With the uncontrolled flying of the Tricopter I decided to buy a smaller and cheaper model to practice on. I actually decided to buy two as it's always better if you fly with a wingman, so I bought the Prof. one too! During the week I have been chatting at work and there are a few others who look like they will buy one as well. After all they are so cheap and really great fun.
The two models I have purchased are slightly different.

First is the X4, which is a basic model with brushed motors. It is very lightweight and comes with a 3.7v 300mah single cell battery. We have found that this flies for about 5 minutes on a full charge when we use a slightly bigger 380mah battery. These are the same batteries I have been using in my Magracing cars.

Second is the Hubsan X4 107D. This is similar with brushed motors, but has a built in camera which can record to an SD card and also transmit back to a LCD screen. We haven't got the screen or a micro SD card yet (The Prof. has ordered some), so this has yet to be tested. Battery life seems to be similar, but it handles slighly better. Maybe this is because of a lower centre of gravity or the higher mass makes it less twitchy when flying.

The above are stock photos, but below is my actual one showing the battery and the hand controller. It does look like a toy, which I suppose is what it is, but I must say the hand controller had a quality feel to it and the quad flies like a dream. In the past week I have smashed this thing into everything in my house and crashed it multiple times in the field. I do break a prop every now and again, but this is rare as they tend to just spin off. There is a prop guard, but we have found that this isn't really necessary as after a few hours you learn to cut the power just before a crash and this tends to save the props.

I am really impressed by this and for around 28 for the basic one (51 for the camera version) I think this is a real bargain. It also has some red and blue LEDs on the prop booms so orientation in flight is fairly easy and they flash when setting up and when the batteries are nearly dead. All very easy to understand.
    On Friday night I tried the quadcopter with a second battery attached under the belly for a weight test and the quad flew really well, in fact altitude control seemed better as it was a bit more docile. I have now ordered some slightly larger batteries for the quad (500mah) to see if we can get a bit longer flying times. This would be nice particularly when flying in a field as the quad can be quite away a way. If I can achieve 10 minutes flight this would be OK. The body cannot take a bigger battery, but I am sure it would still fly with an even larger one. Maybe this is something to try in the future.
    Using this quad for a few dasy has really helped me learn to fly. I can now hover quite well and can move about the room as I please. I am onto the "Walking the dog" excercise, which is where you walk around and turn and follow whilst flying. I have not yet mastered flying towards me as my brain is not there yet, but I can turn about 45 degrees either way before I panic. Many recommend that you need to go through over 60 battery charges before it begins to sink in and I am probably at about 35 now. I am also trying to practice indoors as this is much harder and is good training. Maybe in a week or so I will have the basics sorted. I was actually quite suprised how hard they are to fly. Altitude control is very hard as you need to compensate whilst turning. It is also affected by surrounding objects. This is why learning to fly indoors is really good practice.
    Of course I cannot resist having some thoughts about modifications. I have been doing some research on the subject. Probably the best controller for me would be the Micro MWC (25 from Hobby King). The board is an integrated reciever, controller and brushless speed controller so would be a direct replacement for the board in a Hubsan X4. One big advantage is that it uses the DSM2 transmission protocol which means I can use my Spektrum DX6i transmitter to control it. This will be great as I can begin to learn the feel of the transmitter for when I return to the larger machines. One further advantage is that the board is arduino based and runs MWC 2.2 firmware. This is written in C and can be uploaded and modified with the Arduino IDE which I use for all my other projects. It is connected to the computer using a FTDI port and adapter which I have. I have one of these boards on order and also another second hand Hubsan (18 from ebay) so I don't have to butcher up this quad and be in the situation where I have nothing to fly.
    Another modifcation I would like to do would be to make the quad slightly larger and to take a bigger battery. There is a large community for this quad and many 3D models for alternative frame design. The size of this model is ideally suited to 3D printing on my Sintron Delta printer. Unfortunately it's not quite finished yet, but this weekend I got it almost complete. Hopefully next weekend I can get the printer up and running (I still need to make a better bed levelling probe). In the meantime I will download a few models from Thingiverse and see what I want to do.

Another Hubsan arrived today. This was the one I had spotted on ebay for 18. At first glance it looked in very good condition, but it was sold as "Maybe not working". The first thing I noticed was that the propellers were fitted incorrectly. It is important that clockwise and counter clockwise propellers are fitted on the correct corners or the thing simply won't fly. The Hubsan has a non conventional arrangement opposite to the other quadcopters I have seen. I fixed this and then fitted a battery. The transmitter bound to the quad, but the throttle didn't seem to work..... and then did all at once. It's a good job these props don't cut you like the larger ones! I then thought to calibrate the joysticks and it worked like a charm. Clearly someone had been messing and  got into all sorts of problems, but a little thinking paid dividends. The quad is now my first choice to fly as it's is new and handles really well. Here is a picture of the two side by side. The one on the left is the one I have just bought.

The one on the right has now been disassembled. I was interested to see what was in one and also I needed to model the parts in CAD so I could design an upgraded frame. Below is an image of the only part inside. It is a reciever, microcontroller and motor driver all on one board. I'm no electronics expert, so I looked up online what everything is. Here is a description

  • At the top of the board are two blue LEDs .The antenna is between them. it is etched onto the PCB.
  • A7105 This is a wireless tranceiver which handles radio communication.
  • To the left of this is a gold box. I believe this is a crystal. It is used as an accurate "tick" signal, probably for the microcontroller (I think 16.000MHz)
  • Left of this and not labeleld is the  Invensense MPU-3050 3 axis gyroscope. This tells the quad which way up / what angle it is!
  • The chip labeled Hubsan is a  Nuvoton MINI54ZAN MCU . This is the microprocessor "Brain" of the Hubsan
  • The chip labeled 3210AK is a 3 axis accelerometer. This tells the quad which way it is accelerating in 3D space
  • Below and to the left of the 3210AK and labeled 65Z5 is a  3V Low Dropout Regulator. This regulates battery voltage for the various PCB components.
  • 4 identical devices below this labeled 124K I believe are  N-Channel power MOSFETs. These take a small signal from the microcontroller and control the large analog voltages to power the 4 motors.
  • Battery power input is on the red and black wires at the bottom.
I got this information from this website .It has lots of other useful information.

All the connections for the Motors and LED's are on the underside. There are a few more connections unused on the board, probably for testing and maybe power take offs for the camera board which is not fitted to this quad.
    I am looking to design my own mini quad, so it is important to measure up some of these parts. I have also downloaded a few models from Thingiverse for inspiration. Here is what I have come up with so far. Click for a larger view

The yellow upper part I have taken from another design, but this will be modified. The controller inside is the standard Hubsan X4 as detailed previously. It would be nice to have a design that accomodates this and has room to fit the Micro MWC controller which I could swap in. The board hasn't arrived yet, but it does look very similar in size and shape. I placed the battery underneath held on with elastic bands. This should keep the centre of gravity low and make the quad quite stable. Most quads seem to have batteries on top. Maybe this is better, but I don't know. Perhaps the high C of G makes the gyros function well or it could simply be for protection. By slinging the battery underneath I can experiment with much larger batteries to see what flies best and for how long. 5 minutes is too short a flight time for me. You also have to take too many batteries with you.
    There's still some work to be done, but I enjoy this sort of designing. I must press on with completing my 3D printer.

29/04/2016 Quadcopter rescue!
A week ago the Prof. managed to land his quadcopter face down on his house roof. We tried a few methods to get it down, such as a tennis ball tied to string, but nothing worked. Today we tried a quadcopter rescue. Firstly we tied some string to a Hubsan and attached a magnet hanging down. Unfortunately the magnet we had was too heavy and swing about very badly. We then tried a hook made out od a split pin, but again this wobbled about too much. last attempt we thought to just land the quad near the one on the roof and then use it to drag both across. To our utter amazement this worked. It was a two man job, me standing close to the house and feeding the string and the Prof. piloting the rescue drone! When we had landed the rescue craft, I gently pulled the two to the edge of the roof. We thought they were going to get stuck in the gutter, but they both fell together. I cought them (Well broke their fall!). Thw quad that had been on the roof was all covered in green mould as it had been rained on many times during its stay on the roof, but we put a new battery in it and incredibly it fired up and began to fly. I am very impressed with the robustness of these little Hubsans. One of the motors did seem a little sluggish, so we wiped the thing down and sprayed a little WD40 around the motors. This improved things and we flew it for about half an hour. later on however one of the motor died. This had also happened earlier in the week to my Hubsan.
   On returning home for the weekend I stripped out the faulty items on both and it was quite easy to change them over. The circuit board has some reasonable sized pads for soldering to. It is interesting to note the hubsans with no camera use 7mm diameter motors, and the camera version uses 8.5mm motors (These are actually sold as 8mm diameter motors!) Luckily I had ordered some of the big type ones for my next custom build and I had also bought a second hand quad on ebay which I could use for the 7mm motors. The three flying Quads in our range are now flying

The postman brought some goodies today. I ordered some microcontroller boards both from the UK and Hong Kong and both arrived.

The one to the lower right is the Micro MWC board. This is a complete system for a micro quad. On the board is a DSM2 protocol receiver, arduino compatible flight controller and 4 brushed drivers. It is basically a direct replacement for a Hubsan board, but I will be able to program it from the Arduino Intergrated Development Environment (IDE). Above it is the yellow coloured board. This is a Future Technology Devices International (FTDI) board. You plug this in to the Micro MWC and it translates USB data to data the board can understand. Normally Arduinos have this chip on board, but to save weight this has been put on a separate board.
   The board on the left is a real find and is absolutely tiny. In fact its a little too small for me so I may look for something slightly bigger. This is a Deltrang RX41d. You can't upload firmware to it, but you can interface with it using your transmitter joysticks to modify the basic setup. Again it is a DSM2 protocol reciever (RX), but has no accelerometer ot gyro, so would be of less use in a quadcopter. It is really menat for use in model trains and cars and this is probably what I will use it for. I expect the range may be limited, but I have plans to do a 1/32 RC car based on my Magracing chassis and I feel that this could be used. I said however that this is very small and may prove difficult to solder, so I may end up using a larger board. I have a few other boards coming from China, so this gives me options. The delivery from China is quite long, so it pays for me to order them in advance. They are also much cheaper this way.
   We are now beginning to collect a few quadcopters and there are a couple of people at work buying them. With this in mind I thought it might be time to do something to identify my quadcopters to avoid any mix ups. Today the formation of the "Black Widow" squadron has occured!

Ok its not very original, but at least it's easy to paint. I'll try to think up something better soon.

I have been having lots of fun flying the Hubsans during the week and at work we have even formed an unofficial club at lunch time so we can get some practice in. There has been another two people who have bought a Hubsan and are busy learning how to fly!
   With the completion of my 3D printer on Friday, I set about printing a new frame. This went quite well and so I quickly knocked up a flying version to test it out.

Ok it's a little rough and held together with insulation tape, but I was eager to get it in the air! It uses the larger 8.5mm diameter motors and the standard Hubsan X4 main board. I strapped a 500mah battery underneath and it took off immediately. I was able to control it quite well. It's very fast and light. I had a slight yaw issue, but nothing I couldn't handle. This was down to a slightly twisted motor (not vertical). Whilst printing the frame I noticed my printer prints slightly smaller than expected (I still need to dial it in exactly) This meant I had to drill the motor mounting holes out with a drill. Clearly one was not quite vertical. It's amazing as soon as I throttled up I could see how I could improve things. I am going to make the next one with the main board underneath the frame, then a second piece below that which holds the battery. This does a few things, firstly it exposes the underside of the main board. On a Hubsan all connections are underside, so this means I could mount the board and then solder in the electrical bits. Secondly A flat upper surface would enable me to add on the top any pod I like. I am thinking of an FPV camera or maybe a shaped body in foam so I looks a bit nicer. The two LEDs mounted forwards work well. They actualy shine straight forward like headlights. I dont miss the rear LEDs, so may leave these out, or just have one. The motors need lowering in the frame as it currently looks like an upturned table! I might even make the frame about 10mm bigger as the motor wires may allow this. I'm really pleased though as my 3D printer produced this in about 20 minutes and I was flying in about an hour. I'm really looking forward to version 2!

It was time to progress the design of my own micro quadcopter. After the print and initial flying of the HF12a (Helium Frog robot project 12) I could immediately see many improvements. One major change was that I thought the main body would be better printed the other way up. The hubsan control board has all the wiring connections underside, so this makes sense. After a couple of nights work this is what I have come up with....The HF12b!

All the electronics are enclosed and fitted from underneath. A plate screws on and below that is a battery box. I have designed a box for both the 500mah battery which is the one I mainly use, but also another one for a 1000mah. This just about flies with the smaller 7mm motors, so I am hoping this might work with the larger 8.5mm motors I will fit on this design. Holes at the front allow air to flow through and out the back to keep the electronics cool. There are 3 LEDS, two at the front and a single one at the rear. Channels inside feed the wires from the motors to the control board. I have also placed some small location tabs which should accomodate both the hubsan control board and also a Micro MWC board should I wish to fit that later. 4 M2 cheese head screws hold it all together. The top surface is completely flat. I could use longer screws and attach a camera pod to the top. There are probably many things I have missed in the design, but the benefit of a 3D printer is that another one is not too far away should I need to change things. I have designed all parts so they can be printed without support material so they should print quite quickly. I am just waiting for the weekend to come!

I managed to print all the parts off at the weekend. The first print came out 95% of the desired size and scaled unequally in X and Y. This indicated an error in the 3D printer machine dimensions, but I checked everything out and could find no obvious errors. I wondered why the Marlin firmware doesn't include simple scaling variables, so you can stretch the print as needed. This is what you have on most firmware. I worked my way through the code and added in one. I then ran some simple test prints and calibrated the X and Y sizes. The next print came out really well 100% size in all directions and all holes required very little reaming out. The design above had a problem and that was the motor wires can get trapped easily and break during assembly  I have now fixed this in the updated design by having a small gap along the motor arms.
   After the quad was assembled I took it for a test flight and it flew well, however on the second flight one motor refused to turn. I checked the motor by directly attaching it to a battery and it was fine. It seeems that the 8.5mm motors may damage the main X4 board (Probably overload the FETs). I replace the main board with another and the same thing happened (Even with new motors). I'm not sure what to do now. Either go for a board from the 107D (which has larger motors), build a 7mm motor frame, go for the Micro MWC board, or abandon this and just build a bigger brushless quad. I don't really like being defeated, so maybe I'll have another go when new parts arrive. I have now an alternative 7mm motor design (HF12c - 7!)which I can use for trials, but this will almost certainly fly no better than the standard Hubsan. In fact the HF12a design I started with might be worth a  revisit for a lighter design with 7mm motors.
   The Micro MWC controller board is programmed using a PC via a separate FTDI board. In order to connect to the controller, some means of easy access is required. This lends itself to an inverted design from the one above with a removable top plate. Luckily CAD is ideal for this sort of design change.
The design also enables me to fly the quad with the top plate removed so I can easily connect the FTDI board for tuning..... Yes I think this is the top candidate for the next print (HF12c - MWC)!

Home for the weekend and the 3D printer was running all night printing quadcopter parts! Whilst this was ongoing, I took apart a Hubsan X4 and documented all the wiring. I also weighed all the individual parts to see how my design compared.

Weights are as follows for the basic Hubsan X4 :-
Frame 10g   Motors 13g   Control Board 2g   500mAh Battery 11g   LED's 1g   Props 1g   Feet Neg   Total 38g

The Hubsan X4 107D with on board camera and 8.5mm motors  weighs 51g

Quite light and I don't think I could improve on this, but my design is sligtly larger and made without the benefit of plastic moulding.
   My first design of the evening was the HF12b with 7mm motors. This is basically a Hubsan with my frame design.

56g total. Not too bad considering it's a bit bigger and rapid prototyped. I was very impressed with the first flight. It flies much better than a Hubsan and reacts quickly. This must be due to the props bieng about 10mm further from the centre of Gravity. I noticed that i was at about 80% throttle to make it fly and after about 2 minutes flying I was at full throttle. So i learned that 7mm motors won't do for this design, which is what I suspected. However this encouraged me as It flew so well at first.
    A few hours later, the 3D printer had spit out the next frame parts. A HF12b with 8.5mm (Hubsan 8mm!) motors.
62g  A 24g heavier than a standard Hubsan, and heavier than the camera hubsan which is 51g. This one flies really well, very punchy and flies on about half throttle similar to a standard hubsan. I'm very pleased and it assembled quite easily from underneath. One slight problem was the wires to the front motors were only just long enough. I think I will move the Control boars a few mm forward in the design to improve this. Also in a crash the battery comes out, So i need do add a bit more of a retainer in the battery tray. I might even ditch the tray and go back to an elastic band setup as this may save a few grams.
  The next job was to print out the HF12c, which is the top loading design specifically for the Micro MWC controller. Whilst fitting the controller I decided to add in some pcb pins, so If I want to experiment I can resolder to a pin rather than to the board. This means I may need a slight bulge in the top plate. Unfortunately I ran out of 8.5mm motors, so this will have to wait until next week! The Prof's quad is also in for repair and I had used two motors on that. I do have enough 7mm motors and parts to make up another standard Hubsan, so I may build up a spare quad.

Tonight I completed the design for the next version of my quadcopter. The HF12d is a lightweight version of the one above which weighed in at 62g. The above frame top half was about 22g, and my new frame I estimate at about 15 - 16g. It's difficult to be exact as the 3D printer prints at about 50% infill, however there is probably not much area in this frame for a hollow interior, so the estimate should be near. This makes the weight less than 55g plus the removal of the lower battery tray and bolts. 50g or less seems a good target to aim for.

As you can see the battery tray has been removed (and replaced with a small flat plate. The battery will be held in with a few rubber bands. Unfortunately the wiring will be visible, but hopefully I can do this neat and cover with small bits of aluminium tape. I don't think there is much more room for weight saving, but we will see how it flies. The top is again flat to allow any camera gear to be fitted later. I'm keeping the characteristic shape as I liked it when I flew the last one and it was very easy to orientate in the air. The hammerhead front is also very strong.

I finally have a design that I am happy with. The HF12d shown above has been printed out and is flying.

Total weight ready to fly with a 500mAh battery is 53g. So only 2g over the weight of a camera carrying 107d. However it must be noted that this quad is 20mm wider and longer, so not too bad. I could remove the battery tray underneath and carry the battery on top and this would save 3g. It flies really well and is very quick and responsive. As it's bigger you can fly it a bit further away and still be able to see its orientation. Underneath it's a bit messy, but it works OK.

Currently I am using aluminium tape to secure the wiring. this tends to come off, so I will look for something better. I may try kapton tape or possibly having a few holes and use thread to sew the wires in place. The circuit board is held in place properly with two screws and is protected in the centre of the craft. Routing the cable from the two front motors is also a bit tricky as the wires are only just long enough. I have come to the conclusion that the secret to a great quadcopter (particularly a small one) is the balance between motors, props, battery size / weight. I don't think I am getting the benefit of the larger motors, so if I can locate some slightly larger props I might give that a go as larger props tend to be more efficient if they are coupled with the correct motor. Hubsan have clearly hit the sweet spot for this equation and I am sure there is one for my craft. Even so as it is I think this is the one I will take flying with me from now on. Flight time is about 6 minutes plus.

The Prof. paid a visit today to use my lathe to modify some rocker spacers for his type 1 engine. This gave us the ideal opportunity to go up to the fields and fly everything that was airworthy. The Squadron had 5 kites in the air!

During the outing the Prof. was determined to loose sight of anything he was flying in a vertical direction. The white Hubsan achieved an incredible altitude, almost out of sight. It was actually a bit windy to be flying, but due to the massive amounts of room we had, we pushed the limits a bit. We ended up searching in the fields about three times, but did retrieve all of them. A particularly handy technique is to walk in the field and max out the throttle and listen for any buzzing in the grass. I managed to get one stuck in a tree (like i said it was very windy!), but a bit of throttle and the wind got it out. We did however have a couple of casualties. The tricopter failed again. One motor seems to have failed and it has a wierd problem where one of the motors stumbles and it veers off for no particular reason. I don't really like the way it flies and whatever we do in the settings is only a partial success. The White Hubsan also fell from a very great height (Well maximum altitude really) and unlickily landed on the only bit of tarmac in the area. One of the booms is completely broken. Not to worry though as it will be stripped for parts and the board will probably find its way into a printed frame.
   The Prof. also flew my two new 3D printed quads. he didn't like how twitchy they were and I must agree with him. We came to the conclusion taht because the PID settings are unadjustable on that board, having a larger frame (about 20mm wider  diagonally) was causing issues. The motors are more powerful and pull from a greater distance from the C of G. This makes the on board firmware over compensate. Not to bad indors in standard mode, but awful outdoors in expert mode!
   All in all a very enjoyable day and I learned quite a lot about what I was doing. Even the tricopter helped as i was adjusting the PID values on the KK2 board and seeing how it affected the craft in the air. On returning home I modified my quad design and printed off the design HF12f design.

This is a 92mm across corner design (The same as a Hubsan X4) with no battery tray and as light as possible. After printing it was assembled up and weighed in at 47g including the 500mah battery. The frame is 12g only 2g heavier than a hubsan frame. So this would be suitable for 7mm motors. The motors I used seem a little suspect and they speed up and slow down, maybe it's the main board or prop imbalance, but either way flight time is 7 minutes 38 seconds so I am happy with that and it would only improve with better components.

As you can see I also changed my 3D printer filament. We will now be printing in orange for a while! The battery holder works really well, in fact I only need the cross elastic band so I will delete the front horn on the next print. I also think that the battery can sit directly on the PCB, so this gives me the opportunity to simplify the holder a little.

01/06/2016 - Crash Testing!
This evening the Prof. and myself took the new quadcopters for extensive field testing in North London. There is a great field near his house with a sort of raised perimeter (for flood defences) and quite long grass. We flew them at full speed and maximum range and of course crashed them several times. Two specific incidents caused an airframe failure. One was hitting a tree (The main trunk) and another was rolling over several times on a gravel path. Both of these caused a fracture in the front booms. The crashes were caused by losing orientation at distance and flying on a nearly flat battery!

First was the newer airframe and this sheared behind the front boom. Clearly the lightening holes at the front are a bad idea. The quad still flew, but a rethink here is in order.

Second was a complete failure of the front end. The crack had begun from a sharp internal corner on the underside (I have removed this from the latest design. Interestingly though the crack had gone around the hexagon holes along the boundary between the infill and perimeter. This is a sort of known weak point as some of my perimeters have a slight gap here.
   I am actually quite happy that we managed to do this as this is the sort of thing I need to do to improve my design knowledge. In my day job I am pretty familiar dealing with stress raisers in conventional metal and plastic parts, but 3D printed parts have different failure modes. It is pretty clear now when you look in the above image (See around the rear hexagon hole set) that there are clear fracture lines around the holes.  Things to do for the next print will be as follows:-

1) Remove the hexagon holes at the front of the quad. I probably could keep a few in the centre.
2) Change the parameters in Slic3r to give more perimeter to infill overlap.
3) Increase radii behind the front booms to give additional strength as this is where the cracks start.
4) Reduce the motor pod diameters. There seems to be no stress here to speak of.
5) The Prof. also suggested that I have two different LED colours, maybe one at the front and one at the back. I will look into this. It might even make the wiring a little easier. We both thought that an LED colour change might be easier to spot in the air as the quad spins around.

Another test we did was strapping a 1000mah battery to the bottom. It flew and quite well, but it is more difficult to fly due to the greater mass. You have to think and anticipate much earlier to overcome the momentum. This makes flying less enjoyable and crashes more likely. As expected flight time is improved a little with the larger battery.
   We were both impressed with how fast they were and they are much more fun than the original Hubsan X4. In fact they are easily fast enough as you can go from far left to far right of your flight envelope in a matter of seconds. I really would like to try some larger props though as I think this may slow it down a touch, but give a longer flight time. We also expect that if we want to break any altitude records that this is the one to use as the gain in altitude on full throttle with the larger motors is incredible.

03/06/2016 All Stressed Out!
After the failures in the week it was time to print out some replacement frames. But first I needed to improve the design to give me more strength in a crash. The target was to improve the design, but keep the same weight.
I am no expert at stress analysis, but I do know the basics. The Von Mises analysis(Finite element) shows the following.

I anchored the two rear motor booms and applied a static load downwards on the front motor boom to simulate what happens when the quad strikes the floor hard. I can see straight away where the high stress is (Shown in red). So the design fails there and the crack propogates to the hexagonal holes or around the weak join between perimeter and infill in the 3D print.
    First change was to beef up the edge. I did this by increasing the outside radii from 8mm to 10mm. I left the inside the same, so the edge gradually thickens up. Secondly I removed the troublesome holes at the front and replaced it with one slot. This is at the centre of the quad frame where there is much less stress. Next change was to add a small bridge between the weak area and the central battery support. Hopefully this will transfer some of the load through to the rest of the structure. To reduce the weight back to what I had before I thinned out the motor pods by 0.5mm and changed the design of the air holes. As the motor is a strong steel tube, the surrounding pod doesn't have to be too strong. I also narrowed the battery supports from 6mm to 4mm and changed the front LEDs from two units to one. By doing these changes I reduced the weight from 11.08g to 9.989g (Not including the one less LED)!

You can see in the image below that the stresses now look much less angry. Hopefully this will improve things when we fly the design next week. I am just printing off the new designs and with luck I can salvage all the parts from the previous two wrecks.

05/06/2016 The Box of Dreams!
Over the weekend I managaged to find 4 Hubsan main boards, so the challenge was on to build 4 quadcopters! As I built each one I took the time to fly it a little and do some subtle modifications. I also had a spare biscuit tin which I lined with some car sound profing to create the "Box of Dreams!".....a small hanger for 4 quadcopters. In top of the tim there is also just room to store the batteries and some spare blades. The design changes in each quad are subtle and mainly for weight saving, but the last one I printed I modified the motor pods to take 7mm motors. This is the red one on the top right of the image below. Interestingly this one weighs in at just 38 - 39g (The scale hovers between the two values) This is very close to a Hubsan X4 at 37g.

The design I am most happy with is the orange one at the bottom left, this is very strong at the front, but I have added lightening slots on the rear booms. It is very dynamic in flight and has good flight duration. Hopefully during the week I can have a good fly of them all and to see how they stand up when crashed. I do think there is a bit more weight saving to be had and I would be happy if I got it to below 37g for the 7mm motor version. As an unexpected bonus, the slots in the booms make the quad have a distinctive whooshing sound in flight..... All adds to the uniqueness of the craft I suppose!
   I feel that the design of this quad is now coming to an end as I am really happy that have a design that I can print off easily and it takes only about 20 minutes to clean up the print, fit the motors and board and get flying. To take this project further I perhaps need to start learning about firmware setup, PID control etc. I still haven't a craft with the micro MWC flight controller on board, so this seems the next logical step. A friend at work also sent me a video of a small brushed quad fitted with a tiny FPV (First Person View) camera on board which enables you to fly via goggles as if you were on board. This is quite common in the multirotor community and it was time I dipped my toe into that.

During the week we have been flying all the "K frame" style quads I have been building. It is amazing how they all seem to have slightly different characteristics, even the ones that are seemingly identical. It must be the way the frames print and the differences in the motors. I really like this though as it makes flying so much better. The Prof. did loose one in a bush and we couldn't find it and we did crack a couple of frames, but again I have modified the design a little to account for the weakness. Unfortunately I cannot seem to modify the Profs flying style though! I also discovered that if you print the frame at 45 degrees on the bed, the printer will print the layer and infill in a horizontal and vertical manner. I am trying this as I think this may make the print stronger in the places where I need it.
   Whilst printing off some frames I decided to try something a bit whacky and go for 7mm motors at the front and 8.5mm motors at the rear. I was sure it wouldn't fly very well due to the imbalance, but guess what, It flies really well. In fact it seems slightly better! This was a complete surprise and it seems to have the punch of a 8.5mm frame, but with slightly improved flight times. I'll reserve judgement intil I have flown some designs back to back a little more, but I think this may be one of those happy accidents that come along once in a while.

I'm going to call this the "Hot Rod Quad" as it has large motors at the rear, like a hot rod car has larger wheels at the back.
   Today I began the next stage of my learning and built a frame with the Micro MWC board. This is an inverted frame design, where the flight control board is loaded from above.

I did try out the quad with whatever firmware it was loaded with and it bonds immediately with the Spectrum Transmitter. All the motors turn in the correct directions, but it tries to flip side to side on power up. I recognise this problem from building the Profs' quad (maybe more on that later!) The motors are wired incorrectly, basically the motors slow down when they should speed up. I wired it in accordance with the diagram, but it appears the firmware doesn't match this. Rather than rewiring I will upload some new firmware (Probably Benedicts firmware) and modify it to match the wiring. I also need a battery connector, which should arrive shortly. This work is something I can do during the week. As the board is 25 I now have about 32 in the air at one time. I think I may add a lost buzzer to this one, so we can find it in the grass. I can do this within the firmware also. I want to take Benedicts firmware, strip out all the excess code and learn a bit more about how things work. This cheap replacable frame is the ideal platform on which to do this learning and if I can keep from losing it whilst flying so much the better.

This evening I cleaned out the software a little, but to be honest I got a little lost in there. The code is very complex as is usual when many people contribute and try and outdo each other for cleverness. This is one of the downsides to open source code it soon gets hundreds of features you don't want. However having said that it does fly the quadcopter like a dream and the little graphic user interface works really well.

I toyed about with the PID settings and found that low P values make it feel really nice like the Hubsan I am used to.
   Next job was to add in a "Lost Quad" buzzer. I had one from a bigger quad, so I connected this up to Digital Pin 5 and the positive voltage next to it. A few lines of code in the firmware and the buzzer can be activated by holding the throttle low and the roll stick to the left. It works quite well and might come in handy.

Of course the buzzer is massive in this quad, but it was all I had. One downside of the Micro MWC board is its very small and soldering is a real pain. I really don't like this board much as it is also quite hard to upload the firmware and you have to use a little pin contact to trigger the upload. The board is expensive at 25, so I don't think this is the one for me.

This weekend saw production of another two variants of the design. . I was looking at the DroneProz Scorpion LTR120 which I quite like the look of.

It appears to be a very neat design and is brushless with FPV. This is the sort of quad I would like to design. In order to get my eye in as to how it looks in the air I did a quick boom extension of my design and printed off a "K frame" of similar size.
Below is the HF12i, a larger 120mm across corners quadcopter.

As you can also see I have taken delivery of some new coloured filament for my 3D printer (Cyan). The print wasn't too good as I was messing about with the print settings, but it was OK to build and is quite strong. I printed it a 45 degrees on the build bed and this means the hatching is horizontal and vertical. This makes the print much stronger for these quad frames. It flies OK with good flight times, but is a bit jittery. I think this is because the P settings (PID) is too high in the hubsan board and the motors being further from the centre of gravity react too fast whilst correcting. It's a bit distracting when hovering, but OK when moving about. Of course with a programable board I could set this up properly, but the hubsan board has fixed settings. It will now sit in the box of dreams so I can fly it a little more to see how I feel abouth this size of quad. One other thing I tried was some slightly larger Parrot "Rolling spider" props. I think these are 65mm diameter (The hubsan ones are about 55mm). I was expecting slightly longer flight times, but it was almost identical at 7 minutes. These props are much more expensive, so this isn't something I will buy again I don't think.
   During the week I also messed about with a Mini MWC board and like the micro MWC it proved troublesome to load in the firmware. I'm not going to buy any more of these boards as they are too much trouble. As ususal Hobby King also give very poor support with almost no documentation, so I can't even use it for my radio controlled car as I have no pinout information.
   Next build was the HF12j. I wanted to make the smallest quadcopter that could fit my new FPV camera from This is the 5.8Ghz LHCP VREG version (I think I bought the wrong LH aerial one though). I drew the Camera in CAD and designed the frame around it. The smallest I could go was 103mm across corners. This was printed and the camera attached with velcro. I also glued in some little 3mm pegs made from 3D printer filament to help locate the camera.

This one is in fetching lime green! This little quad struggles to get in the air and you need full throttle. Once airborne it wobbles about a bit and this is probably due to the high centre of gravity. Again I cannot correct this using PID settings as the hubsan board doesn't allow any changes. I am also struggling to get a decent picture from the camera in to my UFO FPV goggles, but I am working on this. I do believe I am expecting too much from these small cameras as I think they are generally give out poor video quality. The goggles also don't seem to have the correct focal length for me, so I am in the process of doing some mods to improve things and have ordered some new fresnel lenses to experiment with. I'm still learning all about this stuff, so I may be doing something wrong. More research needed!
   I think I am now at the end of development of these small quadcopters for the time being and over the weekend I flew all the different variations I had built. The best one for me is the standard 92mm quad with the 3D printed frame and the 8.5mm motors. This is closely followed by the "Hot Rod" 92mm version with the 7mm motors at the front and 8.5mm at the rear.
   It is now time to move onto brushless designs. This mainly due to the fact that I need a little more load carrying capacity for the FPV gear that I want to add. It is possible to build a brushed FPV quad, but it will struggle and flight times will be quite short. Maybe I will return when I have learned a bit more about what I am doing as I do like the smaller designs better.
   I will always have a few of these quads with me in the "Box of Dreams" should we crash everything and are left with nothing else to fly! They are still great fun. One of the major benefits of them is that you can fly them in public parks and at work and no one minds as they are virtually harmless if they hit anyone and really cheap should you lose them.