For any one that is interested I have tidied up the Eagle CAD files for my lighthouse circuit.  You can grab them here. They’re not perfect and are probably not suitable for sending to a PCB production company but they might give you a head start of making your own version.

They’re saved from Eagle CAD version 5.6.0 so you’ll need to go to thier site to get that if you don’t have the latest version.

UPDATED! : I’ve added a photo to the battery terminals, see below.

Recently my trusty Black & Decker Quatro drill/driver/saw/sander started misbehaving.  It’s been great for years and gotten us through a lot of DIY and IKEA modification.  But now it could barely manage to cut a 15cm length of 1cm chipboard.

Time for a bit of investigation.

I wondered if this was due to the charger not actually charging, it seemed ok, the LEDs lit up but on closer inspection there was a suspiciously low voltage at the charging pins and a distinct smell of “magic-smoke“.  So after a quick trip to the Ferreteria to buy some star drivers (why do they use these annoying types of screws) I could see that there was a very hot and rather charred looking transistor.  Looks like the charger was dead.  I thought about replacing the transistor, but unfortunately the part wasn’t marked and after a lot of Googling I couldn’t find any schematics.  It’s quite an old model and not available any more and plus there’s no knowing what else might have died, so I bit the bullet and bought a new fast charger.

Turns out that the battery pack had died too, it probably died first and took the charger with it in some sort of suicide pact.  A quick search of some online DIY forums shows that this is a common “feature”…  The new charger has a diagnostic “blinky light” that alerts you to a dead pack.

14.4v Black & Decker Firestorm NiCad pack.

The pack is a 14.4v NiCad (12 cells of 1.2v) which should have a terminal voltage of around 14.4v (not surprisingly).  Mine had dropped to just a little over 6v.  Bad. at nearly 60 Euros for a new pack… Very bad. Especially as I had just shelled out on a new charger.

I remembered from my childhood RC car days that NiCads have this problem called crystal dendrite growth.  What happens is that inside the battery crystals form, the crystal conduct electricity so when enough of them have formed they basically short out the battery.  What we used to do in those days was whack the battery across a car battery for a second or two and that’d sort it out.  It burns away the crystals and brings the battery back to life.  Also gets quite hot too! I remember that it works quite well,  not quite as good as a brand new battery but better than you might think.

That’s easy when the battery pack is 7.2 volts and you have a 12v car battery with effectively infinite current to zap with.  Not so easy when you’re living in an apartment with no car and the pack voltage is 14.4volts.

I tried with an 18v 2A spare laptop PSU but it made no difference so it was time for something a bit more adventurous.

Some people have had a lot of luck using disposable camera flash electronics to zap their NiCads.  Here’s an instructable. Others have used welders…

Just so happens I had a disposable camera that I had saved for a project just like this.  It’s a Kodak, not sure what model it is as they don’t print much on the outside. They’re easy to open, especially as I’m not bothered with re-using it, at least not as a camera.

Kodak disposable camera.

After ripping it apart the I had the circuit out and made a note of the polarity of the AAA battery (positive end is the spring on the circuit board).  After a quick test to make sure it still worked, using a wooden chopstick to press the buttons, I de-soldered the flash bulb and soldered in a couple of extension wires across the main capacitor terminals.

Flash circuit extracted from camera. Flash bulb disconnected. Extension cables soldered across capacitor.

Now I was ready to go.

Note:  This could be quite dangerous.  The capacitor is charged up to around 300V (measured) and when triggered the circuit boosts the voltage again to around 1000V (not that I measured it).  So be careful if you intend to try this. Do not short this across your body, it will hurt and might be life threatening.

Charge up the circuit, listen for the whine, then when the light comes on I disconnect the battery and short the extension wires across the terminals of the B&D NiCad pack, positive to positive, negative to negative.
It makes a loud crack and a small spark!

Battery terminals. Temp sensor connection is not used in my new charger.

The voltage at the main capacitor is around 300V.  There’s no need to actually trigger the flash as such, and I think that it’s probably best not to as that would mean the output voltage would be much higher as it dumps the capacitor voltage into a coil to raise the voltage to around 1000V.

After two tries my battery voltage had risen to nearly 7 volts. So I plonked the battery on the new charger for 10 minutes and the pack voltage had risen to 12 volts.

One more zap and another ten minutes on the charger and the pack voltage had risen to 15V  Hurray!  After an hour on the charger the pack was fully charged and back to normal.  The light even went green when the pack was fully charged!

I’ve been using the drill quite a bit since, putting up curtain rails, and have recharged that pack a couple of times. It definitely seems to be back to normal.  The capacity seems to be up to the same level as before the problem.

Not bad for free and certainly better than 60 Euros for a new pack!

Additional photos

Battery pack opened up.

Old charger (Now dead) with temperature monitoring

New charger.

Bottom of drill.

Various gizmos

Life does get rather busy sometimes.  For us it was having a baby and moving house in the same week.  Subsequently I’ve been rather preoccupied.

But that was nearly 6 months ago and things are getting a little more sorted out now, plus it was my birthday recently so I thought it was about time to get back in to electronics and also to add some new posts to this blog.

So to start with I ordered some gizmos from Deal Extreme, most notable was a bunch of tweazers to make working with SMD components easier.  There’s nothing like getting new shiny tools!  And for the price these are excellent value. The Bamboo ones are really nice.

Assorted tweezers from Deal Extreme

I also bought a bunch of stuff from a shop I’ve not used before called Seeedstudio they had some really nice SMD switches as well as lots of other nice things.  Their prices are great and the delivery cost was very reasonable.  Speedy too, I’ll definitely be using them again.

Here’s some of the things I bought.

Tiny surface mount momentary action buttons.

These were just 15 US cents each. Product info here.

Surface mount SPDT switch.

These were just 20 US cents a pop. More info here.

I’m a sucker for tiny switches, I seem to stockpile them.  I bought some lovely tiny smd momentary action switches with built in LED’s last year and I can’t actually bring myself to use them up as they’re so nice.  That’s sad right…

I also got these…

6V 60 RPM geared motors for the Robot that will happen one day.

PSP style variable resistor joystick with push switch.

But best of all, to get myself back into electronics, but without having to break out all my kit which is still packed away in the spare room, I bought this great Oscilloscope kit.

Digital Storage Oscilloscope Kit

I’ve started on it already, but I’ll write up a full post later on when I’m done.

So, there you go.  After a long blog drought I’m back at it.  Hope you enjoyed reading.

I had an email from a friend who has built my lighthouse circuit but whose LEDs were lighting up in the wrong order.  This is exactly what happened to me when I built it the first time too.  But no worries!

The great thing about charlieplexing and this circuit is that it doesn’t matter which order you place the leds on your board as you can change the code to match the correct sequence for your final layout.  This means that you can layout your circuit for ease of wiring rather than having to keep the LEDs in any particular order.  As long as the circuit is correct and that you have an LED everywhere you should have one the ordering can be sorted out later.

Of course if you are just using my compiled HEX file and don’t want to touch the code then you need to be a bit more careful and place the LEDs correctly as the ordering code is fixed into the HEX and can’t be changed without a re-compile.

If you want to change the code the bit you need to edit is from line 41.

// pattern for all 12 LEDs in sequential order.
const char charlieBits[] = {
0b10000010
,    0b10100000
,    0b10000010
,    0b10000100
,    0b10000100
,    0b10100000
,    0b10000100
,    0b10000001
,    0b10000001
,    0b10100000
,    0b10000001
,    0b10000010
};

const char trisBits[] = {
0b10011101
,    0b10011101
,    0b10111001
,    0b10111001
,    0b10011011
,    0b10011011
,    0b10111010
,    0b10111010
,    0b10011110
,    0b10011110
,    0b10111100
,    0b10111100
};

These two arrays store the two status bytes of the PIC output pins (GPIO) for each LED in the sequence.  The first row or the array is for LED 1 the 2 is for LED 2 and so on.  We are using GPIO’s 0,1,2 and 5 for the LED connections so these are the only bits that are relevant in these two arrays, the other bits refer to the button input on GPIO4 and GPIO3 which is defined as the MCLR (reset), bits 6 and 7 are irrelevant.  So for each row only bits 0,1,2 and 5 should change. Eg. 0b--X--XXX.

There are two arrays, the first (charlieBits) is for the GPIO output hi/lo status, the second (trisBits) is for the GPIO TRIS status which sets if the pin is an input or an output.  By combining these two array we have three states for each GPIO pin. When TRIS is 0 the pin is an output and so is either hi or lo depending on the charlieBit status.  When TRIS bit is 1 then pin is set as an input and so the charlieBit doesn’t matter.  When the pin is an input it is effectively non existent as far as our LED’s are concerned.  In actual fact it’s still connected but with a high impedance that doesn’t affect our circuit.

So to match the arrays up to how you have connected you LEDs it’s a bit of trial and error.  When I did  it I started by resetting the circuit and noting which LED lit up first.  There is a test pattern that cycles through each LED individually before starting the main routine which gives you a nice change to note the order.  You can alter the delay used here to make it slower or faster by changing the value on lines 160 and 161.
delay_ms(200);
delay_s(1);
The first delay is in milliseconds the second is whole seconds so we have a total delay of 1.2 seconds here.  You could make that slower by making the delay_s(1) into a delay_s(2) then recompiling and uploading to the PIC.

By repeatedly resetting the PIC you will be able to see in which order the LEDs light up in the test phase.  Once you have them noted you can go back into the code and re-order the rows in the two arrays.  Watch the LED that you want to be the first one, eg the one at the 12′oclock position. And note in which order it lights up – just count them off. Eg. if your 1st physical LED is lighting up 5th then you would need to move the 5th row of the array to the top of the array, remember to do that for both arrays to keep them in sync.  Re-compile, re-upload and check that the first physical LED is lighting up first.  Now do same for the 2nd physical LED (2o’clock).  You could try to do them all in one go but I found it easier to re-compile and upload after each change to check I was on the right track.

Sounds a bit tedious, but it’s not too bad really.
Hope that helps to get things up and running smoothly.

I’ve been using Google SketchUp recently and I’m really impressed with how easy it is to use. It also produces great render image for illustrating building projects.

Our new living room.

We’re moving house soon, fingers crossed.  The license is finally here and the now builder is cracking on with the works. It’s a long, long story… But now finally a year after we bought the place we can really start to wonder what it might be like to live there.

My wife Ruth and I differ in our approaches to planning the arrangement of our furniture.  I’m a measuring type of person, I hate having to guess or make assumptions until I’ve measured it.  Ruth is more of a visual, “It’ll fit, don’t worry” person.  This leads to some amusing disagreements that often end in me threatening to take out the tape measure and Ruth saying that I should show a bit more faith.

So to try and keep both of us happy I’ve been drawing up plans and Ruth has been penciling in the layouts.  We’ve got a great idea for some shelves that I needed to communicate to the builder so instead of spending hours with plans and elevations I thought I’d give SketchUp a go.

It’s fantastic. It’s a simple as that. It’s been years since I’ve done any architectural 3d stuff and I couldn’t believe just how fast and easy it was to get a quick plan in and flesh it out into three dimensions.  If only it had existed decades ago when I used to slug away at my powerMac with Archicad version 1. How many months of my life would I have back…

One of the best features is a special rendering mode called “Xray” that is a combination of wireframe and hidden line that is perfect for showing the construction of furniture or anything else for that matter.  Perfect for showing the builder just how I want the shelves built.

Hidden line with XRay mode view of the shelves I designed for the living room.

One thing I learnt very early on was that to build complex models and to make your life easier you need to use two main features.

The first is layers.  SketchUp can have as many layers as you want, different types of objects can be put on their own layers that can be turned off or on as needed. This really helps to reduce clutter when you’re drawing and can also be helpful when you’re setting up different shots for rendering. Eg I have a layer just for dimensions and another for furniture and one more for walls and floors.

The second really useful feature is components.  Components are groups of geometry that can be replicated throughout the scene.  The difference between normal groups and components is that all the duplicate components in a scene share the same geometry data, whereas groups do not.  This means that if you update the component, add more detail or change the colour then all the copies get updated too.  This is great for managing multiple copies of the same object that you might want to use in a scene, eg doors or windows, chairs etc.  Another great thing is that components can load their data from external files which means you can build all your complex objects in their own sketchUp files where they are easy to manage and edit then load them into an other file for creating your scene.  Components can also contain other components which can be handy too.

The cool thing about components being loaded from external files is that you can share them and lots of common objects are already available for download from the warehouse.  A quick search for IKEA returned 100’s of results.  You have to be careful though as some models are built better than others and some are built too well, containing every tiny detail, so you have to keep an eye on things. I’ve now built a bunch of object from around our house so I can see how they fit in our new apartment.  Some of them were reasonably complex objects, so building them on their own as a component makes it a lot easier to keep track of everything.

Any way enough talk. It’s free. Go and download it.

Here’s a few of my models.

Ruth's 8 Shaft Harris loom and stand I built for it.

Our planned sofa arrangement.(IKEA Manstead sofa with two Hol boxes and our coffee table/storage box)

TV Stand made from two IKEA Norrebo units, top unit accessed from rear.TV not shown. On wheels.

The shelves for the living room.

For anyone who is interested (Hi Richard) here are my project files for the lighthouse project.

The project is written for the PIC12f683 using SourceBoost Technologies BoostC that I use as a tool suite in  Microchip’s MPLAB.  BoostC is fairly ANSI so if you’re using another C then it wouldn’t be hard to convert.  There’s a free version of BoostC too and I think my program will fit within it’s limits.

C source code here (7kb). : Compiled hex binary here (2kb).

The schematic and circuit are drawn up using the free version of Cadsoft Eagle. They were created using version 4, but the latest version of Eagle is v5 which I’ve not used yet, however the circuit is fairly simple so I doubt there would be any problems.

High resolution combined print for top and bottom of circuit for toner transfer. (4800x2222px, 163Kb)

Schematic layout of circuit. Note that in the real thing I added a couple of DIP switches in the GP0 & GP1 lines just before the resistors R4 & R5 to isolate the ICSP data and clock lines when programming. (25kb)

I’ve not included the Eagle files themselves as they make use of some random parts that I’ve made in my own library.  But the circuit is not so complex that you couldn’t copy it.  If you really want them then let me know.

If you have any problems drop me a line.

I’ve really taken to paella since moving to Spain.  Previously i’d never go near that much wierdly coloured wobbley seafood stuff.  But after a few years and some really tasty meals I’ve been converted.  Anyway here’s my version of Paella mixta, that’s paella with seafood and chicken.  It’s based on books, my taste and our friend’s mum’s secret recipe. Check it out here.

Working prototype. The DIP switches are used to isolate the ICSP clk and
dat pins when programming. Only two LED are actually lit at any one time but
due to the exposure time of this shot it looks like loads of them are power up.

After about a day of messing about I finally managed to program the damned thing using ICSP rather than pulling the PIC each time.

Schematics lie! Especially when you draw them yourself. Guess what. Pin 5 of the PIC12f683 isn’t the ICSP clock line… Pin 6 is! Ah! Well there’s a lesson learned the hard way. Next time I’ll be more diligent when drafting things out in Eagle.

Here’s a quick video of the thing in action. It’s a bit fast for a lighthouse, more like a beacon. I’ll slow it down and maybe add a potentiometer to allow speed changes later on. I deliberately chose the input pin to be GPIO 4 as it’s one of the IO pins that is multiplexed with the analog to digital converter.

Apologies for the manky looking video, it’s hard to get
the camera to work well in low light conditions.

UPDATE Aug 7th 09 : You can get the Eagle CAD files here.

UPDATE Oct 2nd 08 : You can grab the source code and circuit diagram for this project from this blog post.

I’ve been working on a variant of the circuit I developed for my Lego beacon. It’s the same basic circuit but this time with 12 rather than 8 LEDs and rebuilt into a circular 40mm dia (1.5 inch) double sided disc.

The idea is to use this as a simple “lighthouse” that when powered up will rotate the beam around the 12 LEDs. I’m not sure why I’m building it, I have no real need of a lighthouse… Maybe it could be useful for a model railway or something similar. But I kind of like the idea of it and had some neat OSRAM SIDELEDs in my parts box that were calling out for some use.

This is the prototype circuit and as such has a couple of features that don’t really need to be there other than for the prototype. The first feature being a full size ICSP connector, the second, a socketed DIP version of the PIC12f683. Actually I used a DIP rather than SOIC pic because I happened to have a bunch of them hanging around and space isn’t so much of an issue here as it was in the Lego thing.

The circuit works fine to a point, but I’ve spotted a couple of problems that I need to resolving. The main problem is that I planned to program the device using ICSP but didn’t allow proper isolation of the DAT/CLK lines which means that my Pickit 2 doesn’t really like talking to it. The 4R7 resistors on pins 5 & 7 aren’t enough to propertly isolate them and the LEDs interfere with data transfer… Doh! I wasn’t thinking clearly there… I’ll hack this prototype with a scalpel and some fine wire to insert a couple of DIP switches to get it working properly.

12 LEDs is the maximum number of you can charlieplex from 4 tri-state pins. The PIC 12f683 is an 8 pin device that has 5 GPIO plus the MCLR pin which can be used as an input. I’ve decided to keep one GPIO spare, GPIO 4 which also happens to be an analog port so I could maybe add a light sensor that turns the lighthouse on when it gets dark (It’s got a button on it at the moment)

I made it using my normal toner transfer technique and it turned out pretty well. The smallest trace is 24mil although there is a tiny bit of 16mil in between the pads of the PIC. Once I’ve debugged this one I might have a go at one about 30% smaller as there’s quite a bit of space especially if I used a SOIC pic and dropped the trace size down to 16 or 12 mil. This version has an LM317 regulator on the back so can be driven from 12V or whatever but I am toying with the idea of making a smaller version that is driven from a 2032 coin cell mounted on the back which wouldn’t need a regulator.

Anyway, here are a few more photos until next time…

Top side of the circuit after etching

Bottom side of the circuit. Alignment was pretty much spot on.
Drilling later on was not quite so good… )

Populated top side.

Populated bottom side with LM317 (sot223 package)

I’ve been playing with this shape for a while. It’s supposed to be one of a number of engine pods for some sort of larger VTOL craft. They mount at each corner and swivel to allow the ship to maneuver. I’ve not got enough Lego to build the whole thing so I’ll have to be content with just building the engine pods for now…