Archive Page 2

Bring Back The Beko!

Our washing machine, a “Beko WMA 641 W”, broke down; no bangs, no smells, no intermittent behaviour, just dead. Having checked the fuse, the socket, the fuse again, it was confirmed broken.

Predictably, it decided to do this after passing the 1 year old mark, so no warranty. And like many appliances, getting it repaired is the same sort of cost as buying a new one, so time to go shopping…

But that felt just a little like the wrong answer. I’m not sure whether it was the idea of having to go shopping, paying again, shifting heavy white goods, or it ending up in a landfill that prompted it, but I became determined not to be beaten so easily. So instead, time to get naked…

Naked Washing Machine

The machine came apart easily, and seemed very well designed. It was very quick to confirm power was getting to the control board and that full disassembly was needed. Top tip: I have always clustered groups of related screws when taking things apart to help when re-assembling, but I realised for the first time that putting these groups in line is really useful. Probably obvious, but I had never done it before. So re-assembly is just working backwards along the line.

The power of chronology should never be underestimated! Our brains seem very good at searching things by time. We may spend ages organising photos in to folders, but given instead a huge list of photos ordered simply by date, we’d probably find what we were after more quickly.

Anyway, the PCB is out…

Beko Controller PCB

The controller is made by a company called Invensys. Some searching uncovered that these boards are about £50, but after firing off a few emails to potential suppliers I was met with silence. Knowing power was getting to the board, its time to try and understand what is going on. Multimeter in hand, some patience and PCB probing…

Beko WMA641 Schematic

The thing that stumped me for a while was that what I thought was the logic DC power supply output seemed to be connected through to ground. Short circuit? It turns out after a little more head scratching that the power supply was designed to produce -5v; suddenly with that insight the whole world flips on its head and things start looking a little more sane.

Having uncovered what looked like a switch-mode power supply design, I attached the power again and investigated that. I wouldn’t recommend probing a live board hanging out of the front of an appliance, but it helped me confirm the problem; the rail wasn’t achieving the voltage it needed to power the logic.

It Works!

By replacing the BC556 used in the power supply, I was welcomed with the sight above. That’s the “Ready” led that is!

So after all that, a 7p transistor is what brings the Beko back! Lucky i’m not charging for my time, but also cost is not really the issue. If nothing else, it saved me the mind-numbing process of choosing a new machine, and another big white box from landfill!


BarCamb 2 – Embedded Can Do The Internet Too!

I was lucky enough to make it to BarCamb 2 at the Genome Campus for a day of discussion about “the interface between science and technology”.

Alf showed off some cool Statistics Mapping he’d been playing with. He was using MapTube to show population demographic information (age in this case) on maps. Looking up Cambridge 20-30 year olds confirms what you’d expect; age increases as you move out from Cambridge to the surrounding villages, and Mill road is a bit of a beacon…

Cambridge 20-30 year old location

Michael was back with another incarnation of Ndiyo called Camvine. They’ve applied their hardware VNC-like technology to digital signage, with the focus around providing the content as a web service. Looks interesting and has the potential to be really slick, and I’m interested to see where their differentiation ends up in such a populated and competitive area.

I wonder if the same concept could be served better with some simple internet-in, video-out web boxes. I’d expect these to become really common over the next few years as single-package computers like that found on the BeagleBoard hit the market.

Andrew gave a great introduction to tangible interfaces, in particular Reactable. There is some really interesting stuff going on in this area and the technology is pretty accessible so you can knock up your own. The presentation also shows how bjork is in on it!

The hot topics of last year seemed to be the semantic-web, micro-formats and RDF. This year none of them featured much, and the new buzz thing was definitely git which got its own panel! It seems a great model, although I couldn’t help feeling that there was a slight mismatch between what people were saying was the great benefit (there is no central repository) and what they were actually getting the benefit from (being able to do local/distributed version control separate from, but still maintaining, a central repository). Worth a further look…

My topic was “Embedded Can Do The Internet Too!” – not really a talk, more a collection of examples and observations I pulled together on the day. The main point was that the internet is currently thought of as Browser/Website/Keyboard/Mouse, and I wanted to get people thinking about the potential when you break out of that and have microcontrollers using the internet.

One example was a project Chris did with a local school; they used an mbed microcontroller to build a data logger in to a wall power socket. It also exposed itself as a webserver using ethernet-over-powerline. The result is pretty cool; point your browser at the wall socket ip and see the log of the appliance power consumption!

Another was the MASHED08 hack we did which accesses the internet using GPRS over the mobile phone network.

BarCamb 2 Talk

I also demoed a simple example I put together of an mbed running a web server, where the webpage it returned showed the state of an analog pin in near real-time (input), and controlled some digital pins using check boxes (output). Not a particular exciting demo in itself, but it was more about the potential and I think people saw where it could go.

It was great to be in the company of such an enthusiastic bunch again, so thanks to Matt for setting it up. I got a few questions about SHDC too, so I’m hoping a few new faces might drop in sometime 🙂


If your a linux hacker and want to play on a computer that is a bit different, you might want to look at the BeagleBoard


btw. that chip in the middle *is* the computer!

The chip is an OMAP3 which means its got an ARM Processor (Cortex-A8, so nice and powerful and includes the NEON SIMD Engine :), a 3D Graphics Processor, RAM/FLASH and a load of other goodies all crammed in to that tiny package. Notice the lack of fan/heatsink too! This is the stuff that is used by the big boys in their next-gen mobile phones and set-top boxes etc, so it is great to see this type of technology being made available to anyone!

It seems like there are various linux distributions running on it already, and there are a few demo videos around. It is just becoming available now, so if you are interested then go and have a look:

Robotics meets Designers!

Browsing around Artbots (thanks Jonny), I came across something very similar to our Packet Network mashed creation. When we were first brainstorming on day 1, we really wanted to have the interaction part of it somewhat like a physical pull. We played with the idea of a dog leash pulling you, divining rods, or some sort of off-centre weight to indicate direction, but it never got much further than the basic tone as (being techies) we spent our time playing with the GSM and compass hardware.

Well these guys have done the tipping idea with real style! Take a look…

More details on their Momo website.

The real skill this demonstrates is knowing how to wrap up the complexities to give you something wonderfully simple. For example, try not to smile at this…

This is “Keepon”, a interaction research project. Lots of complex technology in a squishy yellow shell. He even features in his own VH1-style Pop-Up Video…

Worth searching for some more of the videos as he has some nice movement tracking, and some great youtube video responses (like this). All the details of the “Keepon” research can be found at


Chris and I went down to london and spent the weekend at MASHED08!

It was basically a hacking event where web-hackers from around the country were mashing together various feeds, in particular the stuff from the BBC and Microsoft/Multimap.

We decided we wanted to try and do something a little different by building hardware that interacted with the internet. Here is what we ended up with…

Packet network

It was the result of our 24-hour hack and an attempt to bring the all the web 2.0 “community” jazz in to the real world; a community based courier service 🙂

The Idea

Packages have RFID tags. People have readers. You scan a package and using a web app set the destination (click on a map, type an address, whatever). You put the package down. Someone else picks it up and scans it. It connects to the internet (mobile phone network) to find out where it wants to go, and starts pulling you in that direction. No indication of where it is going, just the direction it wants to go. Walk for a while, but now you want to go in a different direction? Put it down and leave the next leg to someone else. Hitch hiking packages!

The Implementation: ARM + RFID + GPRS + GPS + COMPASS + PIEZO + WEBAPP!

You swipe an RFID tag of a package and the micro detects and reads the ID. It then reads the current location using the GPS module and packages that and the RFID id up in to an http request, made to our server over the internet using the GPRS module. The server looks up the package identified by the ID, looks where it is going and using the GPS co-ordinates we sent, calculates the direction you need to go (0-360 degrees, clockwise from north). The server replies to the http request with this direction. The micro then plays a note via the piezo sounder based on the difference between the angle it wants you to go, and the angle we work out from reading the compass on board. Highest pitch = the direction you want to go. A bit like a metal detector, but it’s the package pulling you in the direction.

Mashed 2008 presentations

At any point, someone with the map web front end can see where the packet is (the reader updates the location by continually sending that over the network too, so the ajaxy map keeps that up to date) and change where the package wants to go by clicking – this just updates the database and the next time the package checks which way to go, it’ll get told its new heading.

I kept a log as we developed it, so you can see more of what we did at

It was pretty hard work, great fun and lovely to chat to so many interesting people as they drifted by. We aim to be back next year!

mbed Microcontroller Beta Launch!

We decided the best way to launch something was to put it in a rocket…

Take an mbed Microcontroller, add an 18g accelerometer and attach a PP3. Wrap in a polybag and tada, a rocket datalogger!

mbed rocket logger

The idea is to put this in the body of the rocket. Model rockets have a motor which after burning through pauses for a while before firing on the other side. This is designed to blow off the nose cone to release the parachute. As this is where our logger is going to be, that also means the logger will get blown out! The green wire coming out the top is our solution to tie it to the rocket nose cone so it doesn’t go to far.

We had been hoping for good weather. It was sunny but windy. Actually, very windy, to the point of being hard to stand still easily. So that unfortunately meant it wasn’t suitable for firing rockets.

But we did anyway…

After running to collect the rocket before it blew away, we opened it up, plugged it in to the laptop and fired up excel. The data looked good – first time too!

Here’s the resulting data from the flight, annotated with what we worked out was going on…

Annotated flight data

A rocket datalogger built in an hour, and fired the same afternoon giving us results first time. I’m happy with that for a launch!

Light Box!

I wanted to get some photos of a small product; the sort of shots you get with a totally white background around it. Luckily Shareef has a flashy camera, so a homebrew attempt was on the cards!

Although I know you can basically “fix” pretty much anything in photoshop, i’m not convinced by it. I think bad attempts to mask out a background look worse that if it was just presented on a dirty kitchen table. I think you basically have to *really* know what you are doing to pull it off properly, and to be honest, most of us don’t. Stick to simple limited global adjustments, like normalisation and croping!

It is very similar to my brief experience in music recording; don’t think you can fix the sound in the mix. If you want a good sounding snare drum, spend time fiddling with the tuning, tension and mic position/type. If you record it crap, it’ll sound crap. Doesn’t matter how much EQ and reverb you chuck at it. The trap that is easy to fall in to is overdoing it by focusing on the aspect you are tweaking; a louder stereo usually “sounds” better in a snap comparison.

So the challenge was to get it right at the point of taking the photo, to leave as little post-processing as possible.

The first attempts centered around a white piece of card with as many lights as we had around. We quickly found that pointing lights cast harsh shadows, which were pretty much impossible to “fill” with light from the other side. Conclusion 1; no direct lighting.

We moved to firing lights anywhere but at the point we wanted, mostly firing the lights at the ceiling. This had better results, but still fairly dark and with some shadows, meaning the quality was not that great…

Uplight attempt

After trying to fix the results in photoshop and failing (metal parts becoming white etc), it became clear we had to try harder to get it right before hitting photoshop.

Some googling convinced us we needed something a little more adventurous, so a box, white tablecloth and the white card was quickly combined to make a light box. Slightly more challenging to get the uplighters firing in the right direction (they are about 2 meters tall, so lying down required openning the patio doors to fit!). Firing all the lights in to the box gave immediately much better results.

We were still getting some shadows, and reflections on the floor off some metal parts so the results weren’t perfect. The final inspiration came when realising that lifting it a few cm off the floor was enough to disperse all shadows. So a little bit of magic (pencil/bluetack) and the solution was complete…

Light Box

The photos we got out of this were great, and just needed some normalisation (or whatever it is called in image-speak; basically shift where is full white/saturation) to make the whites whiter than white!

The only problem was we were croping a lot of the background of the final images, so weren’t ending up with the full resolution potential. So time to find a macro lens to borrow…