flight simulator

Crazy Last Days of Summer

IMG_20170924_121344.jpg

 

The weather seems to have got the memo this week and we had warm sunny conditions with enough breeze to hand launch an autogyro. You can see from the picture how I’m stopping the blades from spinning with a rucksack.

It was crazy this week as I think everybody saw this as probably the last good day before the weather turns Autumnal. We had three drone parties in various positions around the field not talking to us in the middle. One was a father and his two kids with a toy drone (Syma?), another had a more professional one at the opposite end of the field, while the third was a couple I’ve seen most weekends with a Mavic. They were right in our landing zone, which wasn’t great. As for everybody else, there was a guy flying a Mavic when I arrived, plus someone I’ve known for ages on his bike with a UMX Fokker DR1. Shortly after this we had the guy with the Cobra, along with his son and daughter, then the Skysurfer, a guy with a DLG from a couple of weeks ago and a lady with a micro drone and FPV. Finally, towards the end we had another two guys from last time with another Skysurfer (or Bixler?), a small helicopter and a drone.

As for flying, in order of take-offs I flew: the Fokker DR1, the Skysurfer, my ATOM Autogyro, my RS352 and finally my 3D printed Dimorphodon quadcopter. The DR1 was a bit of a handful in the wind as flying a 30g box kite on a windy day is always interesting. Anyway, it was really pitching up badly this time, which I can’t remember it doing when I first flew it after its front end repair the other week. I ended up putting in about 10 clicks of down elevator, but I think it’s partly thrust line as things definitely improved when the motor was off. I did some fairly impressive vertical aerobatics while I was busy hitting the down trim, but then the motor wouldn’t come back on, I tried going back to zero, then 100% to clear the ESC, but there was no power so it was time to land. Back on the ground power seemed to be restored, so I can only think that the aircraft had had enough and decided it wanted to be on the ground more than getting blown about in the air.

Next up was the Skysurfer which I did the maiden for a few weeks ago and thought was an absolutely wonderful aircraft to fly. This time, the launch was a bit too interesting. I elected to hand launch and fly, but once I released it, the aircraft dived into the ground. There was a moment when I would ordinarily have cut the throttle, but the motor’s at the back and up top, so I just bounced it off the ground and slid my way into the air. This time out it was pitching down badly. Why I’m not quite sure as it had only had its wing fixing improved since last time out. Anyway, I got it sorted with about 10 clicks of up elevator and a bit of right, then let the owner have a fly. It wasn’t liking the conditions much, as it felt a bit spirally unstable in a tight left turn. It still flew well, though and the owner had a few more flights on his own too. – NOTE: it occurs to me afterwards that I’ve just flown one aircraft that was pitching up and needed me to fly with 50% down elevator, then switched straight to another plane that was pitching down and required me to fly with 50% up. It’s all good experience.

 

OK, so I fixed the ATOM autogyro yesterday and now it was time to fly it. One alteration that I made was to put the receiver on the rear wall of the pod rather than inside, as the internals are rather busy. Facing into wind and holding the nose vertical, I waited for some significant spinning to happen. Sure enough, it started humming and the wheels started to vibrate, which is the signal that the rotors are spinning fast enough to launch. But, it failed to really fly. I find myself having to tilt the rotors forward to prevent it going over my head, it flies away from me, but never seemed to be really on step enough to gain height. I had three attempts and every time it flew forwards, rolled and yawed around to the left, lost what little height it had and hovered in for a landing having not quite completed a full circuit (at about 10 feet height). I’ve switched to a 9×6 prop from a 9×4.5 to try and give it a bit more bite, but it hasn’t made any difference. Also, I’m using 0.8mm shims under the blades, but I did this as two 0.4mm shims on top of each other. The top one was sellotaped on to make it removable, so the next obvious thing to try was to remove the extra shim and fly with just 0.4mm. This didn’t seem to make any difference, so I’m going to have to take this home and have a think about it. The aircraft has flown from a hand launch in this configuration (with a 9×4.5 prop though) as I had a 4 minute flight. My feeling is that it’s the launch and not getting into the right part of the flight envelope. The left bank on launch says to me that the rotors aren’t up to speed and then it never gets out of this stalled mode. I was flying with the throttle at 100%, but with no noticeable effect on airspeed, rotor speed or height gain. Something isn’t right here, so I need to do some more homework on how to fly autogyros.

After that I had three flights with the RS352. I’ve only just realised, but this aircraft’s birthday is around this time of year. I bought it in early October, but I can’t remember if it’s now 4 or 5. I’ll have to have a look, but I like to live with my aircraft for a long time. I don’t know whether it was my new charging regime, but the good LiPos which I put in for the first two flights just didn’t seem to have the punch that they normally have. I used the storage charge of 3.88v two weeks ago, then fully charged them Saturday night and used them without topping up on Sunday morning. Normally I would top up and fly, so that might be the difference? I was a bit preoccupied with the autogyro though.

Finally, I flew the 3D printed Dimorphodon quadcopter that I made a while ago. I’ve never had a real go with this until now, but it was a lot of fun. It’s based around the HubSan flight controller and I had to use Expert mode because of the wind, but it doesn’t fly like most other quadcopters. I always wonder how much is to do with how an aircraft looks, but with its odd shape and more aircraft like wings rather than being square like a quadcopter, it seemed to want to spin round on the spot. It was great fun to fly, but I need to do something about those pointy wing tips as I managed to stick one in the grass at one point. That can’t be safe, so I’ll round them off before I publish the STL file for other people to make.

One final thing is that I’ve taken two sets of photos of the autogyro in order to attempt a 3D reconstruction to build a mesh for my flight simulator. I did one inside and one outside, so I’ll run that once I get back to work. The more I get into autogyros, the more I want to build a simulator for one.

Advertisements

Rain, Sleet, Snow

We’ve had all three forms of precipitation in the last few days. There’s been heavy rain since 6 o’clock this morning and we might get sleet and snow later, so it’s fairly obvious that there’s not going to be any flying done this week. I haven’t flown the RS352 since the week before Christmas now. My BMFA renewal arrived earlier this week, so I’m good to fly for the whole of the rest of the year. It’s always like this every January.

The little blue acrylic quadcopter is flying around indoors nicely though. You can see that I have added some basic prop guards since last week. I’m actually a bit disappointed with the weight as it’s come out at 48g flying weight, while I was hoping for a bit lower. I’ve still got to add the 4g FPV camera, but the test flight this morning was around 5 minutes. That suggests that it’s round about the same as the HubSan X4C, although the flight controller is in a different league. The separate flight controller and FrSky receiver that you can see in the picture above must be heavier than one of the integrated units, so I must be giving away a few grammes there. I’m flying it with my Futaba Field Force 8 which has a FrSky ACCST module in the back. This is what makes me really mad about FrSky, because I can’t use the Taranis radio with this receiver. Our Taranis for work was flashed with the European firmware and that won’t talk to an 8 channel receiver, so I’m sticking with the Futaba. Either that or we’ll have to switch everything back to the US firmware.

Anyway, it was refreshing to fly a quadcopter that I’ve actually designed and built myself. It’s just a bit too much of a conventional design, though, so I might have a go at some more radical alterations next. As for this morning’s first full test flight, I took some video using my RunCam, but it’s not that good, so I’ll have to try another flight later today. Also, I don’t think it’s calibrated properly as it was trying to yaw left all the time. This might be to do with the fact that I hold the flight controller on with double sided tape and the flight controller has a habit of peeling itself off where the components on the back don’t make a flat surface. It needs a better solution really, but it’s so easy and light to hold the flight controller and battery on with tape.

As for the balancing of the autogyro blades last week, that turned out to be harder than I thought. There is only a few grammes difference between the heaviest blade and the lightest, but glue doesn’t add that much weight, so it’s hard to get them to balance. I’m not sure how close they need to be, so I ended up trying to take some wood off of the heavier one, but they’re still coming out as 19.2g, 19.4g and 19.9g. That’s within 0.7g, so it’s probably close enough. When I finally cover them with film it’s going to cause the same balance problems all over again.

That’s it for this week, except to say that I now know how to make the Taranis work as a joystick with our flight simulator. You can reprogram the ranges to match the ranges that Unity expects, so it is perfectly possible to use it as a joystick in other software that was written in Unity too. FPV Freerider is one which immediately springs to mind. So that’s my afternoon planned out: film acrylic blue flying, add the FPV camera, play with Taranis and flight simulator.

Let’s hope it’s not raining next week.

Red Autogyro, Blue Autogyro

I’m a bit late this week as I’ve spent the day covering my ATOM Autogyro. There’s no flying this week as yesterday it was blowing a gale and this morning isn’t much better. It’s taken me a long time to get to the covering stage as it’s just been so hot recently that I couldn’t bring myself to plug the covering iron in.

As you can see, it’s taken me a long time to get this far and I’m not quite there yet. Normally you cover the bottom of the aircraft and it’s a nice and easy way to start, doing something that isn’t terribly visible and is square and flat. Not with this fuselage. I hadn’t realised how many compound curves there are to go around on the bottom. The sides were a piece of cake by comparison. It seems you never lose the knack for solar filming though, even after you realise that the blue metal flake film is partially transparent and is going to magnify every tiny little mistake. If you look at the tail closely, I’m disappointed at how visible the joins are. I was covering it using up my very last scraps of old (20+ years old) film and ran out on the top horizontal stabiliser surface. You can see the overlap where I’ve had to join three pieces together in a funny position. Oh, well, I’ll just have to get creative with the trim graphics. I still haven’t decided what colour to do the fin end plates or the blades, but that can wait for a bit. What worried me a little when I was covering the fuselage is just how top heavy it is. I’ve never covered anything like this before and it was so difficult to hang on to the thing as I was smoothing out the film as it kept trying to twist out of my hands. There’s just nothing to hang on to or rest flat on the work surface.

OK, so that’s the autogyro, but I’ve actually spent most evenings this week working on the quadcopter simulator. I’ve had a play around with a control system that could be used for automatic navigation. In the simulator I could just tell it to head in a particular direction at a certain speed and height and the controller would just get on with it. I was thinking that I could try out some of the synchronised manoeuvres that I’ve seen the GRASP Lab at Penn University put up on the Internet. I’m thinking of simulating it first as I’m currently lacking a vision system, but it would be interesting to explore what’s possible with some of the cheap hardware that’s around. In addition to this, I’ve been improving how the quadcopter flies in the simulator. It’s well worth having a look at the MultiWii rewrite controller in the Cleanflight code to see how it’s done on a real aircraft, but there are some subtle differences between that and how it’s done in a flight simulator. It’s all very interesting once you get into it and it’s making me think about whether I could construct something novel out of all this automatic control theory.

30 Degrees and Windy

The weather is rather interesting today. It’s very hot, but it’s also very windy with the clouds skimming across the sky at some speed. The sunshade also blew into next door, so the decision not to bother going flying looks like the right one.

quadsim_OlympicPark1.png

The quadcopter simulator with London Olympic Park models

I spent yesterday trying to build another scene for the quadcopter simulator. With the Rio Olympics just starting, I wondered whether I could build a London Olympic Park model from bits of free artwork collected on the Internet. Having spent a day on this, I think I’m going to give up at this point as my artistic skills are not up to it. The picture above shows a fairly passable Arcelor Mittal Orbit, Aquatic Centre, main stadium and velodrome, but the textures are seriously messed up. The problem is the export path from Sketchup via 3DS and then into Blender before loading the model into the Unity scene. There are too many conversions between formats introducing errors where an importer can’t handle the format of the data in the previous step of the chain. Basically, it was taking ages and a lot of trial and error, leaving me with Blender models where textures, normals and vertices were broken. I think I have to accept that my artistic skills aren’t up to the job and go back to being a programmer.

Having spent quite a long time flying the quadcopter around the Olympic Park, I realise that I have some serious work to do on the flight simulation itself. The way it flies just doesn’t feel at all right, so I’m going to go back to the code and add in the motor force and moments modelling that I didn’t have time to do before. Then all the pieces should be there to tune it just like a real quadcopter.

Anyway, my plan for the rest of the day is to wait until the temperature is below 30 degrees and then think about covering the autogyro. I really can’t face the thought of solar filming in this heat, so I need it to cool down a bit first.

Simulators and Micro Quads

There’s a very light misty rain falling this morning, which isn’t going to clear any time soon, so no flying this week. I might have given it a go, apart from it being really dark, impossible to see an aeroplane if you wear glasses (think windscreen wipers) and the forecast is for strong winds by lunchtime.

Actually, I’m really annoyed because I spent most of yesterday on the electronics for my 110 size quadcopter, only to find that the flight controller had packed up.

IMG_20160131_140625.jpg

My 110 size micro quad made out of Depron and four HubSan motors.

As you can see from the picture, I built the frame out of Depron, which was originally from an old pizza box base that I saved. It’s surprisingly strong, considering that it’s only a push fit and there is no glue used at all. The flight controller is a Flip32+, which should sit on the brown sticky pad in the centre. Everything was going really well until I connected it up last night and I think the micro USB connection broke. It detects that something is connected to the USB, but I suddenly started getting “unrecognised USB device” errors from Windows 10 when it had all been working perfectly before. It’s annoying because I was just about to connect up the motors and run them using Cleanflight for the first time to see if it had enough thrust to fly. It’s a bit heavy for this class of quad, so the flight controller and radio are going to be switched for smaller and lighter versions. It appeared to be working up to that point as I had test run one motor, then added the other three and suddenly the flight controller won’t talk to me. There’s a lot of wiring there for something this small, and the connectors are the pins from a Futaba servo connector, crimped onto the wires and then covered in heat shrink tubing. It works surprisingly well, but an integrated PCB with tracks running out to the motors would probably be better.

So, I spent the rest of my time on the quadcopter simulator. I’ve worked out how to set the textures on models imported from Blender into Unity. You have to create a folder called “Textures”, put the images in there, open the Blender model from Unity, point at the textures you just copied to the Unity Assets “Textures” folder and then set the albedo on the Unity material to use the correct one which now appears as an Asset. UV mapping in Blender is a completely different problem though. I can’t say I’m really impressed by any of this as, having worked in computer graphics for quite a number of years, it seems to me that their model import chain is just rubbish. There’s no reason why Unity couldn’t do all that for you, but as long as you know about the albedo hack, most people can probably work it out for themselves.

Anyway, here are some of the screenshots so far:

 

I really need to get this playable by the end of the day, as we’re running a Royal Institution coding for kids session in about a month and this forms the central theme. All that is left to do is some of the housekeeping code and then I’ll have an alpha release.

Hurricane

No, not the weather this time, I actually flew my first hurricane. The weather was actually perfect, with virtually no wind and it felt warm in the sunshine this morning. I was beaten to the flying field this morning by a guy flying a DJI Inspire. Not the same one from last week, but he packed up while I was having my first flight, so I never got to talk to him. Then a young boy, his parents and sister arrived with a really amazing little aeroplane. I had to look it up when I got home. It is called an Ares Trainer 100 and is a tiny UMX style plane that uses the same single cell LiPo as the Hobbyzone Champ, but is significantly smaller. With no wind it seemed to fly for ages on that single cell and, for someone who had only bought it the day before, he was flying it really well. Most importantly it survived some big bumps on the ground, so it doesn’t break easily.

Anyway, just after my second flight with the RS352, the owner of the brand new Dynam Hurricane arrived. I had flown his A10 Warthog last week, so the Hurricane didn’t seem such a tricky prospect. After checking everything over, I thought the balance was slightly back, but it was in the suggested range, so we put it down on the runway and I opened up the throttle. I was a bit worried because it feels heavier than you would expect, but it went straight up into the air with plenty of power from the 4S LiPo and that amazing 3 blade prop. There were no problems at all with this aircraft, I think I only put in about three clicks of down the entire flight. I put the gear up after the first turn and just flew it around the sky for about 5 minutes. This feels like a real aircraft, though, as it’s got some weight and presence about it. In the turns you have to maintain momentum and then levelling out it tends to keep climbing on you. All too soon I was thinking, “OK, now I have to get this back down” and flying square box circuits to gauge the speed and rate of descent. There are no flaps on this one, so the eventual landing was a bit fast on the rough grass. I had hit the limit on the up elevator and the flare wasn’t what it should be, resulting in a bit of a bounce, the main wheels catching and flipping the tail over the nose. This was with virtually no speed, so it didn’t do any damage. A check of the capacity left in the battery showed about 50% left, so duration must be about 10 minutes.

Now, I had offered the owner of the Hurricane the transmitter on the first flight, but declined until the second one. Take off was just as easy as before and I settled back down into a circuit at half throttle and handed the transmitter over to him to fly. I won’t go into too much detail, but the flight ended up spiralling into the ground a few minutes later. Standing next to him I thought he was OK, told him to give me the transmitter back if he wasn’t, but he thought it was too late by then. Seriously though, I can crash with the best of them so I would rather have the transmitter shoved into my hands two feet off the ground inverted in a spiral dive because you never know. If you are going to hit the ground, please remember throttle to zero, it always does less damage. As for the recovery, we were both expecting to see lots of bits of aeroplane, but it was pretty much in one piece. That’s really very impressive, so Dynam must build them strong because this aircraft isn’t light. The 3 blade prop was broken, the left landing light cover was missing and some bits of foam from the right leading edge of the wing root had been pushed out by the wheel which looked like it had moved. That was it though, the airframe was still basically in one piece and serviceable. I managed to find the tip light cover in the grass and we suspect that might have been the point of impact, with the left wing tip and prop absorbing the energy as the aircraft hit the ground.

Around that time another regular arrived with his Multiplex Heron, so he saw most of the flight. Apparently, from the car park he said he was sure it was a Spitfire. The wing shape in the air and the underside are the right colour, if slightly more square than the Spitfire ellipse. I managed another two flights with the RS352, while all the time we were watching what looked like a 250 size quadcopter flown by somebody camped out under the trees about 200 metres away. He obviously didn’t want to talk to us for some reason? Seeing the Hurricane flying around close to him must have been quite a sight though. We didn’t get out that far as you could see the quad flipping itself left, right, forward and back. We’re not sure if that was deliberate, but he definitely hit the ground a few times and had to fix it.

That’s it for this week, apart from having to admit that we’ve bought an F450 frame, Taranis X9D, Mini Hawk Flight Controller and Fat Shark V3 to build a quadcopter for work. That’s why I’ve been working on a quadcopter simulator in Unity. We’re doing some 3D reconstruction work using point clouds and have also just purchased Pix4D.

The quadcopter simulator in Unity.

The quadcopter simulator in Unity.

I’ve also bought myself some parts to build my own 250 size quadcopter. Our plan is to film the build and 3D print some of our own parts, so it should make a really interesting project.

OK, so today’s target must be to get the tail on my ATOM autogyro as I’m really desperate to see it fly now. I’ve done nothing all week so I’m off to make the rudders work.

Three Weeks Running

I think that’s three weeks running that I’ve been able to fly on a Sunday now. That hasn’t happened in a while, which just shows you how bad the weather has been.

Anyway, we had seven people this week, including myself, flying a combination of a Dynam Smart Trainer, EFlite Advance, Parkzone Radian glider, Hyperion Sniper II 3D aerobat and a smaller CAP231 (?) profile 3D EPP parkfly plus a small helicopter. The final guy was the aerobatic helicopter pilot, who I haven’t seen for quite a while, but he was doing his thing upside down and right way up with a big electric helicopter.

As for me, I only had the RS352, which I managed to fly three times, along with a long flight of the Radian and three trimming flights with the Sniper, which I had flown a few weeks ago. Neither the Sniper’s owner nor me could understand this, but it was completely out of trim for some reason. He managed to do a really interesting ground snap roll before handing me the transmitter and saying, “you have a go”. OK, so first attempt at take-off resulted in the prop hitting the ground and the nose digging in. I cut the power early, so no damage, but I did check the prop remembering what happened last time. Attempt number two and holding in a little up, it took off. Oh dear, that’s VERY sensitive, so find the operating point, fly the circuit and come back down again. After turning down the amount of movement on elevator and aileron (the TX said 135%, but we don’t know why), and also re-centring the surfaces, I was back in the air again and everything seemed a lot better. That was until I pointed out that the rudder was actually reversed. I really don’t know how I didn’t notice that before as I had done all the checks, but perhaps it got reversed when the rates were changed and I just didn’t notice? Anyway, back down to earth again, reverse the rudder put it back into the air and everything was more or less right.

The most impressive flight of the day was probably watching the Dynam Smart Trainer, which looked absolutely perfect. Apparently it was only his second flight with it, the first had bent the nose leg on landing so he had to straighten it out. For a trainer, taking off from rough grass, it went up absolutely dead straight, like it was attached to rails on the ground. He then proceeded to fly it around the sky and execute a perfectly good landing about 100 metres down the field. Any landing where you don’t have to bend the nose leg back counts as a good one, so he got a second flight before packing it up to go home. Apparently he’s waiting for some new LiPos to turn up so he can get more flights in. The thing we all found astonishing about this aircraft is the sheer size of it. The entire tail section is removable to enable you to get it into a car and both wings unplug.

That’s it for flying this week and I really haven’t done anything more on the ATOM since last week due to the amount of work I’ve been doing. One thing I was playing around with yesterday, though, is the Quadcopter simulator which we’re writing. I’ve just built a Windows 10 machine so I can run Unity and managed to get it flying using my old Great Planes Real Flight Transmitter joystick controller for the first time. This was originally an analogue gameport device, but I’ve connected it to a SiteCom gameport to USB adapter which I got from Maplin years ago for about £10. I now use it for all my simulators.

Here is a quick screenshot of my attempt at drawing a quadcopter in Blender for use in the simulator:

A model of a quadcopter being built in Blender

A model of a quadcopter being built in Blender

Also, I must put a new battery into my HubSan Q4 as I’ve completely worn out the old one flying it around the bedroom. A post on how to do that will follow later.