Pixhawk with DJI E800 tuned Propulsion System 2nd Flight

Well I did the PID tuning procedure for the PX4 side of the Pixhawk Flight controller and I must say I should have done this before my first flight. Actually the parameters were not that far off from the default parameters. I had to increase the Proportional values for the roll and pitch while lowering the values for the yaw channels. When I get the PID values nailed down I will post them.

I found something interesting that I did not experience with the Phantom controller. While adjusting the Proportional portion I had no problems but when introducing the Differential parameters for either pitch or roll my PVC landing gear vibrated violently, like a monkey’s arms flailing it’s own poop. So I had to add a cross brace to the gear, I suppose I could have put the brace higher up but this should do for now. It seems as if the Differential feedback occurs at the vibrational frequency of my landing gear.

I flew for about eight minutes using about 2500 mAh from my 10,000 mAh battery so flying time, with very conservative reserves, should easily give me 25+ minutes of flight time, probably more with a hex. My Attopilot current and voltage sensing board, necessary for batteries grater than 4s, seems to be very twitchy. When I adjust the battery parameter to indicate the proper voltage it changes considerably from one power-up to the next. I must take the blame for this since I changed the setup twice, meaning I soldered three times and desoldered twice, I think that the onboard chips must not have liked the heat so I guess I need to order another Attopilot board.

_DSC2240

I am already working on my next project, a hex-copter, using what I have learned so far. My dilemma is deciding whether to buy additional DJI motors and ESC’s. DJI basically is screwing around with me and won’t even consider replacing the motor that was fried by their faulty ESC unless I go through another seven week customer service hostage scenario by sending them the failed motor, good grief.  They did sent me a new ESC, by the way, at least I hope it’s new since it was not even properly packaged in their own sealed package???? They used the ESD safe envelope that I sent them the original faulty ESC, F***-DJI !!!!!! Their customer service sucks monkey arse !!!! Plus they don’t even have Battery Eliminator Circuits (BEC’s) with their ESC’s.

My Hex frame in progress:

_DSC2242

PixHawk Quad With DJI E800 Tuned Propulsion System First Flight

Well I didn’t wait for DJI to finish my customer service order to finish their process and purchased a new motor and ESC. The PixHawk open source project has turned out to be ultimately successful, but a very fragmented and frustrating path since the documentation is weak and scattered across several web pages. It actually does not help that the PixHawk has two separate competing branches. I think it would actually be beneficial if both branches worked together to produce a reliable and bug free platform from which to program the PixHawk autopilot.

Instead we have two very buggy and unreliable software paths, called flight stacks, that are necessary in order to properly program the PixHawk hardware. WARNING going the open source route is plagued with numerous updates that may, or may not, be steps forward. This is especially true since the two sides are competing for additional features instead of competing for reliability and robustness.

Well that being said, my first flight using the PixHawk and QGoundControl default settings using the DJI 350 model as a starting point was successful. The default settings use very conservative low values for the PID feedback settings and this shows in the video. The aircraft controls were very sluggish and slow to respond from error input such as the effects from strong wind gusts. Obviously more tuning is in order.

One of the things that I noticed with the PixHawk’s GPS is that it is much more sensitive. While I could get a GPS lock with my DJI Phantom placed near a window or patio door the PixHawk can lock on to five satellites a full twenty feet from any window inside of my apartment! I’m very impressed! Also note that while I had some trouble reliably controlling the aircraft, due to weak default PID input values, the aircraft still acted in a positive manner, even at the end of the video when the battery became detached and ultimately unplugged it still landed upright and unscathed! The first flight for the PixHawk were decidedly superior to my first flights using the Phantom autopilot, which all ended up with uncontrollable crashes and in the destruction of several propellers:-(

Here is an image showing My new creation with the DJI Phantom. Note the huge 10AH 6S battery that obviously needs a more robust mount than velcro and a single bungee strap to secure can provide.

_DSC2220 _DSC2219I’m actually thinking that I will use this prototypes acquired knowledge into making a hex-copter, since the extra weight of the battery eats up most of the extra power supplied by the motor-ESC combination. It was obvious from this first flight test that the hover power was actually at about 50% when adding additional weight of a camera/gimbal system will probably overload this setup, especially when used with this huge 10AH battery. DJI is actually sending me a new ESC, I’m still not sure if they are sending me a new motor and props, but I will find out Monday when the shipment should arrive.

Oh ya, I also have been spending a lot of time learning how to use and program my new radio transmitter and receiver, the FrSky Taranis Plus sixteen channel transmitter and the X8R SBUS receiver. Also an open source steep learning curve quagmire of fragmented sources of software ind information. Ultimately I am very happy with my choice, just be forewarned a lot of time surfing the web for weak and fragmented documentation should be expected.

Sounds like to me another trip to Home depot…

 

DJI E800 3510 350KV BLDC Motor Failure

Well I now looked more into the motor failure of my DJI 3510 motor and found some strange happenings inside.

_DSC6395

If you will notice in the above image that the wire failed about 3mm from the solder joint, wait, WHAT? A solder joint on a high current motor lead? Now I think that this is piss poor engineering! And also notice that the wires are not even twisted around each other to help make a low resistance joint. Instead the lead in wire was simply laid upon the other two coil wires (appears to be delta wound) and as such the current must travel directly through a large portion of relatively high resistance solder, and no this is not silver solder since it melts at a relatively low temperature. I also noticed that the other two windings solder joints were also obviously over heated since the heat shrink tubing was also very well cooked. Obviously this is the weak point in this motor.

Good grief, I don’t know what the standards are in the Radio Control (RC) world are but crimp connectors should have been the obvious choice here. Silver solder may be acceptable but only if the joint can be annealed after soldering. (This is why using solder joints in modern aircraft is highly frowned upon) There is also a complete lack of any strain relief making highly susceptible to failure from vibrations in these wires so be damn careful when handling the lead in wires to this motor folks. I now don’t know if the motor connections failed causing the ESC to become smoked or did the failure mode of the ESC overloaded the motor causing it to fail, chicken and egg conundrum. The motor was operating fine before I removed it and the bad ESC, so it must have broke while handling the leads during removal.
_DSC6396Now I will be able to repair the connection and slop some new Gliptol on the burned windings to create a workable motor but this one will not ever be supporting anything valuable like a heli or a camera. I’m sure that portions of the windings have to be hardened and brittle to the point that it will most likely fail again quite prematurely.

I also don’t know why the wires failed where they did, perhaps the wires were nicked causing a high resistance choke point, I just can’t say for sure.

Update, 2015-03-22:

Well, I reattached the lead in wire and with the exception of one of the leads being shorter it functions perfectly. I even did a “Identify” from the ESC Assistant software and it passed all tests, even with the crispy overheated wires. Again this motor will never be used as a heli motor, perhaps I will use it to do some development as a position control motor for something like a gimbal motor, we’ll have to see…

WARNING DJI 620S ESC May Also Destroy Your Motors!

After sending in the faulty ESC to DJI I also found out that the ESC destroyed the motor it was attached to. Here’re some pics showing the overheated lead-in wire to the windings as well as the video demonstrating the failure mode once again:

Humor allert: Six weeks turn around time? I think that the people who say the DJI customer service is actually RTC whereas the normal failure mode of a DJI drone is RTH (Return To Home) It actually takes six weeks to to enable the real failure DJI mode of Return To China (RTC). HA!

When the sh*t hits the fan ya gotta relay on humor, after all.

_DSC6391 _DSC6392 _DSC6393

DJI on the Outs

Well, its been over a week since I sent my defective DJI 620S ESC back to DJI customer service technicians and have not heard one word back, even after unanswered inquiries asking for the status of my order. DJI seems to have forced my hand and I now have a PixHawk flight controller with GPS and telemetry radios on order :-) My only regret is buying the DJI “tuned propulsion system” in the first place.

Now I am in a dilemma, do I wait for DJI to replace the ESC or will they reject my request to replace it and the six f’d-up propellers. Actually I think instead of buying a replacement DJI ESC, when an individual ESC actually becomes available anywhere in the USA, I will replace all four ESC’s with another brand. (I’m not sure why Atlanta Hobby, a DJI recommended supplier, can’t break open one of their kits and then have DJI replenish the ESC.)

Like I said in a previous post: In the drone world, drone flyers seem to fall into two camps, DJI fans and DJI haters. Reluctantly I seem to be gravitating towards the DJI haters crowd, for very good and realistic reasons.

I have given up on my DJI Frankendrone and am currently putting back together the Phantom 2 Vision with its prop guards for the sole purpose of practicing and polishing my manual flying skills. While my newest creation will be used for actual photography work, at least after extensive testing and optimizations. The DJI will be set up with current software and their  spyware programs will then be permanently deleted from both my mac and Windoze partitions.

I’m looking forward to learning a new Real Time Operating System used by the 3DRobotics people who have developed the open source code for the PixHawk. While DJI has introduced a Software Development path they only make available the highest level layers to the developers, the major portions of the code are still unavailable to the public, whereas, with the Pixhawk there are two separate open source branches with every single bit of source code made available to anyone interested.

Sorry, but I simply can’t get myself to trust any communist based Chinese state run corporation, no matter how popular they have become here in the West.

Update 2015-03-19:

Well I just received an email from DJI support saying that they have no information a for me at this time and that their turn around time will be six weeks! GOOD GRIEF! Does anyone out there want to buy a cute little one owner DJI E800 Tuned propulsion system less one ESC and two good  props? Will sell for cheap!

DJI Software IS Spyware

I thought that I would install the most recent versions of the DJI software on my windows partition and then do a virus scan on this partition so as to demonstrate the ClamXav anti-virus software warnings of the trojan horses that are reported. I did some research on the web as to what these actually are and they are quite disturbingly potentially very dangerous. They potentially disable any installed anti-virus software. (this is why they are detected from my macbook scanning my windows partition but NOT from within windows) and allow any system settings to be changed without user validation as well as allowing ANY hardware (including DJI hardware) to be updated in the background without any user knowledge.

I also read many people who adamantly defend DJI claiming that this is necessary because any software that accesses or updates hardware firmware will also trigger the same alarms. THIS IS A FALSE CLAIM!!!!! I have numerous programming software packages that are used to program various hardware and microcontrollers and NONE of them trigger any trojan horse or virus warnings.

The anti-virus was updated to the latest versions, here is a screenshot of the results:

clam_screen

Both installer and program executables trigger two different trojan horse warnings so BEWARE!!!

DJI 620S ESC Commutation Test

While I wait for feedback from DJI customer service concerning my defective 620S ESC I thought that I would do some testing to see if the 620S is indeed a true sinusoidally commutated motor controller. Here is a short video demonstrating that it indeed is a true sinusoidal controller that should be measurably more efficient than most, if not all, other multi-rotor BLDC motor controllers that use trapezoidal or any other commutation, at least in the consumer RC world.


Update 3-16-2015 for clarification:
I would like to point out that once started the commutation of the BLDC motor, by necessity, operates in a closed feedback loop mode. (sensor or sensorless feedback) The PID speed control portion of the motor controller (ESC) may, or may not, operate in an open loop mode. However, in a multi rotor helicopter application the relative speed of the motors, which directly affects the attitude and positional changes of the aircraft, ultimately form an indirect closed feedback loop to the Inertial Management control portion of the flight controller. Remember that a multi rotor helicopter is inherently unstable and could not fly, or be controllable, without operating in a closed feedback loop mode.

For some really good background information see Microchip Application Notes:

AN1160
AN1078
AN1017

For comparison here is a snapshot of a DJI Phantom motor being commutated. The voltage wave is more of a triangularly shaped waveform than either trapezoidal or sinusoidal and has several discontinuities,. At any rate is is significantly dirtier than the DJI 620S ESC:

TEK00001

E800 Powered DJI Phantom 2 Vision Drone

Since getting my Phantom 2 Vision last year I have enjoyed some great fun learning how  to fly it. I was originally dissatisfied with the cameras image quality and hoped to be able to attach a much better camera but found that even a small increase in weight, of about one pound (about 450 grams) of my Nikon 1, it was a bit unstable and had some flight characteristics that were undesirable, at least for carrying around an expensive camera.

I was planning from the start simply to evaluate the Phantom with the idea that the Phantom was simply an entry level point with my sights set on a larger model but I became very hesitant. This was due to there being a lot of chatter that the FAA may outlaw these drones sometime in the near future, thankfully with the recent FAA Notice of Proposed Rule Making this does not seem to be the case. Search for “Drone NPRM” if you wish to read the proposal. Last year the Minnesota Legislature was actually testing the legal waters by floating a knee jerk bill that would have outlawed all RC aircraft. Thankfully there was enough resistance that they tabled the ridiculous piece of legislation, at least for the time being.

So I recently started to re-explore the idea of putting together another more powerful drone. Yes, I would like to get my hands on one of those very pricey ten-grand drones but I can’t see spending a fortune on equipment that cant even handle a Nikon camera. For some unknown reason to me nobody in this hobby seems to have any interest in supporting Nikon products, at least at the consumer end of the spectrum.

So how do I take my Phantom and provide it with the additional power that could carry something like my Nikon D800? Simple, make a new frame that is capable of carrying larger motors with larger propellers. Unfortunately the Phantom uses a proprietary Flight Controller that can only be minimally configured compared to other controllers. The only configuration available is a Quad V and in order to carry a large camera it needed to have a recommended take off weight of somewhere around five to eight pounds (2-4 kg).

Thankfully DJI has a new motor and controller that can provide a quad with exactly that specification, The E800 Tuned Propulsion System consisting of 3510/350kv BLDC motors and the new 620S Electronic Speed Controller (ESC) using 13.5 inch propellers. The same setup used on DJI’s new Inspire 1 quadracopter. The very interesting thing about this new controller is that DJI supposedly (testing will follow) introduced something that I wanted to do with the original Phantom, that is create a motor controller that commutates the BLDC motor sinusoidally rather than using the square or trapezoidal commutation found in most every other RC controller. Sinusoidal commutation should provide much more efficiency and should reduce vibrations considerably. (I’m wondering whether DJI engineers read my suggestions from my previous blog posts)

The E800 Tuned Propulsion system was placed on order, I then needed an airframe, buy a DIY ready made frame or build one myself. The quickest and cheapest route was to simply make one myself. I could have made one from aluminum sheet metal stock or even tubing made from aluminum or even carbon fiber but this would have been much costlier that a ready made frame. I opted to use wood for this prototype since I should have an excess of power and with the new motors weight became much less of a factor. To my surprise my frame (less landing gear) weighs only five ounces more than the plastic DJI Phantom frame. For those of you laughing at my red neck approach of using wooden sticks let me tell you that wood is an ideal material for aircraft, it absorbs vibrations a hell of a lot better than carbon fiber and it can take a hell of a lot more abuse than any carbon fiber airframe. For reasons that will come later I crashed this thing very hard four times on my initial flight testing and it came away unscathed, six plastic props were destroyed, however.

_DSC6379

If you are interested in using wood for aircraft structures here is a very good source that can get you started, it is from the FAA publication: FAA-H-8083-31 Aviation Technician Airframe Handbook Vol. One, Chapter six. Note that there are only a few species of wood suitable for use in aircraft structures, Spruce is the preferred species but luckily Home Depot sells furniture grade Poplar 1×2’s so the total cost of the airframe was under $10 US dollars!

Aviation Technician Airframe Handbook

I simply created my X-frame with a glued-and-screwed half-lap joint with 1/8 inch balsa wood gussets glued to the center providing some additional strength and platforms to mount the hardware and electronics. To mount the GPS and its antennae I created a platform up on a wooden dowel rod with a section of unetched circuit board material to provide some shielding from the electrically noisy electronics and motors, the Phantom uses a copper sheet embedded in foam tape to isolate the GPS. I simply used a aluminum sheet stock  to mount the Phantom camera. I resurrected the extended landing gear that I built for my Phantom with my roll axis gimbal platform. I also temporarily mounted the magnetic flux sensor way up on top of the wooden dowel until I find the best magnetically quiet place to mount it.

_DSC6316_blog

I still had one dilemma, and that was powering this Frankenstein drone. The Phantom is powered with a 3S 11.1V battery and the recommended voltage source for the motors is a 6S 22.2V battery. Since I have no idea if the Phantom main board can handle or regulate the 22.2V 6S battery I powered it from one Phantom battery, I then powered the motors and controllers with an additional Phantom battery wired in series with the first battery. The biggest problem is that the Phantom battery connectors wont lock without being installed in the Phantom frame. Additionally the Phantom main board relies on information supplied by two spring loaded contacts just below the power connectors. I used some Velcro and a chunk of plastic to hold the connectors in place, not ideal but this is only a prototype – after all. The second problem relating to the batteries is that the Phantom will not allow the motors to arm without a genuine Phantom battery with a secure data connection to it, there are NO other choices for powering the Phantom. (I may look into this more closely for a possible future hack.)

One other problem with the Phantom’s “Intelligent” battery is that if there is no current draw it will shut itself off within five minutes, potentially a very serious problem if the battery decides to shut itself off in flight! So far in my initial testing the battery won’t shut itself off as long as the motors are drawing sufficient current. This is something that definitely needs to be addressed.

One last item needed to be dealt with, and that was the LED indication lights used to communicate the Phantom’s flight modes and status. They are located on the Phantom’s ESC, since I couldn’t find a normal Naza USB/LED for sale anywhere I simply de-soldered the motor from one of the Phantom’s ESC and tied the ESC to the dowel rod supporting the GPS with the LED’s facing towards the rear.

After installing the new motors and ESC’s I did some initial testing of the motors, everything seemed to work flawlessly the motors all turned in the right directions and acted properly to elevator, aileron, rudder, and throttle controls. Time to take this beast out and flight test it, I used initial gain settings 50% greater than the Phantom’s default settings.

Initial take off was uneventful, it rose to about two feet off of the ground and hovered in place wondering around fifteen to twenty feet from its take-off point in GPS mode. I gave it some control inputs to bring it back to its take-off point and it started to loose altitude so I gave it some more throttle and all hell broke loose! It took off on its own, I tried to counter the beast and was able to slow it down somewhat but unfortunately it was already loosing altitude fast and heading for some trees so I chopped the motors off. I had to watch the thing pile into the ground. There goes two props just like that. Needless to say I was disappointed, I thought that the Attitude gain setting may have been too high and the basic settings too low due to the wondering in hover mode.

To make this story short I tried several different gain settings and while the characteristics did change noticeably as expected, every time I gave it some more throttle the damn thing would haul ass in some indeterminate direction followed by an uncontrollable crash into terra firrma. After six of eight of the propellers were trashed I was done flying. So I ran the thing on my bench without the props to see if I could isolate the problem which soon became obvious. Every time I started the motors everything seemed just fine, again all of the motors spun up normally and responded correctly to the radio controller inputs, and good grief!  Suddenly everything that was wrong became obvious! After applying full throttle and allowing the motors to spin up one of the motors became uncontrollable at full throttle. I could shut the other motors off normally with the radio but the only way to shut off the renegade motor was to kill the battery power. OH WHAT A RELIEF! IT’S NOT MY FAULT!

After doing some basic trouble shooting and swapping ESC’s, motors around and connecting the ESC’s to varying positions on the Inertial Management Unit (IMU) the problem was obviously with the new 620S ESC. Now will Atlanta Hobby, the place where I bought it, replace the ESC and the six broken props? The answer was no, they referred me to DJI, awe crap, from what I have read all over the internet DJI customer service was supposedly nonexistent. In reality I am in the middle of my dealing with DJI and I found the customer service to have been extremely polite and responsive, They requested I send the ESC back to them and they will replace it and the broken props under warranty, will they actually come through? We’ll see since I just sent it back to them today.

To be honest I have been considering some serious issues with DJI and lately have been exploring other avenues and found that there are basically two camps of drone fliers, DJI fans and those that despise DJI. The biggest advantage with DJI is that they have some very good well tested and very reliable, but expensive, platforms that work very well right out of the box. The major problem is that they are basically closed source so tinkering or modifying any DJI product becomes very difficult, if not impossible with any level of success. They also have some disturbing practices, mainly with maintaining absolute control over the use or usability of their products.

I updated my Phantom iPhone app but did not use it for a week and when I did it would not allow me to enter my registration info and locked me out of the program displaying a irritating message that program registration timed out, the only way to get it running was to delete the app and re-download it, and of course register it right away. Luckily, I was home and had the registration info handy, if I had been on the road or away from cell coverage I would have been totally screwed. Additionally I saw a YouTube video review of their new Inspire 1 platform and the reviewer noticed that his drone would not arm until he updated his firmware. So DJI maintains the capability to disable their products at will! Another area is in flashing their products, it must be done with an internet connection and with only the firmware update that they choose, the last update does not appear to be retained on the local computer, again if there is no active internet connection you are screwed!

Additionally I found that when scanning my Windoze partition from my dual boot macbook using ClamXav anti-virus program it flagged and quarantined all of the DJI software on the Windoze partition as spyware. I’m not sure if it was a false warning, but you have to wonder…

Alas, I’m actually seriously looking at some open source alternatives that are very popular and well supported, like the PixHawk flight controller, which is much much more tunable and is at least as reliable and robust as any of DJI’s product line. A lot depends on my customer service outcome with DJI. DJI does have one advantage with their very expensive A2 controller in that it can make a hex or an octo-rotor easily controllable with a motor failure. But with the open source PixHawk writing control algorithms are easily possible, if an accurate state space model can be successfully created for it, sounds like a challenge to me;-)

So I’m stuck waiting for another ESC, I will report back if I find that my Frankendrone becomes a reliable flier.

CGEM Telescope Mount

My posts concerning the CGEM telescope mount, curiously, continues to be very popular even though many of them are several years old. By back checking to the referring sites I can read the forum posts which links to my blog pages. I find it very interesting that there seems to be several persons vehemently disagreeing with my conclusions that the CGEM is not an acceptable mount for anything but introductory amateur astrophotography or strictly for use with wide angle lenses.

The interesting thing is when I travel to other posts giving advice (from the very same posters) to any would-be astrophotographer they nearly always advise against the CGEM and similar class of mounts. Instead they nearly always recommend  much more expensive AP and similar class  mounts for any serious astrophotography. The very same conclusions that I arrived at. Actually I also recommended  a much cheaper route which is buying time on remote time-sharing systems which tend to use laboratory grade equipment that can be used for scientific purposes other than strictly making pretty graphic arts pictures.

There are several issues brought up with my criticisms. First, they seem to confuse non-orthogonality between the optical tube and the mount axis, called cone error, with non-orthogonality between the axis. Very different animals, cone error can be corrected, inter-axis orthogonality can only be properly corrected by line-boring in a machine shop.

Second, they seem to think that I was doing my testing with an overloaded mount. Wrong! My 9.25 optical tube was on the ground. When I did my testing it was loaded ONLY with the Orion auto guiding scope which weighed less than ten pounds.

Third, my criticism of comparing the mode of operation of the feedback control DC motors to that of stepper motors. Yes, many servo motors, DC and otherwise, use pulse width modulation to produce an effective DC voltage less than the power supply in order to control motor speed. These pulses are very often several hundreds to several thousands of pulses per second, outside the vibrational frequencies of the mount and scope. My point is that you CANNOT use speed control below the effective voltages in which the CGEM servos operates during guiding. Instead the motors are subjected to strong pulses every second or two, depending on the exposure time of the guide camera. These pulses cause detectable vibrations that reduce resolution, PERIOD!

Hence the usability of imaging unguided with PEC corrections during longer exposure times is a VERY GOOD INDICATOR OF MOUNT QUALITY, or in the case of the CGEM the indicator of the LACK OF QUALITY. The CGEM is NOT PEC CORRECTABLE, Celestron is NOT working on the so-called 8/3 error, despite the continued misinformation supplied by such web fanboy forums.

Fourth, the resolution of any lens, or antennas used in radio telescopes, is dependent on the aperture diameter and the frequencies being observed. So when a 2.5 inch camera lens produces higher resolution than a 9.25 inch diameter optical tube that is supposedly optimized for astrophotography – that is a VERY BIG DEAL!

I could go on and on but the facts remain the same, arguing with fanboys only leaves you as frustrated as trying to use the CGEM mount for astrophotography. Like I said many times before the CGEM is a superlative mount for visual use, but for astrophotography you must lower your expectations or look elsewhere, and that just about says it all.