Ducted Fan VTOL UAV / Sci-Fi Craft

Why are simple things so difficult?

2011-06-15T01:33:00.006+01:00

Ok, I've got a couple or three hours to spend on a controller. I need to control an EDF (output a servo signal) and read a ultrasonic ranger (read ADC channel). Shouldn't be too difficult, right?

So I've opted to try it out on the mBed. First there's some faffing with usernames and passwords, then I remember the web-based IDE doesn't work on Opera. I finally get up and running.

Reading an ADC channel is dead simple and works first time. Cool. Outputting a servo signal should be just as easy. It isn't. I check the code (C++ is still alien to me) and libraries, jump through some hoops and then it compiles. Checking the signal on the scope shows it's not moving and this highlights the major drawback of the mBed: no debug.

No debug and my limited C++ knowledge leads me to dump the mBed and switch back to a PIC. It's getting late and I just want some simple algorithm to work, nothing fancy. I've got some bit-banging code working already so now just need to remember how to do an ADC read. I think I've done everything that needs doing, so I switch to debug mode and the value in the watch window changes as I twiddle the pot. Sorted.

Next I have to tie the ADC to the servo signal, but not now; it's late (or early) and I'm off to bed...

The path of least resistance

2011-02-10T10:57:00.000+00:00

I got to play with an mBed at work yesterday, it's pretty awesome.

I'm in the process of designing a small PIC board that I can stick a number of different footprints on to take a range of devices. I was going to allow flexibility for plug in expansion (like the Arduino shield) but still keeping it small and simple, and being compatible with breadboards. The plan was to use it on the UAV and a couple projects for my motorbike.

I wasn't happy with the cost of the Arduino. It's too big and expensive to leave in many simple projects (especially if using shields). The smaller versions have no on board USB, which means more faffing. There is a slightly bigger version that does have USB on board, but that's only a few £ less than an mBed and doesn't do USB host, CAN or Ethernet (unless I'm mistaken).

So I'm going to simplify my PIC board, for simpler projects, keep it smaller and cheap (10+ for the price of an mBed?). When I need USB/Ethernet/CAN, I'll use an mBed.

Happy bunny

2011-01-16T23:25:00.000+00:00

I've actually managed to get something done! OK it's nothing major, but I feel I've accomplished something. Something I haven't felt for a long time, so I'm a happy bunny. All I've done is added three buttons; up, down and home, along with adjustable end stops and home position to my bit banging servo code. I also upped the refresh rate to about 80Hz, from previous experience. I can substitute an EDF and could set the home position to be just below the point where the whole thing gets light on it's feet.

But I need to close the loop, so will dig out the ultrasonic rangers and try remember how to setup the ADC to read the range signal(s). I know, it's going to very crude...

Haven't had any photos recently, so thought I'd change that:

Microchip ICD2, my own dev board and standard servo.

ICD2 sorted (& plan B?)

2011-01-16T12:07:00.002+00:00

Got it working last night by uninstalling MPLAB 8.6 and MPLAB X, then deleting the microchip folder. Installed 8.63 and X again, went through the driver installation and it connected!

So now I can mod my bit banging servo code for controlling an EDF.

But wait... thinking about it last night, there's no point doing all the low level stuff in the uC, that's just going to delay things even further. I need to use an external servo controller that I just talk to over serial comms. It takes a lot of overhead away from the main processor and makes servicing each task a lot simpler. Sound like a plan?

ICD2 is OK

2011-01-14T11:05:00.000+00:00

Plugged it in, had to install drivers, but MPLAB connected to it fine.

Is it a Win7 problem?

Still no luck

2011-01-14T08:06:00.000+00:00

Last night, I tried reinstalling ICD2 drivers, which didn't work. Then I downloaded and installed MPLAB X (Beta), no change. I can only assume my ICD2 died since I last used it!? Then I got bored so went and watched four or five episodes of The Big Bang Theory (thanks for the loan, Gary).

I'm going to take the ICD2 to work today and try it on my machine there.

Been too long!

2011-01-12T20:58:00.000+00:00

I have a little time before my studies start up again, so thought I'd try to get something together to control an EDF for crude testing. I grabbed my PICDev01 board, ICD2 and a PSU and sat down to get to grips with MPLAB IDE again (it has been a long time).

So far I'm not having a lot of luck. It's nothing I haven't done before; I think the ICD2 drivers haven't installed properly and the project won't build (new HDD since last used too!). I can see the advantage of having everything in the cloud, as the mBed does; saves these sort of setup hassles...

Mr Parker - I thank you...

2010-08-13T00:04:00.000+01:00

I got a good telling-off today, by Mr Parker. He said I'm spending all my time planning my revision timetable and not actually revising (working on the L/G - he wants to see it fly). I should just use a couple of bent wire coat hangers and be done with it.

So, I got home tonight, determined to get a chunk of work done on the legs. I want to get some gear in place for the thing to sit on, so it's doesn't rock and skate around (in theory at least). But you are right, Mr P, so I worked like a crazy man...

I got the rear vertical pieces cut out, sanded, drilled and tapped, then drilled the Alu knees of the rear pair. In a flash of genius I solved the anchor and adjustment problem for the tensioners, so they now operate correctly. I've ditched the LEDs too (for now).

I can lock-out the gear (manually locked in deployed position) and move on...

Everything will need dressing up a bit.

Time to work towards a controlled lift-off and touch-down cycle.

Fiddly legs

2010-08-12T01:29:00.000+01:00

I've finished the front pair of landing legs, at least the component parts. I still need to find a neat way to secure the operating linkage, and operation isn't very smooth either. It's really fiddly!

I've been messing with adding lighting to the legs, but I'm not convinced about the effect (lighting not shown in photo).

The leg on the left has the tensioner, LED and wiring in place. Additional wiring will be needed if I add a Weigh-On-Leg sensor.

Ardunio or PIC - Action Priority Matrix

2010-08-06T12:06:00.006+01:00

I remembered this:

I don't want the tool to be a major project, unless it needs to be. Arduino's looking good!

Arduino or PIC?

2010-08-05T23:37:00.000+01:00

A dilemma: what should I use? Should I buy into the ready-made hardware, tons of examples and accessories of the Arduino or do I stick with the more powerful/flexible PIC, which I've used before, which has a rich selection of peripherals and can make a smaller, neater end solution, but needs a lot more effort to get going?

The trouble is it's been a while since I did any serious PIC work and I'm attracted to the quick and easy Arduino. But I'm also not sure the Arduino will be up to scratch, with respect to I/O, peripherals and physical size (I could get a Nano?) and I'd have to learn a new platform.

So if I go for an Arduino, will I have to re-do everything in a PIC solution later? Should I go straight for a PIC solution, but with the extra time and effort?

I don't know. The Arduino does look like a quick solution...

Arduino website
Microchip PIC website

How many times?

2010-08-03T22:17:00.114+01:00

I've fabbed two vertical pieces for the front L/G pair from 6mm polycarbonate and in the process; realised that I could have done it better, sooner, given more thought. But it'll do. Hindsight is a wonderful thing!

Rear pair will be slightly shorter.

I gave one a light coat of paint so it would show up in the photos and give me an idea of what it might look like when done.

I had a small can of "Pewter".

Just resting in place.

I'll add some detail later.

Comparison. I'll have to stop re-designing the gear soon!

Like the caption says, I want to finalise the gear design so I can move on (I've said that before too). I've been thinking about the next step; how to get the desired results without unnecessary work, but that's another post...

New site; UAVpilot.org gives me a mention!

2010-07-24T19:06:00.000+01:00

I had an email from Thomas of UAVpilot.org;

Just wanted to say, I stumbled across your site earlier tonight and was really impressed with your do-it-yourself UAV project!

I've recently made a website, it's about...two days old now...but I hope you don't mind my doing a very small article on your project and providing a link, for anyone who might stumble upon me.

So I'm returning the favour with a link in the sidebar, looks like it could be a good resource... go take a look: UAVpilot.org

Landing gear almost done

2010-06-15T13:16:00.001+01:00

All four retracts now have arms and swinging brackets that the vertical "foot" mounts to, complete with pivots and end stops. I think I might ditch the slider mechanism after all; save the added complexity for little gain. The polycarbonate arms provide a little spring anyway and I'm hoping to get soft controlled landings. Fixed vertical sections (the "feet" or "shins"?) are next, then the servos/mechs.

Keeping at it

2010-06-13T22:41:00.005+01:00

No Worldcup for me, but I did spend a couple of hours tinkering in the garage. I discovered a minor issue with the gear retracts, but think I've got around it. I've worked on the second leg and cut some parts for the rear pair.

I've also been looking at PIC24Fs and will probably look at getting a couple of samples from Microchip. I'm also going to see if it's possible to home etch a PCB for a 100 pin TQFP with a 0.4mm pitch. I'm a bit rusty with PCB layout and C, but I'm sure it'll come back to me.

Longer legs and a blue ball

2010-06-06T23:19:00.000+01:00

I've re-made the main arm of the L/G, making it longer and better positioned when stowed. I'm also half way through assembling the second leg (rear pair will probably have to be a different size). I'm having second thoughts about the sprung slider mech too; it adds complexity, weight and it won't even work too well because of the angle. Should I scrap it and go for a fixed lower leg? Would anyone notice? Would anybody care?

Notice the ball in the nose? That'll be the location of the pan/tilt camera and LED "searchlight".

Different angle of the longer leg, deployed.

The stowed position is much better and the wider footprint will improve stability.

Also, I was originally going to have the cargo ramp on the bottom, opening forward, but the L/G isn't going to give much room so I'll have to go for the traditional rear ramp under the tail.

One leg done, three to go!

2010-05-23T23:05:00.002+01:00

I've fixed the slider to the pivot arm...

...and fixed the shaped vertical "foot"(?) to the slider mech, then temporarily fitted it into the airframe to test the fit. While I was at it, I clamped a servo in place and connected a pushrod to the retract control linkage, then plugged in the servo controller for a demo (see video, below).

Prototype L/G slider

2010-05-18T22:48:00.003+01:00

The slider mech I mentioned in the last post, has to be lightweight and compact. I had come up with a possible solution, but it would've been a bit clunky. The alternative came from a sacrificial plastic vernier caliper that I got free, many moons ago, with some tech magazine. It's very light, free moving and ready made (kind of).

I'm interested in the centre sliding pieces. The cross-section shows the interlocking profiles that would be difficult to replicate by hand.

I hacked a mock-up and glued it to the gear arm, as proof of concept. There's a small spring fitted behind the slide that took around 1.5 kg quite nicely (there will be four of them). My concern is it may be too long to fit snuggly in the fuselage. Video below shows the proof of concept. Now I need to make it more robust.

Obviously, this will all be painted and disguised!

Landing gear prototype - video

2010-05-16T22:55:00.004+01:00

Finally, a L/G design that might work!

2010-05-16T21:59:00.003+01:00

I got some time to work on the landing gear and came up with a design that gives the desired look and motion, while still being lightweight. I bodged together a prototype that kind-of works. Photos below.

Showing gear in stowed position.

Showing gear in deployed position.

I'm hoping that on the real thing; I can have the vertical piece on a sprung slider, to give added effect and possibly some shock absorption upon touch-down. This really needs a video...

It'll happen when it happens...

2010-03-30T21:15:00.000+01:00

My local hobby shop (Power Hobbies) called to say the plastic sheets I'd ordered last week, had come in. I now have 10 sheets of 0.25mm, 10 of 0.5mm and one of the thicker 2mm (all approx. A4 size), so I can go ahead and create the fuselage, in theory.

Damn meerkats have hogged my time recently, on top of my studies, but I'm determined to break this gear issue and make some progress. A friend said they've given up worrying over progress of their projects but I can't help beating myself up over how long it takes me to get anything done.

I had a little time!

2010-03-10T22:09:00.002+00:00

A couple of weeks ago I managed to get some time (at a SHDC meet) to hack together some bit-banging code on a PIC to control a RC servo. This crude code, in theory, will allow me to control the servos for the landing gear, nacelle steering and of course the EDF thrust.

I had a little time over Christmas to discover my planned landing gear redesign was halted by my inability to mill a tidy slot in some aluminium channel with my tiny little lathe/mill combo. I need to come up with a simpler mechanism, get the gear done and put some kind of external skin on the airframe, then I can get to the control system!

Joost doon't 'ave the poower!

2009-11-10T23:32:00.000+00:00

I made the adapter cable with the Deans connectors (cut off the end of the umbilical) and 4mm bullets fitted to mate with the Li-Po. I didn't know how charged the battery was, but I had to try it...

With just the battery (no extra weights) I powered up and "hovered" for a minute or so; all 3 EDFs running and plenty of power. I then added the 2 blocks of metal again. These are just bits I've been using to help in the construction so far, as weights or a square-edge. So powering again, it got off the ground but was obviously loaded. Second attempt and it just got off the ground and I hit max on the controller. I backed off, then tried again - now barely lifting. I took off the heavier block and re-tried - only slightly better. The battery must be dead! As I said; I don't know what the state of charge was beforehand and the charger hasn't arrived yet...

Disheartened by the sub-60 second runtime, I was thinking the battery couldn't have been charged. I wasn't expecting much, maybe 2 or 3 minutes. I also thought I should check and quantify the "other stuff", so went off to weigh the 2 blocks. They were a lot heavier than I thought: the larger block of aluminium was 335g and the smaller stainless steel block was 414g! For comparison, the battery is 408g (listed as 412g). That made me feel a little better; I'd added a lot more than necessary, an additional 759g (1.65lb for those working in old money). So fingers-crossed; if I can keep the weight of all that "other stuff" down and with a fully charged battery, it might just work... for a couple of minutes.

Battery and tail

2009-11-09T09:07:00.001+00:00

A colleague pointed out that I should get a battery sooner rather than later, so I don't waste my effort if the battery can't provide enough juice to lift off. So I ordered one: Li-Po 5000mAh 20C, 30C for ten seconds. That's 100A continuous and 150A burst. Doesn't sound feasible! We'll see.

I got the upper tail fins finished and fitted. They're mostly for show. They may help in forward flight, but I'm no aero-engineer. I'm working on gut-feel!

With as much as possible fitted and two lumps of metal for "some other stuff", it weighs in at 2kg.

Clarification

2009-11-04T07:40:00.002+00:00

I thought I should clarify that although I'm using some RC gear, I'm not using RC in the traditional sense. I have no transmitter or receiver as a couple of MCUs will be doing the work, these will have a radio link to a base station of some kind where the operator can monitor and guide the aircraft. Any operator inputs will feed in to the control loop as error offsets rather than direct control of servos. At least, that's the plan...