TOM 2 Autonomous Robot Bug

This was a little project I'd been wanting to do for a little while, just hadn't got round to it until recently. It started back ages ago when i got stuck in a park for 2 hours waiting for a lift. So I found a little shield beetle and started to watch how it behaved (as you do) and i observed that it's behaviour broke down into a set of relatively simple rules. Later I decided this could be replicated using relatively basic circuitry and a little bit of cleaver wiring. So a few weeks ago, I sat down and started fiddling around with my old favourites, relays, to see if i could come up with anything interesting. after a few hours of soldering and re-soldering, I had built this:

The circuit diagram below is set up to show how it's wired up, and the position of the wires:

This little robot would have worked quite well, if i had used a different propulsion system. as it is, the 3 volt motors had to have a separate power supply to the 9 volt relays, and only worked on very flat ground. The logic itself worked really well. I may re-build it with better motors at a later date if i get round to it. The whole project could have been done with two 2pdt relays instead, but thanks to T.I.M. i'm rather short on those. To illustrate the logic, i drew up the following diagram:

Basically if it hits an object, it turns away from it (top right hand pictures). If it can't do that, then it reverses then turns away (bottom left). It will also reverse and turn away if it gets stuck and a feeler gets pushed hard (bottom right). For 4 relays and a few capacitors, that's not too bad.

Land Of The Dead

OK, so since i said in the subtitle i make games, i thought i should do a post related to it.

This is my latest game, Land Of The Dead. It's a zombie shooter where you run around shooting waves of zombies with an AK-47. Pretty standard stuff really, but i was pleased with how the game came out and it looks almost half like a real game. Gets a bit boring after a while, but ah well.

These pictures are some what out of date as the game has changed a lot since, but you get the idea.

You can download it by clicking here. It is an .EXE file and need uncompressing first. It will only run on Windows with Direct X, and I'm not sure how well it'll work on Windows 7 or Vista... so good luck.

I write all my games in Dark Basic, in case your interested.

The Trike

The trike was the first vehicle i ever made. It was built in one week in the time i had left after school, using no welding as the welder had just been stolen. I believe i was 14 at the time.

It was impossible to drive, horribly unreliable, and uncomfortable. But it was still awesome and i enjoyed it greatly.

It had a 67cc 2.8hp Honda general purpose engine on the back, half a stacking chair as a seat, the front half of my sister's old bicycle on the front and the wheels off my old bicycle on the back. The rest of the chassis was made out of some bed frame. The throttle was a piece of string you wrapped around your hand. The whole front end was put on badly so it was tilted to the right, which meant you could only turn right at speed without falling over. It was so hard to drive i could never find anyone else who could drive it more than 2mph. with a good run up this could make 10 or 15mph which is plenty fast enough.

The best part of the build i remember was when it first moved under it's own power. As it was the first vehicle I had made, I remember the great excitement of it slowly trundling a few feet before the chain fell off (a recurring problem through out it's life).

This whole project was inspired by my mate's mini motorbike which we had been riding up and down his garden. it was great and i really wanted one, buuut i had no money so this was the next best thing.

It has long since been dismantled. most of the back end was re-used in the Snow tug

Here's a video of it going.

TOM The Turtle

Back in year 7 or 8, i was browsing through a book on robotics in the school library. In it i came across Gray Walter's robotic turtles Elmer and Elise. These were two robots built in 1948 which were very simple but completely autonomous and very life like in their behaviour based around the intensity and direction of light. The book had a picture of it with it's cover off. I thought 'That doesn't look too hard'. Unfortunately, I never got around to building it until last year at the same time I was building TIM, but it turned out not to be nearly as easy as I first thought. 


I made the front wheels, steering and chassis out of bit's of old change machines (my dad builds them so there were lots of spare parts). Then it came to the circuitry. I did many hours of research into how they worked and behaved, and i eventually came across the original circuit diagram. It consisted of two tetrode valves and two relays, with a couple of other passive components thrown in. Not having any valves, i redesigned the circuit using transistors. This required a full understanding of how the circuits worked. Now not being particularly familiar with valves or transistors this wasn't easy, but my dad helped a lot and we eventually got a circuit that worked. put it all together and a lot of tweaking and it finally worked.

Snow Tug

Winter 2010 saw the rare event of it actually snowing enough to be fun. Sliding down a hill on a toboggan is all well and good, but then you have to pull it back up to the top again. So i quickly designed a snow tug, and gave myself a day to build it. And at 11:30 that night, I wheeled out :

It was great, 15mph on a toboggan is pretty darn quick. So as a proper test i drove it down the icy lanes to our near by village (which is where the photo was taken). Here i spent the day whizzing up and down the street much to the amusement of passers by. From this i discovered two main flaws. One was the lack of grip, so you would grind to a halt at the slightest incline, and two, the wheels would spray you with a snowy mush which meant you couldn't see. I therefore simply fitted mudguards, and wrapped chicken wire around the wheels like snow chains. It could now pull you up relatively steep hills, without the icy shower; Power slides on a toboggan is one of the more exiting things I've come across :)

Once the snow had melted i made a sort of wheeled toboggan and removed the chicken wire. The new vehicle looked bizarre as you had to put your feet up on the engine and have the handle bars between your legs, but it was great fun to zoom about on. unfortunately one of the tires on the wheeled toboggan burst beyond repair, so i'll just have to wait until it snows again.

This is a video of me driving it into the yard and power sliding sideways, which was always awesome, however this time i over did it and fell off!

Chainsaw Powered Bike

I had been looking at building a powered bicycle for some time; a quick and easy way to make quite a cool vehicle. So in the winter of 2009, i went out and bought a chainsaw engine for £10 from a local guy who I buy most of my parts from. I then set about attaching it to the bicycle i bought for the purpose for £5 a month or so before. It was a relatively easy build, i can't remember encountering any huge problems in the construction other than having to go back 3 times to find a working magneto to get the engine going. However my first attempt, as shown below, wasn't geared low enough, so it took an age to get going. 

So the next day i set about attaching a second lay shaft which doubled the gear ratio.

The main problem now was getting the chains to stay on. It was a recurring problem with this vehicle that was fairly predictable with 3 separate chains and lack of precision, but once you managed to set it up right it went along nicely, and with the five gears on the back wheel usable, you could get it up to quite high speeds.

I once managed to fall off it spectacularly whilst going round an icy corner fast (with no helmet on :s ). Hurt like a b*tch but surprisingly it still worked fine after i put the chains back on. Sadly though, when welding the drive sprocket onto the clutch, the metal must have annealed, so when the chain slipped as it sometimes did, it would ware down the teeth on the cog meaning it would slip more which meant it wore down the cog more... eventually i was left with a completely smooth tube where the cog used to be.

If you wat to see it working then the here's the link for a video: http://www.youtube.com/watch?v=0MwvWK-_vQk

Ok so it's a year later now and I've finally got round to fixing the cog (only took a few hours). I had forgotten what an awful piece of engineering it was. I'm amazed it even moves at all!! but i managed to get it going pretty well and fitted a bike speedo. It even ran long enough to do a speed run: http://www.youtube.com/watch?v=qO-hxl4AYaE
In this vid i got it up to 24.5 mph, but I've hit 30.4 before, and that was in 5th (it has 6 gears but only a five speed selector atm). However having only 10hp per ton, acceleration isn't huge.

Infinite Irritations ( Homemade Synthesizer)

This was a project i'd wanted to do ever since I found out about Schmitt Nand synthesizers, but only reacently got around to it. It basically consists of a quad nand chip set up to make 4 separate Schmitt oscillator, 2 slow, 2 fast. I then modulated the two fast ones using the two slow ones, and then combined them to give a two tone output. The effect of this allows an infinite variety of sounds, all really annoying! It was small enough to pack into an Altoids tin with a battery and a speaker. Now I could drive everyone in a room insane at the flick of a switch... I'm such a great guy :P

For an example of the noises it can make, go to: http://www.youtube.com/watch?v=X8qRPz9VdF8&

Bicycle Canoe Trailer

My mate asked me to build him a trailer so he could tow his canoe with his bike. Sounded hilarious so i said yeah why not. 2 days later:

Nothing special but awesome all the same.

It was a surprisingly nice piece of engineering considering my normal standards, i even used paint! We took it out on the road for a test run when my mate came to pick it up and it worked really well, even tried it fast over a bumpy field and everything was fine.

All round a lurvely machine.

The Life and Times of TIM (My Relay Computer)

EDIT: This page is very out of date and will not be updated as my computer projects have all been moved to their own website: http://www.northdownfarm.co.uk/rory/tim/


I have always been a man of science, and as such i would usually have a rough understanding of how everything works. However one thing that had always bugged me was computers. I could never work out how a computer could turn 1s and 0s flowing through logic gates into what you see before you.

So in the summer of 2010 when i was just finishing my GCSE's, I decided to do a little research in to the matter. So I did a few searches on YouTube, and came across this video, which shows you how to make a simple 4 bit adding circuit out of transistors. It contains a circuit diagram in logic gates of how to make a simple half adder. Intrigued, I went out to the shed to try build the thing.

Now I had never been very good with electronics; the only circuitry I ever really had done was with relays, so I thought I'd try and apply some of the principles shown in the video using relays. I started by making up a few basic logic gates and wiring them together into a simple half adder as shown at the beginning of the video. I then worked out how to do basic latch memory. Lo and behold, TIM 1 was born. You may be wondering why it was called TIM. Basically I just like the name Tim, but the official reason is because it stands for The Intelligent Machine... or something like that. When I powered it up it would add 2 binary digits together. This was surprisingly satisfying seeing as it wasn't very impressive, but it meant I had finally grasped the very basics of computing.

This first attempt, as shown above, was the start of a years worth of design and development and sweating over a soldering iron until the small hours of the morning, but I'll come to all that later.

In the days following, a rush of developments and breakthroughs occurred (some of which seem painfully obvious); I discovered that an XOR gate and an AND gate would do the same as the 3 AND's, 2 NOT's and an OR as used in TIM 1 (which meant spending ages trying to develop an XOR gate made from just 2 relays), That two wires soldered together would make an or gate instead of having to use 2 relays, and that a diode could be used instead of my special 'Isolators'. All these developments meant less relays were used for any particular system. This was critical as I only have the relay's I can find lying around, I wouldn't go out and buy them as that costs money which I don't have. Not that there wasn't enough of them; my dad's an electronic engineer and there's loads of junk lying around which I could tare up for parts, It's just finding them that was the problem. This lack of relays was to become the bane of the entire TIM project.

Anyway, back to the project. TIM 2 was much the same as TIM 1, It just had another bit of memory. TIM 3 on the other hand was a 3 bit full adder, with control panel. This was where things began to get interesting:

I always remember the first time I switched it on. Not a particularly interesting event in theory, but just seeing that power light turn on and the power supply fan starting up gave the same sort of excitement I get when one of my vehicles moves under it's own power for the first time. It felt like a real computer, even if it was only capable of the most basic additions.

TIM 4 was much the same as TIM 3, just now it was a 4 bit full adder.

TIM 5 was my first attempt at a real computer (of sorts). I basically took TIM 4 and added 3 registers (memory blocks) ,a clock and a whole bunch of control circuitry. It looked awesome (in a kind of Heath Robinson-y way ) as I practically soldered up the relays where they happened to be at the time, meaning there was wires everywhere, Look at the picture below and you'll see what I mean:

It could now do multiplication and incrementation as well as addition up to 31, however it was impossible to move anywhere as you had to move each relay individually. Here's a video of it incrementing up to 16. I eventually took it apart to make TIM 6, A more compact and simplified version of TIM 5 as shown below:

I've added the photo of the underside to just to give you an idea of the problems faced when trying to find bugs in the system. I never made any circuit diagrams for it since I made it up as i was going along, so I had to remember where all the wires went. There was one fault which caused it to go haywire occasionally which I never found until I finally disassembled it (one of the capacitors in the C register had become unsoldered for anyone who's interested).

Things started to get serious from now on. By this time I wanted a real working CPU capable of running programs rather than just the simple calculators I'd been constructing. I started the design of TIM 7 at the end of the summer holiday's, and only finished the plans for it around Christmas.

 It was originally going to be a very simple 4 bit adder, basically TIM 6 with punch tape instead of a control panel. Then i upgraded the design to a full 5 bit punch tape computer with all the standard ALU commands, as shown above. However I still didn't like the design all that much so completely redesigned it as a serial ALU, 8 bit punch tape computer which was much neater. Although TIM 7 was never built, i called the last one TIM 8, partly because TIM 7 had so many plans for it, and partly because it better suited an 8 bit computer.

TIM 8 by contrast only took a few weeks to design, this was mainly because I'd worked out all the hard while stuff designing TIM 7. The problem was I'd now found all the relays there were to be found, and that only came up to about 150. I've never seen an 8 bit relay computer built in less than 281 relays, and most of those were 4 pole relay, where as mine are nearly all single pole. So I had to take the simplification to extreme lengths. The advantage of running it straight off the tape reader is that you didn't require a program counter, or a finite state machine, or a clock, which saves a good hundred relays at least. Then there is the serial ALU, which although slower than a parallel one, allows greater flexibility and uses far less relays.

Another problem that had occurred was that I refuse to use a RAM chip for memory, because it's a RELAY computer, not a relay-with-several-million-transistors-thrown-in-because-I'm-to-lazy-to-do-it-properly computer. As such I'm only allowing technology which was around before the time of  ENIAC. Almost every other relay computer I've seen uses a RAM chip, which is only vaguely acceptable as they're using the Von Neumann Architecture (Computer data and program are stored in the same memory) , where as I'm using the Harvard Architecture (Computer data and program are stored in separate memories, in my case the program is stored on the punch tape). This means the amount of memory I have doesn't limit the length of program I can write, however it still limits the amount of data the computer can process. So I designed in five 8 bit registers. But that uses a huge amount of relays and only stores 5 bytes, which isn't really enough for anything exciting. I had also just redesigned my memory addressing system so that it allowed 16 bytes of memory to be stored. The problem was now that i had very few relays left, so making all of that out of relay latch memory wasn't possible. I therefore set about researching all the different kinds of computer RAM that had ever been developed to try find one which I could use, however they all were horribly complex and required large amounts of decoding hardware. Therefore I had to design my own. I decided to use capacitors as they were the only component I had which could store data easily. So I spent the next week or so trying loads of different designs out, looking at the problems encountered and redesigning accordingly. eventually I came up with the following design, which only uses one relay pole per byte:

This meant I could now use all of my addressable memory with out having to resort to modern IC's. It does mean I'm going to have to buy around 100 capacitors and 200 diodes, but they are much cheaper than relays, and it should look cool when done (I haven't built a full scale version yet, only a 2x2 bit prototype)

Here is the finished design of TIM 8. As of this post, I have built pretty much everything except the capacitor memory so far.

Below is a picture of the ALU panel with it's supporting hardware to give you an idea of what it looks like. The photo is  somewhat out of date as various things have changed, but for the most part it's the same. The ALU in the centre does all 8 logic processes for the computer, so is basically TIM's brain. It took a long time and a lot of effort to get it down to just 12 relays (The one on TIM 7 would have used 40), However it does require the parallel to serial converters and command selector etc. which brings the relay count up slightly, but it's still pretty decent. The 1 bit nature of the ALU does allow useful tricks such as variable bit width, so you can string 2 or more registers together if you need to process numbers greater than 8 bit. Alternatively, you could use half a register for one 4 bit number, and the other half for another to increase memory capacity.

This is a picture of the actual circuit diagram for the ALU, showing all 8 function outputs. It was drawn up in Circuit Wizard, which was where the design for this ALU was finalized:

Below is (quite a poor) photo of an almost completed T.I.M. on my desk. I had to build it on the sheets which eventually became the case, or it would have been very hard to make it fit:

This is one of an almost completed T.I.M. in his box. From the photo above to the photo below was about one school week. This rapid development was due to the fact I had heard there was to be a science fair that Friday, and decided it would be a good deadline in which to finish TIM, or it would never get done. Some very late nights later and TIM was pretty much done. I took it into school the next day almost entirely untested, and after an hour trying to fix a fault (turned out the tape reader head had moved slightly), it ran an increment program for 2 hours continuously without fault (We eventually turned it off, the clicking was driving us insane). Admittedly the increment command was all he could do at that point, seeing as i had wired up the A and B inputs to the ALU the wrong way round, but i only discovered that the next day. To see a video of it running, click here.

The BIOS of T.I.M. is called B.L.T. (Basic Language of Tim/ My favourite sandwich). It is designed in the way to use the smallest possible amount of relays for decoding, and yet keep it Turing complete. It is a little bit of a Turing Tarpit language, but it's still surprisingly flexible. As the ALU is serial, each bit has to be addressed to pass through it. This means you have 8 commands for any ALU instruction. This may seem excessive, but it means that, for example, if i wanted to store two 4 bit numbers separately, i could store them in the same register or alternatively, as the carry has it's own register and is fed back in to each computation, I can link registers together to hold larger numbers. So when T.I.M.'s memory is finished, it should be able to hold numbers equivalent to around 3.4x10^38... which is pretty big (128 bits). However, for addressing commands you only need a 3 bit address, and the address in is 4 bits across. Therefore I've used this to allow me to address to working registers (B and C) which means I can work on more numbers where they are instead of having to move them into a different memory location (which there are very few of). This means that if I was counting using a flag (for example) then I could increment my flag in one register, and do all the workings in the other.

I have also left out commands I can do in other ways using more commands. E.G. there is no INC command as this would take an extra command and be hard to do on a serial ALU. However to increment you 'set carry to 1', 'Disable input A' and then ADD A and B to increment B. Again this means the computer is more flexible as you can increment single digits or huge numbers with much less relays. Despite being so simple I'm always having to update the language where I've hardwired it into T.I.M. wrong or written it down wrong. For example none of my more complex programs worked previously, and this was because I'd written down 'Disable input A' as 'Disable input B/C' which meant I was always trying to increment A which won't work.

Another thing I did to save relays was doing away with the clock and finite state machine that times all the commands in the computer. If you look at the tape, the commands are much longer than the execute line to let the relays settle in the right position before a command is processed. To do any other timings I have used capacitors. This reduces the speed I can run the computer, but speed wasn't what T.I.M. was designed for. 

Here is a picture explaining the language (commands in brackets have not yet been added):

Programs are written as bytes in note pad, then compiled into a .bmp file using a little program I wrote. These are then printed out on to tape using a recipt printer. This method is substantially easier than the original punched tape programs that you got on old computers, as I would have had to punch each program out by hand, and most mistakes would require a complete re-punch of the program.

I have written a program that almost amounts to an operating system now; basically what it is is a basic calculator program, where you input a number, give it an operation such as + ,- ,X ,or /, then input another number. The program then runs, then places the result in the first input register, loops back, and allow's you to give another function and another number. It's sort of working, but since the tape is around 4 meters long it took an age to print (only allows me to print in 30cm sections for some reason) I haven't printed out the debuged vertion, as i want to find as many bugs as possible first.

Below is a little program that just increments the B register. This is the form in which it would be printed out, with annotations added. This program can be seen running in this video.

This page is no longer updated, but if you want more recent information on TIM please visit the official website at http://www.northdownfarm.co.uk/rory/tim/
 (The information is mostly the same, but it includes recent updates which this page does not.)

The Banter Buggy!

This, in my opinion, is my best machine to date. Not necessarily the best engineered or the best designed, but by far the most fun! Built in 2009 when I was in year 10 this was my second vehicle to be built after the trike, which was a little special...


years and years ago I got lent a book by one of my Mums friends called ' Build Your Own Off-road Buggy for as little as £100 '. I used to read through it regularly and kept asking my dad to build one, but unsurprisingly he wasn't too keen. Then, in early 2009 a friend of mine offered me 3 motorbike engines for £50 that he recently had taken out of his own buggy (built from the same book as it turned out). It was a good bargain so I snapped them up with little idea of what I was going to do with them. Then I remembered the book. So that Easter holiday I got started. It was a very steep learning curve as I had never done anything particularly complex before, and had only just learnt to weld. I got a whole load of steel off cuts from the local engineering shop, and a lot of box section from an old horse box we'd recently destroyed, and started work. The first bit wasn't too bad, just welding up the basic frame shape from the measurements in the book, but then it started to get more complex with the steering.

We'd recently cannibalised our old lawnmower tractor for parts, so I nicked the wheels and front steering mechanism off that. However all of the steering rods on that thing were worn to nothing. Fortunately, my cousin worked down the road at Ariel Motor Company, and so was able to get a couple of old track rod ends from them, which made it a little easier. I then had to work out all the geometry's in such a way that didn't get in the way of your feet, or lower ground clearance. The mechanism I ended up with was far from neat, but it seems to have done the job fairly well so far.


Next (I think, was a while ago now!) was the seat, which was also far from neat. I planned on using the springs from the lawnmower seat to counter the lack of suspension, but these had to be attached to the glass fibre Atom seat that was again supplied by my cousin. It was out of a crashed Atom so half of it was pretty broken, but the other half was almost complete. So the broken side was cut off and this was then mounted on top of the lawnmower seat with a single bolt, and then expanding foam was used to hold it in place. It is an almost painfully bad piece of engineering, but it does the job.


Next was the engine, which was easy enough as the one I wanted to use already came in a rear subframe from my friends old buggy. There was a lot of debate as to whether I should have suspension or not, but I finally agreed to put it in (which was a good move as it puts less shock loading on the rest of the vehicle). So two bolts were used to hold it in place on a pivoting join.


The back axle was one of the hardest (and most expensive) part of the project. it came from one of my dad's friends, and was the back axle out of his camper van (or something like that). Unfortunately it was a different size to both the bearings I had and the back wheels, so I had it sent to a professional mechanic who had it machined to size at either end. Then I had to fit the rear cog, which had been bought from a local bike shop, which was mounted on part of an old bearing. This actually worked pretty well as it turns out. This then all had to be mounted with the bearings onto the rear subframe and connected to the engine.

All that was left to do was the rear suspention. All this consisted off was a pair of rear mororbike shocks attached between the main frame and a plate above the rear bearings. Simple, but reasnobly effective.

This all took 5 weeks from start to first drive (although it was still hardly finnished) and had cost £125. I havent a clue how your supposed to do it for £100 as most of the parts i had already or got free!

A bit of driving

 Despite the poor build quality, this thing is awesome to drive! The sheer volume of the engine is impressive enough (i actually got complaints from the neighbours about the noise, and they live about half a mile away!)
It has big problems with understeer as you would expect what with having no diferential, although it is now fitted with brakes. which is a job I've been putting off for the last 3 years. The handleing leaves something to be desired, but for sheer driving awesomeness it is brilliant!