farday cage first. If that doesn't stop the problem, I'll start looking for a new camera. I do believe the Ion Pro with metal body has to be better at reducing the problem.
Old problem, mentioned numerous times on this forum.
Many video cams generate spurious signals well into the gps frequencies.
You can try to wrap the camera in tin foil, obviously leaving the lens and controls open :-)
I almost lost a quad due to a HobbyKing Wingcam, haven't seen any problems with my GoPro, yet.
Try fitting the gps on a mast, raise it away from the electronics, as you have it in your pictures.
Even better, get the gps/compass combo, cut the trace in the apm, and make an apm 2.6.
That has also been covered on this forum.
Good Luck !
Sure. This is simply the first time for me it was an issue. I do like the tin foil suggestion. That would at least provide a cheap way to see if I want to stay with the camera. If GoPros had this issue, I would have found more like it. I was thinking of an Ion Pro WiFi for light fpv duties, but that simply adds to the freq spectrum I have to be concerned about. I have already changed the rc receiver location along with my Bluetooth location to minimize any effects.
Over the past few years I have slowly rethought locations and configuration to provide the most reliable flight. The times between failure is much longer, but considering a good test will help me keep it that way. Now if I can figure out a way to keep the stupid pilot errors out of the mix too. I am moving to autonomous flight within sight to allow me to plan flights in advance. I haven't used the rally points yet, but I do like that option.
BTW, tin foil verified the camera interference issue. A rather short piece of flashing strategically set to block the GPS module direction worked well. I'll continue to use the Midland camera for now.