November update

Blimey.  The last time I posted anything here was way back in June, how time flies when you're having fun :-)

A quick summary of projects during that time follows:  you at the back, please try to stay awake

First up is picture scanning.  I bought a cheap negative/slide scanner from Maplins a couple of years ago with the plan of scanning all our old photographs - the idea was to keep a permanent digital record in case the originals went missing, and also to try to correct some of the colour drift that has affected some of the early stuff.  Good plan, but a couple of evenings of feeding the scanner with negatives soon convinced me that it was a job left for another day.  A few years on and the 'another day' has arrived, so I set to work a couple of months ago.

We have three boxes crammed with assorted photos and negatives spanning the time since 1970, and another with a mixture of older pictures and other peoples photos - I finished scanning the contents of the first three boxes in around 6 weeks, the remaining one may never get done (especially as there are lots of photos where neither of us know who the people in the picture are). Windows reports that there is now around 8000 pictures in my library, with a total mass of around 20Gb, and the whole lot has been backed up on to three DVDs in case the server bursts into flames.

Apart from the tedium, the cheapo scanner worked pretty well and I could probably flog it on Ebay for about what I paid for it.  There's also been lots of 'hey, do you remember this...?' moments, and a lot of fun to be had looking back, especially from when the kids were little.  Not only that, but of course we can now run a continuous slide show on the TV of all these old pictures to bore visitors to death - a bit like getting out the photo albums, but infinitely worse.

I've also been turning some perfectly good wood into sawdust again, with the completion of a small wall-hung jewellery cabinet, and a new jewellery box.

The cabinet is made in cherry with some trimmings made in imbuia, and it turned out rather well - see picture. It has my usual brass hinges with ball bearings to make them work smoothly, and the latching mechnism is hidden inside the central divider and operated by means of a pair of concealed levers underneath the front edge of the base.  It's presently sat in the Dansel Gallery near Weymouth with a hefty price tag and waiting for 'the woman who has everything, and needs somewhere to keep it' to arrive with her credit card.

There's a bit more of a story to the box.  A few months ago, a chap in the village who I know slightly turned up at the door and asked if I wanted some wood - he has seen me floundering about in a cloud of sawdust over the years when he's been out walking his dog.  It turns out he's working on a new school being built in Bridgwater, and spotted some of the chippies about to throw some offcuts into the skip, so he stuffed them into the boot of his car and brought them to me as a little present - very nice of him.  The wood is ash, reasonably easy to work, and with a dramatic stripey grain, so to try it out I built this box.


This one has a couple of other features that are new to my boxes.  The corner joints are made using the jointing jig that I've had for a few years and never had the time or patience to set it up properly, turns out it works pretty well.  I havn't made a box with drawers in before, and this works pretty well too.  Finally, I designed a new style of hinges for this box which make the back edge of the lid partly disappear down the back of the box - no real useful reason for this, it just looks a bit different.

Oh, and the chinese character on the front means 'harmony'.  I've got enough wood left to build another box as well, and it never cost a red cent.  Excellent.

This neatly leads on to a couple of other things that I've been looking at recently.  The first concerns my radial arm saw, a fantastic bit of kit but when ripping timber the blade spins towards the operator and in an upward direction, completely opposite to a table saw where the blade cuts downwards.  With larger chunks of tree this is not a problem, as the weight of the wood holds it down, but these boxes use lots of very thin and/or narrow pieces of wood - its real easy for the blade to catch on a loose edge and pick up the piece and hurl it straight into your face, and even with the lexan face mask it can be a scary experience.  And of course, it completely boogers the piece of wood that I'm trying to cut, which is invariable the last piece of that particular colour or grain...

This has always been an issue with these saws, and I've noticed it much more since having my blades re-sharpened a few months ago - you would think that the sharper teeth would make things better, but I noticed that the sharpener has ground the teeth to a more aggressive angle than before and while they cut a treat, I suspect this is not helping the 'throw the wood at the old man's face' game.  Time to try to resolve this, so:-

I made a featherboard - a chunk of wood with multiple fingers that lightly presses against the piece being cut and holds it down on to the table.  The first attempt worked really well, but had a limited range of adjustment for different thicknesses.  It also had the fingers too far away from the blade, so the last couple of inches of wood was not held down by the fingers, and whoosh; whizz; bang, bits of wood banging off the face mask again.

The new version has a pair of slidey fingery bits that allows the fingers to go right up to the centreline of the blade, as well as a better way of adjusting the whole contraption for wood thickness.  Tested it on a few sample pieces and it seems to work fine, but we'll only know for sure when I feed it with that last-piece-on-earth-like-this irreplaceable timber.  I'll try it out for real and then finish off the metal bits with a coat of paint.

I also thought that it would be a neat idea to have a motorised arrangement that slowly draws the wood through the blade.  Apart from the smoothness of the cut that this would deliver, it also has the advantage of keeping fingers well away from the sharp bits.  Into the junk box, and out with a rear window wiper motor from an Austin Metro (remember those?) and some assorted metal chunks and off to the lathe and milling machine.  On to Ebay for some rubber tyred wheels (£5 for a set of 4, thank you very much), bolt the whole caboodle together and hey - it works!. 

A little too fast though, ok for thin plywood but anything thicker causes the blade to struggle.  No problem, I've built and bought DC motor speed controls before so I can do it again, but lets just check Ebay first....  Ah.  Parts to build a DIY controller is around £15, ready built one is around £25 (but will go backwards as well), but my pals in China will sell me a completely built unit for £4 with free postage.  Ten days later, the unit arrives and works perfectly.


A few modifications later and the unit is finished.  Like the featherboard, it still needs to be tested in anger, but it seems to work fine, although the rubber wheel is a bit slippy especially when covered in sawdust. 


And then the most recent workshop thingy.  I have a little Axminster milling machine, an excellent bit of kit that I really enjoy using.  It has a couple of things that are less than ideal, and one of these is the fine adjustment hand wheel on the spindle feed - the wheel is too close to the machine casing to let my fat fingers get in and reset the dial to zero.  Making a longer spindle for the handwheel would be a bit of a hassle - it has a tiny keyway machined in it for one thing - so I made a new handwheel out of some thick pespex that I happened to have lying around, and machined the old handwheel down so that I could screw the new one on to the face of the old one.  Works a treat.

In the process of doing this I realised why I've had a few problems with using this handwheel for setting the depth of cut.  All of the other handwheels on the machine are graduated with 80 small divisions that give a feed of 2mm per turn of the wheel - slightly unusual as most machines are 1mm per turn, but OK once you get the hang of it.  The spindle handle must have been designed by a chimp on crack, because it has 36 divisions and a feed of 1.8mm per turn.  I've tried to think of a plausible reason for this, and I can't - they must have just had some spare 1.8mm screwthreads kicking about in the junk box and decided to use them up. Ah well.

I also realised that one of the great pleasures for me in doing these things is having the ability and the tools to make the various parts from materials that are mostly junk.  The nut that holds the handwheel on is made from an offcut of aluminium bar, and both that and the handwheel itself were made on the milling machine using the rotary table.  Apart from the improvements in usability of the finished project, there is enormous pleasure to be had from just making a nut.
I will try to update this blog around once a month from now on, its not as if I'm sitting around with nothing to write about :-)  Projects for the near future are possibly a radio controlled beach crawler (inspired by the RNLI launchmobile seen at Burnham a few weeks ago), and a reworking of that old nitro model plane engine that I've had for years and can't decide what to put it in - it might end up with a flywheel and clutch, and an on-board electric starter, and it may finish up in the beach crawler, who knows.  I'm also thinking about ways to flog my jewel boxes - galleries are either only interested in paintings, or want 90% commission, yes, you read it right, 90% - and it seems to me that there must be loads of people like me who can make good original stuff but can't get it to market: time for a 'makers community' effort perhaps.

Toot toot

Miscellaneous ramblings

Been a while since I posted an update, so this is a quick summary of some of the things I've been doing lately.

First up is the new dragon, made from strips of thin copper foil cut with scissors and my wife's pinking shears (shhh! don't tell her!), with each piece glued and/or soldered to a framework made from copper wire coiled round a series of formers to give the different parts of the body.  I don't have any pictures of the work in progress, but this is the finished article in the final location in the garden:-

The present shiny tarty colour will darken to a purple-ish brown with exposure to the weather.  For those interested, his name is Brian, and his job is to scare away the bad spirits who may happen to wander past.

The second interesting thing I've been doing is trying the latest version of Ubuntu Linux. Over several years, I have made strenuous efforts to get linux working sensibly as an alternative to Windows, and every past attempt has been a disaster. 

I really would like to see any version of linux succeed as a viable alternative for mainstream users - Windows just keeps getting bigger and more complex, and Microscrote's obsession with security is becoming ridiculous - so every six months or so I download a couple of different versions and try it out. The one I have grown to like best, mainly for its user interface, has been Ubuntu: however, all of the past versions just would not work reliably with my network - even though other flavours of linux and good ol' Windows all run fine.  Also, I play a lot of music, and all of the media players that I tried would play anything except MP3s, which is what all of my music is.

Then there's my other long-standing gripe against linux, which is that if you want to plug anything in to the box, or find out why the network won't run, or install some software that you got from anywhere else, you have no option but to jump into the command line.  Back when I started in computers in around 1978, this was normal, but the world has moved on a fair bit and we are all used to just plugging in a web cam, or a USB stick, or some such, and it just works. The linux command line is even less user-friendly than the old MS-DOS was back in the day, and questions to various linux forums are generally greeted with incomprehensible comments like " .. you can just use Woodchip using the /fek switch.." while further enquiry seems to invariably result in  torrents of abuse.  Not helpful

My latest attempt uses Ubuntu v12.04.  The news is that: a) it installed and worked without a hitch on my four year old Clereon based Shuttle machine; b) it found and used my network with zero drama, and c) best of all, it seems happy to find and use all the devices I've plugged into it without complaint, and with no reason for me to tackle the command line.  After playing with it for a while, I decided that I didn't much like the latest Ubuntu user interface, so I downloaded KDE instead (again, an easy install), which is much better IMO.  Strangely, KDE doesn't come with a printer module, so you have to find one on the web and download and install it manually - a small bit of command line here to do this, but just type in the words and hit return and it loads and works.  (I later found that there is a version of Ubuntu called Kubuntu, which gives you the same end result as I have, although the downloaded version that I have won't boot the machine from the CD, it just sits there looking at you pretty much for ever)

I'm still playing with it, but I have yet to find anything that doesn't work, or requires a Master's degree in guru to be able to make it work.  At last, this may be the way for me to ditch the Windows bloat and security paranoia - I'm going to set it up on the music machine in the garage (yes, in the garage, I like to entertain the neighbours with a bit of rock an' roll when I'm working out there) and try it for a while on that.

Next item concerns weight.  As a result of my regular fitness training, the old body has slimmed down a bit and I have lost a couple of kilos, generally a Good Thing.  However, this means that I am now under the minimum weight for my kart racing class, and I need to add some more weight to the kart.  I had some scrap bits of lead sheet in the garage, and decided to try melting them into slabs.

My initial plan was to make a mould from damp sand - the same way that iron has been cast for a few centuries - but reading up on the web suggested that this would be a Bad Idea.  Commercial sand is a complex mixture of sand, clay and other stuff designed to both give a smooth surface to the casting and allow the moisture to escape, and  I quickly realised that my original idea of making a dent in some wet sand and chucking molten lead into it was a non-starter - lead at around 600 degrees C turns water instantly to steam, and if it can't escape it causes bubbles in the lead which burst, splattering molten lead all around.

What I did find out was that you can cast lead into a wooden mould.  The wood chars from the heat, but the metal is not hot enough to set it on fire, and the charring on the surface prevents the wood from burning any further.  Also, the lead contracts slightly as it cools, making it easy to get it out of the mould.  Off to the workshop to make a flat open-topped box out of some scrap MDF, which has a smooth surface finish and is easy to work. The dimensions were calculated to give a slab that would weigh roughly 1.5 kilos and was around 15cm long

I then weighed around a kilo of the scrap lead and cut it into strips with some old scissors and put them into an old saucepan over a camping stove. First put on my face mask (lead fumes are generally bad for you), lit the gas blowlamp and played that into the pan, and the lead melted really quickly, developing a thick scum on the surface which I dragged to one side and scooped out with a flat wooden paddle (a scrap bit of apple wood).  I poured the lead carefully into the mould and left it to cool for around ten minutes, then turned the mould over and smacked it hard on the floor a couple of times until the slab fell out.

It worked so well that I did the same thing again, then drilled holes in both slabs to use for mounting on the kart.

The slabs are smooth and professional looking, and the only damage to the mould is the surface charring and a lingering smell of burning fish.  The saucepan won't be much good for making porridge any more, but that's a small price to pay :-)





The last item is a bit of irony.  I have serviced my own cars ever since god was a boy, and always get a hard time from dealers when coming to sell them because the routine servicing - despite being complete - has not been carried out by a main dealer.  I changed my Civic Type R last autumn and replaced it with the same model with a lot less miles and a couple of years younger, and the new car came with a splendid full Honda dealer service history.

I did a service on the car last week, and discovered that at the last service Mr Honda had replaced the oil filter with some cheap and nasty aftermarket crap; the washer on the sump plug had been used several times, and the air filter looked as if it was the original one fitted in the factory back in 2005.  Based on this, there's no reason to expect that the oil was anything other than cheap cooking oil, or that any of the other service checks had been done properly either.

I'll carry on doing my own servicing, thanks.  At least I know the work has all been done, the correct parts and decent quality fluids have been used, and - most important - I won't have to pay heavily for the pleasure.

Arduino and Home Easy problems

Whew!  Just spent the best part of a day and a half working out what was wrong with the Arduino-based  magic box that controls the lights and a bunch of other stuff in the house.

Its been working fine for a few months, but around a week ago it decided to either not turn stuff on, or turn it on and then forget to turn it off. Pretty poor show.  Most of the HE kit has its own switch in the room where it lives, but some can only be operated with the remote control which can be a bit of a pain.

I scribbled a quick bit of code for the arduino to flash the various pins that operate the remote control, and discovered that the one that turns things 'on' wasn't working.  The 'off' and 'unit' worked fine.

I started by checking the remote control which is hard wired to the arduino box with a length of CAT5 cable and a standard RJ45 plug - this lets me move the remote around the house using the CAT5 in the walls to get the best wireless signal for the HE stuff.  Opened up the remote, no obvious signs of wires hanging off.  Checked the voltage on the three pins that operate the control - two at 0v, but one at around 0.6v, no surprise that the odd one out was the 'on' connection.

I assumed that the remote was the most likely place for a fault, because I had to butcher it about a bit, so I replaced the opto-isolator for the 'on' button.  No joy.

So the control box had to come out of the rack, and be opened up.  I plugged the remote directly into the box and checked the resistance in the connections, all near enough zero as you'd expect.  From this I assumed that the problem probably lay in the arduino - a dead I/O pin probably - so I switched the connections on the processor and changed the code to use a different one.  Put it all back together, but no luck.

Out with the control box again, check everything again.  Still no joy.  While I had the bonnet up, I changed the 12 volt power supply that also lives in there - it was a switched mode supply previously and it had a lot of problems delivering 12 volts - sometimes it would drift up to 15 or 16 v, then other times it would drop to 4 or 5, playing havoc with my mailbox detector.  I binned the old PSU and replaced with a conventional transformer/rectifier/capacitor arrangement and bingo, 12.5 volts bang on the nose.

After replacing the original I/O pin and checking the software yet again, I concluded that either there was some strange fault in the arduino, or there was a cable fault.  Searched online for reasons for the arduino to play up without success, so I went back to the cabling.  This time I checked every single connection of every cable between the processor and the remote, including individual patch leads and the cables in the wall.

All still tested good, but then a fluke - I happened to move the end of one of the patch leads while the meter was still connected and lost the connection:  moved it a bit more and it came back.

Much happiness in the camp, the problem was an intermittent break in the patch lead (which is why it worked ok one day and not the next).  Cut up and binned the offending lead, then celebrated by flashing all the lights in the house on and off a few times without leaving my chair, which didn't go down too well with 'er indoors.

Grandfather clock re-build

Well, its taken a solid two months of hard work, but the new case for the family grandfather clock is now completed. The pictures above and on the left shows the new one, and the one below shows how the old one looked before the re-furb.

The clock was originally built in around 1820, so he's nearly 200 years old, been in my family for six generations, and during that time he's taken a bit of a beating.

What's wrong with the old one?

Well, although the clock itself ran OK, the case was in a terrible state. Originally built with a cheap pine framework and veneered with mahogany, it must have looked fine to begin with but the years caused the veneer to split in lots of places and bits have fallen off, to be replaced by various of my ancestors with more enthusiasm than skill.

At some time the whole clock had obviously fallen face down, hopefully not when someone was walking past it, because the front of the case and the lower door had clear evidence of major full-frontal impact. The key for the lock for the lower door must have also been lost at some point, so a chunk had been gouged out of the case so that the door could be opened.

My mother told me that at one time it lived in a house with low ceilings, and it was too tall, so the owners simply dug a hole in the dirt floor and placed it in that. OK for a while, but the bottom foot of the case rotted away, and ended up being sawn off. I guess that was good in one way, as it meant that it would then fit under the low ceilings, but it meant that the weights had a shorter distance to fall and it would only run for around six days instead of the usual eight. My dad solved the problem by making a box out of some rough timber that he had lying around and balancing the clock slightly precariously on top of it.

Apart from the woodworm, last but not least, I found no less than 33 holes in the back panel of the case. These clocks are supposed to be screwed to the wall, but even allowing for numerous house moves I am mystified why it needed that many.

The clockwork bits

The mechanism ran fine, but it was absolutely filthy - my dad used to oil it with a mixture of engine oil and paraffin, liberally applied with an old paintbrush (that noise you can hear is the clock makers of this world screaming in pain), so the whole thing was covered in sticky black goo mixed with fluff and old spider's webs. I thought it best not to dismantle the old chap, so instead I carefully cleaned the whole thing with petrol and carburettor cleaner, and then lightly oiled all the shaft pivots with tiny drops of very thin oil (the same stuff used for my high speed engraver). One of the cords for the weights was a bit tattered, so I replaced it to avoid a 10 Kg lump of lead dropping through the floor in the middle of the night.

Time for the new case

I started with carefully measuring the old one, and working out how to make replicas of all the mouldings that had been used.

Next, with the assistance of my friend and grandson, Owen, we built a timber framework to provide a solid structure for the whole case. Not only did this give it strength, it also made sure that the structure was square and level. These clocks are picky about being level, so the base had adjustable feet concealed within so that we don't need to mess about packing the mechanism to make it level.

The frame was then sheathed in MDF. This may seem a bit non-traditional to the purists, but remember that these clocks were originally made from cheap materials and veneered with more expensive wood, and if the makers were building one today I expect they would have also used MDF. This material also has the advantage of being strong and dimensionally stable, so it should last a fair while.

The veneer used on the original was mahogany, but these days rocking horse poop is easier to get, so I used sapele for the various mouldings and other solid wood parts. Once the mouldings were made, using my trusty router and various modified cutters, the whole case was veneered and the mouldings fixed in place.

I re-used as much of the old case as possible - the brass finials on the front door; the lower door hinges; the lock for the lower door; the glass in the front door; the wooden pillars on the front, and a few other bits. The door which covers the clock face has concealed ball bearings to swing on in place of the rusty nails used on the originals, and a proper catch to hold it closed (the original was just jammed against the frame to keep it shut).

The whole thing was then given a double coating of mahogany stain to even out the colours of the various woods, and to make it look a little older, followed by several coats of danish oil and finished off with beeswax.

Placed it against the wall, used the hidden feet to bring the whole thing vertical in both planes, and screwed it to the wall. Placed the mechanism on the original seatboard, and hung the weights and pendulum, then adjusted the swing of the pendulum to give an even tick - if its uneven, the clock will only run for five minutes and then gives up.

I can hear it ticking reassuringly as I write this, and although its been a monumental job I'm very pleased with the result. Time to pack the tools away, clear all the dust out of the workshop, and start on the next project...

Arduino home control part 4

Made some good progress with this over the past couple of days.

I experimented with getting the Arduino to read the time and date from the PC, but its too 'ard, mum. Instead, I bought a real time clock module (RTC) from a nice man in China for around four quid and hard wired it to the arduino. Searched the 'web for a simple bit of code to set the clock up, and eventually found this -

http://www.glacialwanderer.com/hobbyrobotics/?cat=4&paged=2

its dead simple, even I can understand it. All the other stuff I found was far too complex for my purposes (and I'm too stoopid to understand most of it).

Then comes the programming. I have to say that software is not one of my core skills, but perseverance wins out over my lack of talent. It ain't pretty, but it works.

What it does is the following:-

- reads the time and date off the RTC

- calculates today's dusk and dawn times based on the current month - it will only update these times once a month, but that's near enough for my purposes

- sets up a series of times to switch things on and off. These are either 'real' times, eg 23:30, or a time which is based on the calculated dusk or dawn times. Some lights come on at dusk, others have a delay to turn them on later.

- reports current state back to the server - at the moment, this is a simple VB application which just displays the data on screen, but in future this will find its way into a web page which will run on an Ameo either in the kitchen or by the front door.

- information displayed at the moment is: internal and external temperatures; lock status of external doors; whether there is mail in the box, and which home easy devices are on and off.

The next stage of the plan is to make the arduino control the central ventilation system based on internal and external temperatures (turn it off on a cold night, etc.), and to run a pair of water leak detectors in the kitchen and utility room - these will be automatic, and shut off the water using solenoid operated valves if there is a leak, but the arduino will tell the PC that a leak has occurred and the PC will wind up a bloody great siren. Looking forward to that part :-)

Some interesting times during the development process, where the software kept running in an endless loop, switching things on and off at random, so much so that Mrs G had to retire to the kitchen as the flashing lights in the rest of the house were giving her a headache. There was also a report that the pulsing of the lights on the deck had encouraged a Sea King helicopter to try a landing, but I don't believe that.

It seems to be working OK, its so far managed to turn all the lights on at the right times without any major malfunction, although I will have to fine tune the post-dusk delay a bit on a couple of the lights. Some alarm from MRs G along the lines of 'what happens if it goes wrong and you're not here? Do I have to sit in the dark?', but I still have a spare remote which can be used to turn things on and off, and all of the important bits have their own switches in the room where they live.

More stuff to come, watch this space.

Arduino home control - Part3

Success at last!

Well, I've lost count of the days I have wasted trying to get this to work using my Chinese radio modules and code off the 'net. Some of the code that I downloaded had errors, but I suspect that I could have made it all work if I'd had a bigger brain and an understanding of C++. Can't be arsed to learn C++, so I looked around for an alternative.

The best one I stumbled over is a chap on Instructables who had hacked one of the Home Easy HE100 remote controls so that pulses on the arduino switched opto-couplers to operate the on

and off buttons - brilliant! Original source for this is here -

http://www.instructables.com/id/Arduino-Home-automation/#step1

I searched for a HE100 to butcher, but could only find them bundled with more HE modules that I didn't need. However, my son Sean had a HE200 spare, so I scrounged it and set to work with my little hatchet.

The HE100 is a simple remote with four pairs of buttons to switch a maximum of four devices on and off. The HE200 is a more complex remote with a small display panel and the ability to run timers and to control up to 16 devices on the same house code. However, to select a specific device, you need to press the 'unit' button until the number of the device you want comes up on the display, then hit the 'on' or 'off' button. Like most of these devices, the buttons simply momentarily short out contacts and it is easy to mimic a series of button presses using the arduino.

First step was to dismember the beast and see what's inside. There are two screws in the battery compartment which hold the two halves of the case together, you will need to gently prise the two halves apart with a thin screwdriver to release the clips. The PCB is held in with another two screws and clips at the top end which force the contacts on the PCB against the rubbery terminals for the display screen. With a bit of care, the board pops out leaving the display and the switches in the case half. PCB looks like this from the switch side:-

...and like this from the back. The wires are not part of the remote, I soldered them on to do some testing.

The PCB is double sided, and luckily the connections needed for the switches also appear on the back side of the board. Even better, the copper tracks have small bare copper pads, probably for testing, and ideal for soldering fine wires. Best of all, these pads are numbered, so pad 65 on the switch side appears on the back and labelled 65 as well.

This remote has two batteries, there is a 3 volt button cell which runs the timers and the switches etc, and a beefier 12 volt cell to power the radio transmitter - it will probably work off a lower voltage, but 12v gives maximum range. My arduino is in a box which also contains a 12 volt power supply, so I left the battery out and connected the terminals to the CAT5 fly lead which also feeds signals back to the arduino. I left the button cell in place, they last a long time and I didn't want to mess about dropping the 12 v down to 3.

You can't just connect the arduino outputs to the circuit board. I didn't try it but I imagine a small puff of acrid smoke and another visit to Mr Credit Card to replace the burnt bits. Instead, I used opto-isolators - these are small chips that connect to the arduino pins on one side and to the HE200 switch terminals on the other, a high signal from the arduino causes the terminals on the other side to be connected.

I wanted a neat job, so I fitted the opto-isolators inside the case of the remote - as you can see in the picture below, I just glued the back of the chips to the PCB with a dab of superglue, and then soldered the wires from the CAT5 fly lead to the various solder pads and the legs of the isolators. With some careful bending and trimming of the isolator legs, I managed to get the whole setup low enough to fit inside the case.

I screwed the case back together again, inserted the button cell and - amazing! - the display fired up fine, and all the buttons worked. Connected up the fly lead to the arduino box, no blinding flash or puffs of smoke, always a good sign. Finally I set a HE receiver into 'learn' mode and pressed the 'on' button on the remote, and the receiver did the usual reassuring clicking and turned on and off when I pressed the buttons on the remote.

I wrote a simple bit of code for the arduino to test it - basically, I 'learned' the receiver to be unit 10, then wrote some code that pulsed the unit button 10 times, then pulsed the 'on' button once. It worked! There is no way for the remote to tell the arduino which unit it has stopped at, so I set the remote to unit 1 using the buttons, then amended the code so that it counts up to unit 10, does the switching, then counts back to unit 1 again. This will be fine as long as the unit button doesn't get pressed accidentally, I need to find out how to prevent this - maybe a dollop of Mr Super Glue in the button

Too pleased with myself to do any more tonight, but a good day's work. I need to modify the code so that it reads the time and date from the PC when the code first starts up (the arduino has no internal real time clock) and I can then either write code that calculates when its dark or use a light sensor to tell it. Turning lights on and off at set times is one of the key functions of this box, and once this is working reliably I will add the rest of the stuff - switching the vent system with temperature; monitoring door lock status; watching the mailbox and the water leak sensors. Then tomorrow, we take over the world!

Thanks to all those who have helped with supplying me with information of this project, a pity that I am too stoopid to make the radio modules work - however, this hacked remote will do what I want it to with minimal effort, so I can spend more time doing other stuff :-)

Mail waiting alert - update

A quick follow up on this following an investigation today into the power issues and some mods to cure some erratic operation over the past week.

First, the power supply was only dishing up around 9 volts which fell to less than 8 by the time it reached the gatepost. Luckily, its a PSU salvaged from an industrial-strength network switch, and the output is adjustable with a small potentiometer, so a quick twiddle and voila! around 12 volts at the gate.

Still erratic in operation though :-( A bit more research later and I think I've found the cause. The Vellemann kit is designed to run off a couple of 9 volt PP3 batteries, one each for the transmitter and the receiver, and their web site/forum says that its OK to run the whole shebang at 12 volts. However, I discovered that if I drop the power to 9 volts on the receiver and leave it at 12 on the transmitter the instability goes away - no idea why, and finding out is well beyond my electronics knowledge. I've added a simple 9 volt regulator in the box and all seems well.

Just as well, as Mrs Grizzly was getting increasingly ticked off by trekking out to the gate to find the box was empty - a bit ironic, as when I announced I was going to make this gadget I got the standard roll of eyes and 'what on earth for??....' :-)

Mail waiting alert

My mailbox is on the gatepost, and the erratic timing of mail deliveries round these parts means sometimes several trips per day out in the pouring rain to check if there is mail. I built a detector from a Velleman kit to let me know when there was mail in the box, and this is how I did it.

The Velleman kit is an MK120 IR light barrier kit, available from Maplins in the UK and a number of other suppliers for around £10. It creates an infra red barrier, and sets off a squawker when the beam is broken. The kit needs to be soldered together, and for my purposes I needed to modify the kit slightly, as follows:-

1. make the infra red LEDs separate from the circuit boards

2. remove the squawker and replace it with a bright LED

3. also make the device send a signal to my Arduino based monitoring system.

First, the easy part - mounting the circuit boards in a weather-proof box. I had a spare aluminium box left over from another project and simply marked out the position of the holes and drilled and tapped them, then mounted the circuit boards on small plastic spacers to ensure that they couldn't touch the box.

The wires all come through the bottom of the box, and I put rubber grommets in the holes to seal them from the weather and stop the cables from chafing against the edges of the holes. Once all the holes were drilled, I sprayed the box and lid with satin black from an aerosol can.

In the picture, the green and yellow cables are the feeds to the IR LEDs, the other loose wires feed the LEDs that are fitted into the lid of the box.

To mount the LEDs remotely from the circuit boards I used a short length of coaxial cable (from and old audio lead) for each one. I tried CAT5 twisted pair cables and they were very prone to false alarms, I think because they picked up noise from each other and from the power cable - the coax solved that problem.

The LED end of these cables is exposed to the weather, so I encapsulated the end of the cables and the soldered connections in epoxy resin.

The 'mail waiting' alert is a bright orange LED that is fixed into the lid of the box (drill a 5mm diameter hole and secure the LED in it with a dab of super glue) with a short length of flexible cable to connect it to the squawker connection on the PCB. Note that a 220 ohm resistor is used to limit the current to the LED, and the LED was selected to give a bright light from around 5 volts - the box sits in direct sunlight most of the day, and a standard brightness LED can't be seen easily. Keep the unused squawker in your spares box for another day :-)

To seal the box from the weather I ran a small bead of silicone sealer around the edge of the lid and screwed it down tight. The sealer gets squeezed out of the joint, so the excess needs to be cleaned off before it dries (scrape most of it off with a screwdriver, then scrub it down with a wet cloth)

I feed the device with power from a central 12 volt power supply fitted in my central Arduino box, and a second (green) LED fitted into the lid of the box shows that power is present. This power supply also feeds the water leak detectors in the house (more on these later). The Velleman kit is designed to run on a 9 volt PP3 battery, but will happily work on 12 volts. When I built the house, I had a duct laid under the driveway to the gatepost and laid both CAT5 cabling and a mains power cable in it, so power is fed to the mailbox via the CAT5. This is not ideal, as the long cable run means that the voltage drops considerably and the 12v at the power supply ends up being around 8v at the mailbox. I plan to modify the power supply to fix that, although the mail detector works fine at this low voltage.

The same CAT5 cable also lets me feed the 'alert' signal back to the Arduino so that I can see on a screen in the house if there is mail in the box. I originally tried just using the raw output from the squawker connections, but could not get the Arduino to reliably read it, then realised that this was because the power supply is separate from the Arduino - instead, I use the squawker output to fire a small 5 volt relay which switches a connection between the Arduino's own 5v power rail and the data pin which monitors the mailbox status. Works a treat, although the low voltage sometimes makes the relay chatter instead of latching solidly.

My mailbox is a simple tin affair with an opening door at the front. I mounted the LEDs into some short plastic tubes glued to an aluminium plate across the back of the box, and angled in towards each other at around 45 degrees. A cheap make-up mirror is fixed to the inside of the door with double sided tape and reflects the beam when the door is closed. Drop a letter into the box and the beam is interrupted and the warning light comes on. Hooray!

It works fine, I just need to jack up the power a little so that the relay latches reliably. No more abortive trips in the rain out to the gatepost for me!