Syma X25PRO loves freedom... and flies away!

Discussion in 'Getting Started' started by Manfred, Sep 13, 2018.

  1. Manfred

    Manfred Member

    A while ago I did the first flight of my new (and first ever) drone. And I ran into a problem that puzzles me.

    I prepared everything as best I could: Fully charged drone battery, fresh alkaline transmitter batteries, I calibrated the sensors (gyros/accelerometers, and compass) according to the procedures outlined in the manual, watching for the proper light flashing codes, then confirmed (also by the drone lights) that the GPS had locked in and was ready to fly. Then I took off, using the automatic takeoff function.

    The drone took off fine - but kept climbing slowly, and drifting sidewards, slowly too. I could easily compensate for that with the sticks, but I had expected a GPS drone to be more stable than that. It drifted sidewards, front and back, up and down, as if the slight breeze with its turbulence was affecting it, and I couldn't see nor hear any automatic action to counteract this drifting.

    Anyway I flew about 20 meters forward, then back, noticing that it went forward far easier than back. Also I tried some sideward flying, this too had a bias, going much faster to one side than the other. I didn't try trimming it at that time, since the drifts were easy to compensate for. Then I rotated the drone over is axis, that went fine. Then I flew an extended circle, by giving it a constant forward speed and controlling its heading. Several times during this I had to correct the altitude, because it kept drifting up and down, but mostly up.

    And then, suddenly it stopped responding to control input! It kept hoovering, drifting up and sideways, and there was ABSOLUTELY no reaction to anything I did! After noticing that it didn't react to the sticks anymore, I pressed the return-to-home button, to no effect, and then the automatic landing button, to no effect either. I could do nothing but watch the drone climb slowly and drift away ever faster, out of the field and toward a forested hill. I ran after it, trying not to lose sight, repeatedly trying to activate the return function, and trying to get any control by the sticks - no reaction. Finally - and fortunately - it had an encounter with a treetop at that forested hill, and fell about 30 meters down through the tree.

    The ground of that forest section is thickly covered in blackberry and other thorny plants, so I had to get the machete I had in my car for precisely such needs, and hack my way to the place where the drone was. At first I couldn't see it, but by repeatedly arming and disarming the motors via the remote control and listening for the prop sound I finally found it, hanging low in the tree with one prop blocked by a twig.

    The good news: It suffered absolutely no damage.

    The bad news, of course: It's useless if it loves to just fly away!

    As you surely noticed, the control was working fine after the crash, allowing me to arm and disarm the motors. And at the time it stopped working in mid flight, the drone was at most 30m away from me. After retrieving the drone I did a range test on the ground, by arming/disarming the motors several times while walking away, and I had solid control up to a distance where I could barely see the drone, easily 100 meters. So it clearly wasn't a range problem.

    Now I need your help to try finding out what happened. I see some possibilities, but since I have no experience with drones, nor with 2.4GHz remote controls, I need help. I do have experience with "antique" 72MHz remote controls, because I used to fly model aircraft in the past, in the days of glow plug engines and digital proportional FM remote controls. I can't remember any case of any of my Futaba radio controls stopping to work in mid flight.

    My ideas, subject to your judging:

    - Could it be that since the drone was set to slow (beginners) mode, it just didn't have enough speed available to compensate for wind? Does that speed limit act before or after position control? I mean, is the limit a net limit, by GPS, or a gross limit in the differential motor speed?

    - Could it be that the drone drifted out of an "electronic fence", and instead of returning into it, stopped reacting? That would be a software bug, of course.

    - Or could it be that the GPS signal was poor, and the GPS position reference drifted, so that the electronically fenced area moved away? But even then it should have returned to whatever it thought was its takeoff point!

    - Could it be that the drone self-interfered in some way, for example by brush noise from its motors? When it loses the control signal it should return home automatically, which it did not. So the interference would have needed to be so bad that the GPS too lost its signals. External interference is almost impossible, because the location is far away from any interference source. Like 400 meters from the nearest power line, and 600 meters from the nearest house, which happens to be mine, at everything was off at home. The nearest neighbor lives more than a kilometer from that place.

    The place where it first lost control was significantly closer than the fartherst place I flew it while it still worked.

    The whole flight might have lasted for 2 minutes, and probably less.

    I don't know what to do now. If I take off again and it flies away again, there might be no kind tree in its path to catch it for me.
     
  2. RENOV8R

    RENOV8R Well-Known Member

    Sounds to me like faulty compass calibration. This needs to be done in an open area, void of any interference. This includes anything in your pockets, including car keys and cellphones, which are both excellent sources of magnetic interference. I'm not familiar with this unit but the other thing that comes to mind is that with some GPS quads, the GPS function is disabled in certain flight modes. This is less likely to be the issue than the first scenario, as it would be set out in the user manual, which I'm assuming that you have perused in detail.
     
  3. Manfred

    Manfred Member

    I did the compass calibration in an open place, without any cellphone, keys, etc in my pockets, and no watch. The nearest magnetic object was my car, about 40 meters away, surely not a cause for significant field distortion. I'm familiar with this kind of compass calibration, because I have several other gadgets that include electronic compasses. My only doubt relative to compass calibration of this drone is this: The manual says that after rotating with the drone in horizontal position until the proper light flashing code confirms that the first part is done, one should hold the drone in VERTICAL position and rotate it until the next light code. It does not say which end should point up - the front, the back, or a side, or whether it doesn't matter. I held it with the front up.

    I will redo the compass calibration before the next attempt to fly. But I don't think it was bad. Because the only stable thing was the heading! The drone would drift front/back, sideways, and up/down, but the heading didn't drift. When I flew that slow circle, it reacted well to my steering input. After loss of control response, I don't know whether the heading went crazy or not. I really can't see from a distance which end points where, and not having video downlink my only way to know the drone's heading has to push the right stick a little forward and see where it goes. When it stopped reacting, I stopped knowing where it was pointing.

    The manual doesn't tell anything about flight modes without GPS. Only that indoors the GPS won't work. But indeed some GPS problem might have been involved, because the place where I flew is surrrounded by hills, so there might be reflections causing the GPS fix to jump. But if that happens, I would expect the drone to either follow those GPS jumps, or disregard the GPS and go into a GPS-less mode - but I would never expect it to stop reacting to control input! And anyway, the worst short-term position drifting I observe on my handheld GPS receiver at that place involves a few meters.

    How good is the wind resistance of such low cost GPS drones, generally? Can I expect it cope with a slight but turbulent breeze? If not, that might be the explanation for the drifting, but of course not for the loss of control.

    I keep wondering if the problem was the SLOW mode. When such a drone is in slow mode, will that also limit the intensity of its reactions to drift, or will it only limit the the range of speed the pilot can fly it at?

    I will make the next test flight at some time when there is absolutely no wind. Not today, because today there is more wind than yesterday.

    And one more thought came up overnight: My area is located roughly 700 meters above sea level. That's not a lot, but the air is roughly 10% less dense here than at sea level. Could it be that the drone was struggling to stay airborne, and had its motors pretty much topped out, limiting its maneuverability? Maybe with the fully charged battery it had just enough power, but as the battery started to wear down, limiting the maximum motor speed, it ran out of reserve motor speed for maneuvering? Anyway in that case I would have expected it to lose altitude, not to climb! And it did climb. From loss of control to that tree collision it climbed about 15 meters.

    Many questions. I hope some of you can enlighten me, at least regarding what's more likely or less likely.
     
  4. Dugdog47

    Dugdog47 Well-Known Member

    If that was it's only flight you'll have to try it again. You never know sometimes these things just work fine on another session. I have an old ford that's the same way lol.
     
  5. Manfred

    Manfred Member

    I WILL try again, that much is clear. But right now the weather is horrible.

    I'm wondering whether to try it indoors, in the smoothest and emptiest room I have, or else try again outdoors but with a fishing line attached, so it can't fly away, just crash! ;-)
     
  6. Manfred

    Manfred Member

    I cleared out a room for indoor test flights. So I can rule out wind and GPS trouble. The results: No loss of control so far, but the stability is indeed poor. I can't do any long flights, no longer than 30 seconds or so, because of poor stability. Just after takeoff it hoovers almost stable, then starts drifting off slowly, to the right and front. At first I can easily compensate for it, but the drift becomes ever faster, until I run out of joystick range to counteract the drift! At that point I land, take off again, and the game begins anew.

    So it seems to suffer from a drift that increases steadily during flight (caused by heating? or battery voltage drop?), soon exceeding the speed range in the "beginners" setting. Interestingly this is the exact same drift I observed in my only outdoor flight.

    The altitude hold is also rather poor, I have to keep correcting the altitude to keep the drone from touching the ceiling or the floor. It will rise to the ceiling or sink to the floor within less than 10 seconds, after starting from a stable hoover.

    The heading is stable, at least. No drift at all in the heading.

    So, it looks like the compass works fine, the accelerometer drifts, the barometer drifts too, and it also seems that the GPS stabilization doesn't work at all, since the drift outdoors is the same as indoors.

    None of my indoor flights so far lasted as long as that one outdoor flight, so I still don't know whether the loss of control happens at a fixed flight time, a certain distance, or at random, or was a one-time event.

    From the point of view of an electronic engineer, the stability seems poor. But from the point of view of a complete beginner in drone flying, I really don't know what to expect. Is such a behaviour normal for a low cost GPS drone?
     
  7. RENOV8R

    RENOV8R Well-Known Member

    When testing indoors, you can rule out anything to do with the GPS, as it is probably not working due to lack of signal. If your drifting was constant, it could be easily corrected with trim. But the fact that it increases exponentially tells me that there's a bigger problem, maybe a hardware issue. You need to contact the seller and explain the issue.
     
  8. Manfred

    Manfred Member

    Today I made some more indoors test flights. I started with the battery fully charged, and tried to just keep it hoovering in the middle of the room for as long as possible. At first it wasn't hard, I only needed about one third of the joystick range, to the lower left, and make small corrections from that position. That lasted for about one minute, with the amount of drift correction required slowly increasing. Then the drone started to get very unstable very fast, and a few seconds later I could no longer keep it from touching a wall and crashing.

    I repositioned it in the center of the room and took off again. It started stable but became unstable just a few seconds later. The instability was not just horizontal, but also in altitude hold. I had to use the left stick a lot to keep it from touching the ceiling or the floor. Only the heading remained stable - it never rotates by itself. A few seconds later the instability exceeded my reaction speed, and it crashed against a wall again.

    I took off again right away. This time the drone shot up like a rocket, hit the ceiling hard and fell down.

    I switched it off and let it cool down. Then I tried again, and again it was controllable for one minute at most before starting to act up seriously.

    After letting it cool off again, I decided to try headless mode. It was a total disaster! The drone was totally unstable right from a cold start. Like: Switch it on, link it to the transmitter, switch it into headless mode, arm motors, push the left stick slowly upwards - and the drone takes off, and barely off the ground shoots to one side and crashes into the wall.

    After switching off the headless mode, it behaved like in the first flight.

    So I'm pretty convinced now that I got a bad drone. One thing is that the accelerometers seem to drift badly as the electronics warm up - and they do get quite warm quickly. Another thing is that the barometer either isn't working at all, or is having jumps in addition to drift. And it's quite obvious that indoors I'm getting the same behaviour as outdoors, which means that the drone isn't using the GPS data for stabilization! When outdoors, the LEDs do indicate satellite search and position acquisition, and indoors they indicate continued search for sats without finding them, and that's all correct - but outdoors the drone was just as unstable as it is indoors, which suggests to me that it isn't using the GPS at all.

    If I had bought it in a store, I would go there and ask for help. But I bought it over the internet, on AliExpress, and just trying to explain such a problem to a Chinese seller who doesn't understand English and is using Google Translate, is a nightmare.

    So I suppose that I will just accept my loss and put this drone in the trash. Which means a quick end of my voyage into drone flying, because I'm not willing to spend several hundred dollars on a drone that actually works as it should. And ordering another cheap drone from China seems to be too risky.

    I'm a professional electronician, and amateur programmer, but I don't suppose there is any way to get into Syma's firmware to try and diagnose what's happening. I will look inside anyway, just out of curiosity, but I don't expect that I can do anything.
     
  9. Dugdog47

    Dugdog47 Well-Known Member

    You should try to get a refund since the product is defective. They might even send you a replacement for free lol.
     
  10. RENOV8R

    RENOV8R Well-Known Member

    You are overthinking the matter. Any tampering with the "firmware" will just void any type of warranty you may have.
     
  11. Manfred

    Manfred Member

    I opened the drone to look for problems I might be able to solve. And that was a good idea! One of the motor freewheeling diodes had shifted during soldering, so that one of its leads wasn't soldered at all, and was completely in the air!!!!! It's funny that the drone flew at all. Just see the left side pin of the right SS14 diode:

    NonsolderedDiode.jpg

    The solder work is quite sloppy. There were many solder balls and whiskers, some sticking to the board, some sticking to parts (like the one visible in the photo, sticking to the left-side diode), and some flying around loosely. I also found a screw that was sticking to the self-adhesive shielding of the GPS module. This screw was extra, it wasn't missing from any position in the drone! So, now I have a spare screw... Such a loose screw or solder blob can easily cause intermittent short circuits, causing anything from erratic behaviour to fire.

    These problems prove sloppy assembly and nonexistant quality control.

    The GPS shield is a self-adhesive copper tape that was simply stuck all around the back of the GPS/compass module, over components and all. The only insulation was the copper tape's glue! That's not very safe, I would say... I added some kevlar tape between the shield and any points that had better stay insulated.

    The 2.4GHz antenna was supported only from its lead, and it's pretty heavy. Not a good idea - a shifting weight is a stability-damaging factor. I fixed it with two tiny dabs of hot glue.

    I thought that Syma was a well-reputed brand, making low cost but decent drones. Now I know otherwise.

    I wonder what effect the unconnected freewheeling diode had. Probably it caused the MOSFET that controls that motor to heat up far more than normal, which also must have shortened the battery endurance. The heat from the MOSFET might have reached the accelerometer/gyro and barometer, which are very close to that MOSFET, but sit on a vibration damper (foam and metal weight plate) which should also provide rather good thermal insulation. In addition, the lack of a functioning freewheling diode should cause more interference, but I don't know whether that affected the 2.4GHz receiver.

    Anyway, after correctly soldering the diode the drone behaves somewhat better, but still not good. The trim still drifts away, but slower than before. I can now fly for approximately 4 minutes continuously before running out of joystick range to counteract the drift. If I land at that point, and do a "sensor recalibration", which basically zeroes the accelerometer and gyro with the drone standing on level ground, I can again take off and continue flying for one or two more minutes - that's the point where the battery gets weak. By the way, the advertised flight time is 12 minutes. The true value is about half that. But this can be partly due to the altitude where I live. And the 12 minutes are probably meant to be until the moment the drone drops out of the air with a totally dead battery.

    While doing this I discovered another bug: When the battery warning comes on, the video recording stops! It can be restarted, but it's still a nuisance. I guess I will lose the footage of many landings due to this bug.

    I would say that a sensor drift so bad that after 4 minutes one runs out of joystick range to hold position is bad. Now I wonder if this is a feature or a bug. I have a very evil suspicion: All the chips I could identify, such as the BK2425 radio chip, the MPU-6050 accelerometer/gyro, the MMC5883 compass chip, are rated for an absolute maximum supply voltage of 3.6 volt or so. Nominally they should run from 3.3V or less. But both of the voltage regulators I found in this drone are 5 volt types, marked 117-5G, which must be 5 volt versions of the NCP1117. Could it be that by mistake they used a wrong regulator, and are powering the 3.3V chips from 5V????? With two regulators present, it would be very logical that one should be 5V and the other 3.3V.

    Unfortuately I forgot to measure the actual voltage those chips get, and disassembling the drone again, with its 36 tiny screws, just to measure the voltage, isn't something I would love to do right now. Also I don't want to wear out the plastic threads through too much disassembly.

    If anybody has an X25PRO that flies well, and happens to have looked inside, I would love to know whether that one also has two 5 volt regulators, or one 5V and one 3.3V!

    The point is that if really my drone was fitted with the incorrect regulator, and is putting 5V into 3.3V chips, that can easily explain all the trouble of sensor drift and even of chips suddenly stopping to work!
     
  12. Dugdog47

    Dugdog47 Well-Known Member

    Wow you are an electronic wizard! If what you suspect is true it probably happened on a large scale because those drones are known to have problems.
     
  13. Manfred

    Manfred Member

    I would love to get into contact with other people who own the same model of drone, or a similar one, to compare experiences and find out whether there is a generic problem.

    Looks like I will have to take it apart again, and measure if the 3.3V chips are really getting 5V, and if so, try to trace whether there is a regulator for the 3.3V chips and another for the 5V circuitry, and by mistake 3.3V regulator position was populated with a 5V one.
     
  14. Manfred

    Manfred Member

    I opened it again and measured the voltages. No problem there. The 3.3V chips actually get 3.3V, from a second stage of regulation after the 5V regulators.

    I made some more indoor flights and discovered that I need to do the sensor calibration with the drone NOT level, but slightly tilted to the left and back. When doing this, it's quite stable. Go figure... Probably there is some asymmetry in weight distribution or prop shaft alignment that requires this tilted calibration to keep it stable.

    Using the trim function instead is very hard. As soon as I activate the trim function, the drone drifts away so fast that it hits a wall before I have a chance to control it.

    Headless mode is crazy and uncontrollable, at least indoors. But that could well be due to the compass working very poorly inside my heavily steel-reinforced home. I need to test this outdoors. The weather here has been nasty for the last two weeks, so that will have to wait.

    And one more thing which is a downer: The camera, advertised as 720p resolution, actually has a much lower resolution sensor, and the image is interpolated to 1280*720 pixels. That largely explains the rather poor image quality.

    Why does Syma lie? I mean, this is a low cost drone. A 640*480 pixel camera isn't bad for an $80 GPS drone. But claiming that it's a 720p camera, only because they interpolate the images to that resolution, IS bad.

    The interpolation algorhythm works well enough on large solid surfaces, but is terrible with fine details such as tree twig structures.
     

Share This Page