A Bespoke Solution - Bat Plane Mk 3

Following on from our less than successful weekend of testing (check out the blog post here if you haven't seen it already) we decided to build a better plane.

In this post we're going to cover a bit about what we wanted the air frame to achieve along with a description of how the major bits were built and finally the maiden flight. This is partially to keep you up to date with what we've been up to and partially to pass on the information required to reproduce what we've made.

Design parameters / goals:

• It needs to fit in the car so we need either a wingspan of 2m or less or a foldable/collapsible wing.
• As large a wing area as possible in order to reduce the wing loading.
• Building materials must be readily available and the building method can't be overly complicated (we don't have loads of experience building planes!).
• A flight time of 45 minutes or greater.
• A distance of 2m or greater between the propeller and microphone of the detector. This probably means a collapsible fuselage so it fits in the car.
• V-tail to keep the surfaces away from the ground when landing.
• Take off - hand launch or bungee
• Landing - possible on rough ground
• Slow flight possible for landing and potentially for use in surveys

With the above parameters in mind a rough 3D model was designed:

The model shown above has a wingspan of 2m and a chord (the distance from the leading edge to the trailing edge of the wing) of 30cm which provides a wing area of 60dm2 (6000cm2). Generally, the larger the wing, the more lift it can produce which for our use is a good thing.
The 'fuselage' or at least the bit that joins all the pieces together is made of a carbon/glass composite fishing rod as seen in the Mk2 design. It's collapsible which was a necessity of the design as it wouldn't fit in the car otherwise!
The weight was estimated to be around 2.5kg and this was factored into the power system choice.

Power system

In order to achieve a long flight time the current draw from the battery would need to be low so a relatively high voltage system was chosen.  The battery is a 6S LiPo (though Li-ion would also be suitable), the motor is a Tiger MN4014 400KV and the prop is an aeronaut 15x6 folding. It's estimated that this setup will produce circa 3-3.5kg of thrust (greater than a 1:1 thrust to weight ratio) and cruise at about a 5A draw. If we had a 5000 mAh battery and used up all the juice (which you shouldn't do as for maximum longevity and battery health you should always use only up to 80% of the capacity) then this set up would theoretically fly for 1 hour. The motor was supplied at reduced cost by Duncan at heliguy.com and a big thanks goes out for that. The prop and yolk (along with ESC and a few other bits and pieces) were supplied by Gareth at Gliders.uk.com at a reduced cost so a big thank you to him as well.
In case you were wondering, the prop is folding so we can protect it upon landing. Previously we had a number of issues in breaking props and motor shafts so hopefully this will cut down on those!

Here's a quick look at the motor and folding prop:

Bat plane Mk3 MN4014 motor and folding 15x6 propeller

​After a discussion with a local flying club (special thanks to Steve and all at Ham FC) it was decided that it would be better to place the motor at the rear of the aircraft in order to put the microphone into clean air and away from the prop wash.

Building technique and materials

The main considerations here were cost and ease of use in building. Given that we have limited experience of building planes the construction techniques needed to be simple and effective.

Fortunately there's a lot of information about simple building techniques online and especially on YouTube. There's a great channel called experimental airlines which provided most of the building techniques used in the construction. Specifically it's worth checking out the videos for the Armin wing and carbon-fibre tail booms.

Wing construction

The material for the wing and tail surfaces is Kapa foamboard which is an expanded foam sandwiched between cardboard. We've been brilliantly supported by modelshop.co.uk who have supplied all the foam and a fair few other bits required for the build. Special mention to Andy there who's been incredible.
The wing stiffness is partially provided by a 2m carbon fibre spar which runs the length of the wing and is held in place using foamboard 'formers' which help create the aerofoil shape of the wing. All the carbon fibre used the build was supplied at a reduced cost by easycomposites.co.uk so a big thank you to them. Below is a photo of the wing mid-build showing the foamboard formers in place. The carbon tube is in the groove between the foamboard. As a weight saving technique holes were drilled in the foamboard which can be seen in the photo.

Bat plane Mk3 wing mid- construction

After a bit of gluing, sanding and taping, the result was a wing that measured 2m x 30cm along with a further 5.5cm of control surfaces. Both ailerons and flaps were made out of the foam and covered in packing tape to waterproof them and provide some structural strength.
The wing has a flat bottomed aerofoil section with a rough 15% height to chord ratio which should provide relatively high lift.

Here's the completed wing laying on top of the fuselage and v-tail section:

Bat plane Mk3 - completed wing

The next matter to attend to was how to attach the wing to the fuselage. The answer to this was handily provided by the Experimental Airlines YouTube channel in their carbon tail boom video.

This technique uses a sheet of carbon fibre/glass fibre/metal attached to the boom (fuselage) and elastic bands to hold the wing in place. Although the elastic bands provide an excellent form of damage limitation in the event of a crash (see the video linked above); we weren't 100% convinced that they alone would be sufficient to attach the wing to the carbon sheet. So, we used a couple of carbon strips and nylon nuts and bolts to attach the wing to the sheet in addition to the elastic bands.
The carbon we used was actually a carbon/glass composite (as it is about half the cost of a pure carbon sheet). It comes with one side not covered in pre-cured resin so as part of the build we covered this in resin to ensure we weren't going to pick up any splinters in use. Here's a photo of the drying carbon:

Carbon fibre sheets with resin applied drying outside

And below is the intended structure of how it'll look without the wings in situ. Note that the two parallel strips will be inside the wings.

Completed wing attachment assembly

Finally, below is how the wing looks with the wing attachment assembly in place. Hopefully it makes sense for how the elastic bands fit over the tabs on the large carbon sheet which is attached to the fuselage.

Wing and wing attachment assembly in place

So that's one major part of the construction completed. Once the v-tail was constructed, the large carbon sheet that the wing sits on was then glued to the fuselage along with a couple of small diameter (approx 5mm) carbon tubes to hold it in place and finally secured with a couple of cable ties.

V-tail construction

Prior to any of the building actually taking place a few numbers were worked out to ensure the plane would have a fair chance of firstly flying and secondly being able to be controlled in the air.

Part of this is working out the size of the tail and a couple of numbers called the horizontal and vertical tail volume coefficients. These are just formulas which you plug your numbers into and they tell you whether the tail will be able to provide the stability that its meant to.

To calculate the v-tail area and angle we used Dr Mark Drela's formulas here. He's a professor in the Department of Aeronautics and Astronautics at MIT so seems to be a pretty reliable source! We also used a value of 20% of the wing area for the 'horizontal' stabiliser and 10% of the wing area for the vertical stabiliser. 
The horizontal and vertical tail volume coefficients (Vh and Vv) were calculated as 0.41 and 0.031 which are satisfactory but not brilliant. There is a restriction on the placement of the v-tail as the fuselage telescopes in the middle so the tail and wings can only be placed on one half of the fuselage (over a distance of about 1m). If we could have used a non-telescoping tube for the fuselage then we could have distanced the wing and tail more which would have resulted in a better Vh and Vv.

We'd worked out that our v-tail would therefore be 20cm x 45cm and have an angle of 110 degrees between the tails (ie 35 degrees from horizontal each). Now all we needed was a method to attach the tail to the fuselage.

Quite a few ideas were brainstormed and although some would have provided more strength than the one that was chosen it's a trade off between weight and strength so weight was prioritised. A number of ideas involved carbon tubes going through the fuselage which was also not practical in our case as this would have meant a fuselage which could not telescope.
The idea we settled on was to attach two carbon strips to the fuselage at the required angle and have a single section of foam forming the two tails which would be glued to the bottom of the strips and the fuselage. Another layer of foam was then placed over the strips and tails and glued in place to provide further security.
Here's the carbon strips being glued in place:

V-tail attachment

There's no close-up images of the tail currently but what we do have is an image of a previous attempt to glue the tail in place that wasn't successful. It gives you an idea of how the tail attached to the fuselage and carbon strips, or rather in this case, how it didn't.

V-tail attachment failure

The issue was that the tail is covered in packing tape which helps protect it from the elements (well, just water really) and also provides a bit of structural strength.

The glue used on the carbon strips didn't adhere well to the tape and when it was placed down quite hard it simply popped off. This does give an idea on how the tail is attached to the fuselage though. The issue was fixed by removing the packing tape and cardboard underneath so that the carbon strips attach directly to the foam in the foamboard which provides a much stronger adhesion.

The control servos for the v-tail ruddervators (a mixture of elevator and rudder) were initially placed on the fuselage just behind the wing and long control rods and carbon tube guides used. After experimenting with this it was decided that the control rods were flexing too much and this would lead to poor modulation of the control of the surfaces. The servos were then moved to a direct placement on the tails and a much shorter control rod used resulting in a satisfactory setup.

To pod or not to pod

In the initial design shown in the 3D model at the top of the page there's a pod attached to the bottom of the fuselage. This was intended to hold all of the electronics equipment (including the detector) and the battery and also to act as a form of landing skid.

The final design of the plane doesn't incorporate this pod as it was realised that the majority of the electronics could fit within the free space of the wing with the battery and detector then being external. This was worked out as during the build the bits and pieces being used and constructed were weighed and it turns out that foamboard is actually quite heavy! The initial thoughts on the weight were that it would not be over 2.5kg but this turned out to be unrealistic when using foamboard due to its weight. Given that the battery alone weights 960g perhaps 2.5kg was an unrealistic goal for a plane of this size. The final weight of the plane is circa 3.7kg which did lead to some concerns over whether it would actually get in the air!

The lack of a pod also lead to one other concern:

Landing gear

The plane doesn't have any landing gear to speak of and instead relies on skids and a slow landing speed. The skids are there to protect the wing attachment assembly and the v-tail attachment assembly and motor. They are constructed of a high density foam (not sure what its real name is but in the RC world it's called blue foam). Here's a couple of images taken during the construction of the skids:

Essentially the foam was cut with a saw and then filed and sanded into a vague aerodynamic shape. It was then glued with Gorilla glue to the fuselage:

Fuselage with landing skids attached

Although the skids would provide some protection for the wing and tail attachment assemblies it's very likely that the wing will strike the ground upon landing. This isn't an ideal solution but the foamboard is flexible and should be able to cope with this. It's also easily replaceable should it tear or break. The main factor to limit any potential damage during landings is to have a slow ground speed.

Also worth noting in the above image is the wing in the background and the small pieces of blue foam used to protect the servo arms and control rods for the wing control surfaces. The servos are buried in the wing with only the arms sticking out so these are the only bits which require any protection.

The brains of the plane - the flight controller

The final piece of the plane is the bit that's actually going to do most of the flying for us, the flight controller.

We're sticking with the Pixhawk flight controller from 3D Robotics in the US as it's a tried and tested board and one we're very familiar with from our other testing.
There was a little bit of additional set up required using the built in v-tail mixer for the Pixhawk and with the addition of flaps but aside from this it was quite straight forward to get going. There was an issue with the compass being affected by magnets which were used to keep the wing opening closed but once these were removed we were good to go!

Below is the Pixhawk being configured and tested on the wing. You can also see the v-tail attachment in this image as well as the tail servo placements. You might also notice that we've got Mission Planner loaded up on a laptop. The laptop was kindly provided by a mutual friend, Mike Davis, so we'd like to take this opportunity to give him a big thank you as well, thanks Mike!

With the plane now completed all that was left to do was to test whether it actually flies!

The construction of the Mk3 was done in London at Tom M's house so the first test would be whether it actually fits in Tom A's car! Happily it fitted with a few cm to spare though we still managed to crush the end of one of the ailerons in the boot lid. Fortunately it straightened out to nearly new once we had it out of the car in Oxfordshire where we do all of our flight testing.

Maiden flight

We really had no idea whether the plane was going to fly as although all the numbers looked ok the real test is in the field! The major concerns were over the weight (as this was much higher than anticipated) and whether the motor would be able to put out enough thrust to generate lift quickly enough in order to fly. There was also a little concern over the flexibility of the foamboard for the control surfaces as they are quite long.

Here's what happened (you'll probably get a clue from the thumbnail!):

It worked!!!

Tom A was doing the throwing with Tom M standing behind at the controls. As you might be able to tell from the video it had no problem in taking off and the prop was generating more than enough thrust to propel it into the air no problem. Tom A actually mentioned that it's almost a case of holding it back.

The first flight was all with mid-flaps and the flaps were then tested during the second flight. The landing with full flaps was nice and slow and no damage was done to the plane.
It was also noticed that elevator authority could be a little better so this will be dealt with prior to the next round of testing.
In general though the flights went brilliantly and the plane was stable in the air and controllable which is exactly what we're looking for. We're really happy with the result.
The additional weight of the plane is also advantageous for windy conditions which we tend to get a lot of in the UK so at least there is a benefit to this.

The next step is to go through some auto-pilot testing and ensure the plane can fly autonomously and then do some surveys!

As always, watch this space!