Discussions

Hi Guys,

Firstly, let me start with the obligatory disclaimer that I am a total noob to this stuff and am reaching out for some valuable advice from those much more experienced and knowledgeable than myself. My name is Jason, I am 31 and based in North Parramatta, which sits (near as makes no difference) in the geographic heart of Sydney, Australia.  I have an interest in geography and history, and all things geospatial and live adjacent to some of Australia's most significant built heritage items.  BTW - I don't facebook, I don't blog and this is my first forum post in a decade, but i thought it may be worth sharing my experience, and hoped to draw on the feedback of others to help guide me forward.

For some context, I have had a few electric gliders, I have never flown a heli heavier than 500g and the last time I held an RC transmitter the Revo 2.5 had just hit the shelves. (>6 years ago).

About 3 months ago I stumbled across this video.

I had to google what the FPV stood for but was immediately captured by the concept.  This kicked off a period of about 10 weeks watching countless videos, reading blogs and absorbing forum threads to effectively bring me up to speed on what is possible and what, perhaps, my first FPV airframe might consist of.

Initially, I had my heart set on building a solar assisted wing not unlike that which features in the video that inspired my pursuit.  Something that might be able to reliably stay aloft from sun up to sun down.  I came up with a design from the HS522 profile with a wing span of 3-metres (3000/900/600mm) and wearing 140W (17VDC@~8A) worth of PV which i call 'Mithra'. Renderings below.

There are also images of an early gimbal design for adapting JVC's venerable Everio series camcorder for aerial use (making use of it's excellent optics - 40x optical/70xdynamic/200xdigital zoom with both optical and digital image stabilisation.  This gimbal design was so early, i had yet to discover the virtues of brushless gimbal motors.

By the time I got to the point of resolving things like ground station requirements and designing a catapult capable of launching the 3-metre wide craft I realised that, considering my existing experience with UAV FPV tech and primary geographic location, i should completely reconsider my entry point.  After specing out a smaller but similarly complicated X8 with Range Video based electronics, I realised that even this was all too much to start with, so I ultimately decided that a multirotor would be a much more accessible platform upon which to build initial experience working with the various aspects of the craft.  Also relevant is the fact that I live just a stone's throw from a convict nations earliest and longest serving Goals, one of the nations oldest stone arch bridges, the first Government House, The People's Park/Governor's Domain (a UNESCO World Heritage listed site), one of the last intact fully landscaped Edwardian gardens, Australia's first observatory, and a bevy of historically and culturally significant structures designed by a respectable list of Government Architectects - dating from Georgian, through Victorian, Edwardian/Federation and later period designs (that's really old in Aussie terms).  I even have an 11ha lake held back by a stone arch dam dating from 1855-56 and surrounded by 300-acres of remnant woodland virtually on my doorstep... The allure of getting above it and getting amongst it (so to speak) is all to great to ignore, and only possible at low altitude via a multirotor.

Hence I commenced the obligatory research of multirotor focused forums, parts and suppliers, spent about 15 hours moving things around in ecalc, settling on the below parameters.

and ordered the following list of parts.

Approximately 4 weeks later this shows what I had came up with (note - I was still waiting for the M2.5x16mm hex bolts required to fit the 10mm upper plate extensions at this point so the battery is not yet fitted).  The current airframe and related ground station components can be viewed here.  Video flight log can be found here.

Assembling the various components was reasonably straightforward and rather rewarding, and integrated a range of techniques which I picked up through my online research.

For interconnections, I did my best to minimise path lengths, maximise distance between sensitive data lines and power cables (primarily by maximising the opportunities offered by the vertical layout - with noisy at the bottom and sensitive higher up.  All wires were also platted in an effort to maximise rejection, and ALL unnecessary power and ground wires removed to eliminate redundant weight and risk of ground loops.  Overall, I think it makes everything nice and neat, but it also minimises the need for fasteners and dampened movement.

For transmission duties I reinvigorated an old Futaba FB-T8AUF, which I had lying around from approximately a decade ago, with a 2650mAh Lipo battery and DFT FRSky 2.4Ghz transceiver.  After a quick internal repair of a fatigued pot wire, a bit of a clean up and a few other mods like removal of the metal carry handle to allow for better reception and signal propagation and removal of the racket on the throttle axis (by sanding it smooth and increasing tension of pressure plate - I am really happy with what I spent here! All up it was less than $100 for RCTX radio gear (including the V8FRII and I am wanting for nothing at this stage (other than maybe a 2W upgrade, which will likely be added soon).  It even offers full 6-way mode control via combination of 2 and 3 way switches, while the air brake function and other programmable offset options fulfil other duties like RTL and tuning functions.  Channels 6, 7 and 8 are also on pots, which will be useful for gimbal control in the future.  Also gives my setup a retro feel - as the transmitter is very dated!

For AV RX and RX duties I really needed to better understand the various elements of this, so I opted to build my own.  Bruce Simpson's helpful videos sold me on this concept, and proceeding the with assistance of Tim Creque's (see this thread for more info) Eagle CAD board designs, I am very happy with the results so far (learning lots).  Note - I am using 3.3VDC version TX purchased from infinitegadget on ebay. Making antennas is also an art I am keen to hone, with average results so far (my 5.8Ghz clover leaf performs as well as purchased item, but my Skew Planner leaves much to be desired.  I am yet to try a helical, but might wait until I have a CNC bench for that.

Hoping to simplify installation and interconnections, I opted for Hobbyzone's 25/30A 4-in-1 ESC, which is located underneath the frame.  This was mounted upside down (with the heatsink radiating heat from below the surfaces it is cooling, rather than from above), however it doesn't get overly hot with this configuration, so i expect this is not going to impact on reliability to much (I could simply flip the ESC, or add a tin heatsink extension, if it was of greater concern).  The 3A BEC in this unit also seems to be quite clean in comparison to a few UBECs i purchased, and it runs everything with ease including the AV Transmitter, with plenty of headroom to spare.  I cant recommend this enough to anyone wanting a cleaner frame setup.

To give me something to receive the live video feed I dug out an old portable DVD player i had collecting dust, made some internal space by removed the disc carriage, and installed a receiver which I spliced in to the main board via it's composite AV input and switched 5VDC rail.  To this I intend to add recording capability via the SD DVR available from HK, which will complete this item (Actually, I will likely add antenna tracking GS to this unit also for compactedfunctionalitiness).  It provides 1hr45minutes of operation on the original unaltered 7-year old 2800mAh NiMH battery pack which is excellent, but these cells will eventually be replaced with li-ions offering 9000mAh capacity. Telemetry is received via 433Mhz link via PC or (for increased portability) my little nexus 7.  Both the AVGS and Android telemetery GS with mounts, antennas and accessories all fit neatly into the travel pack provided with the DVD Player (now the AV GS), and the RCTX clips conveniently to the back to provide a small single handled package for all relevant equipment no bigger than a small shoebox.

Not wanting to stretch for a gopro, at least until my ability to build a reliable airframe had been evidenced, I decided to opt for something cheaper - hence the modified #26 808 1080 HD camera.  This was chosen for it's compact size, low weight, full HD capability and it's relatively affordable price.  Not happy with it's comparatively (to the gopro 3 black) narrow field of view, I decided to have a go at improving it.  The current lens is a cheap 0.47x wide angle taken from a 3-in-1 lens adaptor designed for the iPhone5 (cost $6 shipped).  It gives a great live output, the recorded files are sharp and of good quality and it records continuously for just over an hour between charges (recharge is quick too). It appears to have great potential, however wont be useful on the airframe until I learn how to deal with the jelly effect (Made more difficult by the fact that the camera itself has such a low mass in comparison to the airframe it sits on).  It does get very hot though, as the case is a rather undersized heat sink.  In winter this is not a huge concern, but i am pretty sure that this thing will go into thermal shutdown if operated in the Aussie summer for more than a few minutes.  I am planning to sandblast the case back to bare aluminium (removing the black paint finish) in an attempt to mitigate this issue to a small degree.

My idea to increase the height of the top plate (hence the name Quadzimodo - the hunchback) to make room for the 8400mAh 11.1 Turnigy Nano-Tech pack seems to have worked a treat.  It not only fits wonderfully, but also provides the following:

1. Exceptional protection in the event of a collision (2 crashed now have done no damage to anything but the rotors (other than some scratches on the receiver and prop mount tips).

2. An easy way of accurately and repeatedly placing the battery in the exact position for ideal balance.

3. Plenty of flexibility in movement forwards or backwards to counterbalance a wide range of payloads.

4. Facilitates a reasonably neat and tidy layout.

5. Due to tight fitting tolerance, the craft acts as a shape monitor, immediately identifying any swelling of the cells.

6. In combination with the choice to position the ESCs under the frame, the battery functions and a shield between the ECSs below and sensitive electronics above.

The technique used to lift the top plate is super simple and could no doubt be implemented into any DJI arm based multirotor such as a hex 550 or TBS quad to achieve a simular result.

With almost everything ready to go including the AV link, camera, APM Autopilot, RCTX, ground station components all sorted, I was left waiting for the boom stand mount for the GPS receiver and the external magnetometer (the latter was ordered because an external mag is recommended for 3.0.1 update of arducopter).  Burning to get in the air, I resoldered the onboard mag and temporarily mounted the GPS above the APM (a position partially obstructed by connecting wires) and decided to persevere with some preliminary flights.

The first 3 flights worked a treat, with no problematic issues and just about everything working as I would expect.  I was truly astonished at just how capable the APM 2.5 with 3.0.1 is straight out of the box with default settings.  As is evidenced by the flight video log, it seems pretty idiot resistant. Flight 2 was particularly enjoyable and revealed that the 1147 props are not too big for the airframe.  I was elated by how smooth it was, and surprised the rate it gains altitude via small momentary throttle pops. Interestingly, while ecalc estimated hover time to be up around 26-minutes and at 34% throttle, I estimate (based on my first 3 flights or approximately 30-minutes of flight time) it's hover flight time is a slightly more modest 21-minutes and approximately 50% throttle is required for hover.  Average current consumption is around 400mAh per minute.  I am still happy with this though, with the battery working well within it's constant 40C rating and unlikely to ever get a chance to challenge the burst limit of 80C for more than a fraction of a second (unless, of course, I go and load it down with another 1.5kgs attempt to do battle with the wind.

The one exception at this early stage was the current sensing capability, which was way off and over reading by approximately 100%, meaning that the APM would think it had expended 8400mAh when in fact it had only used approximately 4400mAh (Gauged based on how much the charger could put back into it after each use - B6 genuine from HK on 5A balanced setting).  I am yet to track down the issue here, but while it fails on correctly reporting the current consumption (both real time and cumulative) it does report voltage with absolute accuracy, so i am not at any real disadvantage until I fix it (other than continuous flights being limited to 10, rather than 20-minutes).

By flight 4, I had build enough confidence to head to the end of the street and take advantage of a space larger than my backyard to stretch the legs a bit more.  While the first 10 minutes or so went great, it appears that the GPS gave a false location which hit the eastern most edge of the 150m geofence i had set, which triggered a failed LAND sequence, then RTL mode - which (based on the fact that it didn't know where it was) caused it to rather violently make a beeline for the treeline.  I thank my lucky stars that all I hit were trees before it realised what it was doing.  Frankly, this episode really put the fear in me, and totally reframed my respect for these things - they really are dangerous in the wrong hands!  Amazingly, the only damage sustained was to a single propeller blade - which was replaced and the new set of 4 rotors rebalanced.

Still waiting for my GPS boom to arrive, i tweaked my failsafe settings a little and headed back out into the backyard.  Flights 6 and 7 went great, but again, flight 8 went bad - with the craft again triggering into FTL mode and heading (at speed) to the place which it thought it should be (which happened to be on the other side of my Barbeque (a collision which can be witnessed from no less than two perspectives via my video flight log referenced above).  Well... this is what I think is going on based on the kmz file from the park fly - as can be seen below.

and evidence of the GPS behaviour - as can be seen below.

This stuff really pulled the wind from my sail, and I resolved not to fly again until the boom stand for the GPS receiver had arrived.

Luckily, both the magnetometer and the GPS boom stand arrived the next day, so I didn't have to wait long.  However, have now lunched 5 1147 props, i was forced into wearing something smaller.  I had 2 sets of 1038 carbon fibre from RC Timer and few sets of 1045 plastics from ebay... I chose to balance and fit the plastic items due to safety concerns (I expect that carbon does way more damage than plastic) while I wait for more plastic 1147s to arrive.

I noticed an immediate improvement in the speed at which the APM would establish a 3D fix after relocating the antenna, and the cheap (less than $2) 3-axis mag I purchased worked great hooked in via the IC2 port.  However, errors in the location are still happening, especially indoors (sometimes it thinks it is over 1km from here, which I just don't understand).

I also do not like the performance of the craft on the smaller diameter props, and I don't like the idea that they require more current to hover.  It seemed to prefer flying on the 1147s, and I actually plan to order a few sets of 1238 props and a set of arms extenders (unfortunately adding 80grams to the frame in the process), just to see how much more efficiency I can extract from this configuration.  I am a big fan of efficiency over power, thus I see the size of the props as the most obvious area to maximise - as just as cone area relates to the efficiency of a loudspeaker, so does the prop area to a multirotor.

Following the crash which ended flight #4, and when i switched away from the 1147 props, i have had a problem with unwanted yaw (rotating in a clockwise direction).  The yaw now drifts gradually under normal hover and is rather more pronounced under increased thrust or normal manoeuvre.  In an effort to eliminate this, I have recalibrated both the accelerometer and compass a number of times, changed the external mag board to a different type, and even switched back to the onboard mag, but nothing seems to eliminate the problem.  Next, I will try reloading the firmware and if that doesn't address the issue, i will start swapping out motors to see if i may have an issue there.  However, i have my doubts that the latter will help, as all motors appear to be working just fine and perfectly consistent with the way they were originally (however the crash ending flight #4 appeared to strain them pretty hard and the crash that ended flight #8 actually dislocated 2 props along with their rotor mounts and shattered the remaining 2 props to pieces).

For Flight 10, which is not uploaded yet, I took my disabled mother with me down to the local cricket oval on a particularly windy day.  It flew fine, but the unwanted yaw issue is still present and problematic.  While flights 1-9 were all under 13-minutes of flight time per charge, for this trip I logged a little over 20-minutes on this charge.  The voltage (under hover load) at which I ended the flight was 9.45V, which came back up to 9.7V after the flight and prior to charging.

Unfortunately, after flight #10 I noticed an ever so slight bulging of the battery pack.  We are talking less than a few millimetres on each side (if it came like this, it would have seemed fine), but it was enough to a difference in shape in the hand.  Upon charging, it actually accepted 8656mAh - meaning that I over discharged it by approximately 3%.  Upon sliding it back into the frame, my suspicion of it's slight bulging was confirmed, as it is now about 1-3mm (total/cumulative) wider than it was (although still fits great).  Considering I had only cycled 10 charges through this battery, always charged it on balanced setting and in all but one instance only pulled between 50-75% capacity - is this normal?

Anyway, with 10 flights and 2 pretty hairy (yet forgiving) crashes under my belt in as many days, I have the following questions:

1. Can anyone help me understand why my GPS isn't perfect?  And if it's behaviour is normal? My iphone, android, ipad, navigation unit never think there a km from where they are, so why would my APM? Or to be Expected?  Should I invest in a more expensive GPS?  Could it be the fact that I have extended the length of the coaxial wire joining the ceramic antenna element to the Ubloc board? (it was about 30mm, it is now 250mm)

2. Anyone got any ideas why my current sensor isn't behaving as expected?

3. Based on the evidence above, can anyone confirm that my theory as to why it went rouge in flights 4 and 8 is accurate?  or perhaps, if my theory is wrong suggest what actually happened? Is there anything obvious I can do to mitigate the risk of this happening again, that I haven't already addressed?  Has this happened to anyone else?

4. Have I lunched the battery?  IE - is it dangerous now it has deformed (even though it is deformed to such a small degree?).  Should I discard it, speak to HK who supplied it, or should i simply persevere with it?

5. Can anyone see any issue with my approach to the wiring?

6. Can someone help with identify the cause of the unwanted yaw?

6. I would greatly appreciate any other input or feedback!!!

Thanks so much,

Jason.

Note - log file is of recent flight were yaw drift issue is present, while the kmz files are from the day of my Park Crash (flight #4).