Sunday 31 July 2011

Laser cutting magic clock

One of the great things I'm finding about having a laser cutter is just turning an idea into a product in a few hours. Such an opportunity presented itself this evening.

After playing petanque down on the seafront at the Brighton and Hove Petanque terrain, I got chatting to Ray - a longtime founder member and one-time president of the club. Ray is selling his house and soon moving to France, to spend his days drinking pastis and playing the game he loves, so it seemed a fitting to use my new-found skills to make some kind of memento - something that incorporates plastic and boules.

The obvious answer, of course is a set of child's plastic boules! The slightly less obvious answer (and the option we finally plumped for) is an all acrylic boules clock!

We had a sheet of A4 mirrored acrylic so quickly knocked up a design - drawn in Flash, exported as .wmf, then loaded into Inkscape and finally saved as a .dxf. This then allowed us to import the shape(s) into NewlyDraw, ready for cutting.
It seems like quite a faff but actually makes sense: you do all your drawing and editing in Flash (I personally love the Flash IDE for drawing!) then save the result as an image. You use Inkscape to prepare the file - no drawing here, but you can split shapes up, put them onto different layers, simplify splines etc. Then save as dxf and import into NewlyDraw. We just use NewlyDraw as the cutter driver - no messing about: load the dxf to cut shapes from, move the laser head to the start position and hit go. That's it!

Heres' the outline drawing

And here's a link to the dxf for cutting.
After assembling, we did hit a few snags, but only because we're still waiting for a delivery of various sized clock movements, so only had one movement to hand. It shouldn't have been a problem because it's on an acrylic-faced clock, but it didn't take long for the problems to start mounting up....

... firstly, with our clock consisting of a layer of 3mm mirrored acrylic on top of a base of 3mm blue opaque, we had a 6mm thick clock. That doesn't sound like much. But our clock movement was designed to fix to a 4mm or thinner face. So we had to re-cut the bit of the boule with the numbers 0-6, making the hole at the bottom wider, so that the clock movement attached onto to the blue layer only.

Then we spend ages getting the hole right for the hour (spiral) hand.
The clock movement was supposed to have a 5mm shaft. It took no less than five goes for use to find out that the hole needed for a snug fit is 4.7mm (we went 5mm, 4.9mm, 4.8mm - oh, this is going to take ages, let's skip a few and go for 4.5mm, oh that's a bit too small, what about 4.6mm, bugger it, if 4.7mm doesn't work, we'll have to give up and go to bed).

So what's with the weird spiral movement and peculiar numbering?
Well, of course, it's a petanque clock - so it's fitting that it should go from zero to thirteen (scoring in petanque goes up to 13, not 12). The spiral hand allows you to tell the time still, even though it's got these extra digits. If you think of the spiral hand as the hours and the short, straight hand as minutes, it's actually quite easy to read the time - in the example drawing above, the clock is reading half past three. Whichever number is covered by the spiral hand represents the hours, the minute hand works as normal. The photo, by the way, shows the time as 01:35am, which is what time it was when we finally finished making the first, working clock!

And the zero and thirteenth hours?
To be honest, it's best just to ignore them - they were only put there to make the clock look cool ;-)

Friday 29 July 2011

Miniature instruments make a return

Using the invisible instrument idea, and the early DeskJam prototypes (originally developed for a concept for Mattel in Sept 2010) we're soon to launch a range of miniature, playable instruments.
After being turned down by Mattel - their target audience was 7-10 years olds, and the instruments had to be dumbed right down to the point of losing their impact - they were also presented to Ravensburger in Germany. They too loved the idea, but couldn't get the cost of production down to less than $4/unit (apparently this was to be able to sell them for $40 each!). So we've decided to hang the expense, and make the little instruments here at Nerd Towers.

The original DeskJam range included guitar and bass (both worked in the same way) a miniature synth and the original instrument that kicked it all off - some miniature usb drums. As we're no longer going ahead with DeskJam - there were two other companies involved in the original project, but all the electronics, firmware and software development was strictly our own work - we're building the entire range again from scratch.

The first instrument in our new range of instruments is the guitar.
We're working on a range of guitar shapes and styles, but each will basically consist of the same set of components: a touch-sensitive neck for selecting the chord/riff to play, and a set of touch-sensitive strings, which the user can strum to trigger the sound(s). For anyone reading the blog a few weeks back and wondering why the topics had jumped around to include multiple touch sensitive inputs, things should become clearer from hereon in...

The first thing we did was create a multi-layered guitar shape.
For the main guitar body, we're working with two layers of 5mm acrylic - into which we've cut a void to give a 10mm deep space for the electronics:

The front and back are cut from 3mm coloured opaque acrylic (this laser cutter has been soooo handy!) and a scratchplate made from some 1mm HIPS. This super-thin material is just right for the scratchplate (3mm acrylic is far too thick) and means that when we introduce some 3mm thick pickups, they stand proud by just the right amount - just like a real guitar!

Finally the layers are all assembled and the final guitar is starting to take shape:

The channel along the neck is for the touch-sensitive board. PCB pins will be inserted onto the "back" (non-copper) side of some copper clad board, and soldered to the etched front. To allow space for the soldered pins, we needed to keep a channel running along the length of the neck.

So it's off to the cupboard to dig out some Ferric Chloride and to get etching. Finally, it feels great to be making PCBs again - and this time, the end product is ready and waiting to be populated with some tried-and-tested electronics. This is one project that might actually get finished within a reasonable timeframe....

Laser splash back

A few weeks ago we tried cutting some miniature guitar shapes from MDF with mixed results. At best it could be described as a partial success. With our swanky new LS3020 laser cutter, however, cutting such shapes is a doddle!

Like most laser cutter users, we've had to spend some time with different materials of different thicknesses and types, finding out which settings to use for the best cut rate.

For 5mm acrylic, we're cutting 15mA at about 8mm/sec.
For 3mm acrylic, we're using 15mA and cutting at about 18mm/sec.

These values give an incredible performance, far better than we'd hoped, especially considering how slow and difficult the laser cutter at BuildBrighton has been. We've tried cutting slower, with less power, and faster with more power - but 15/18 gives a nice balance between speed and power use (apparently, cutting with less power makes the CO2 tube last longer; I guess it's a matter of time before we find out how true this is!)

However, whatever speed/power settings we use, we're getting a bit of what can only be described as "splashback" on the reverse of each piece. It's more obvious on some cuts than others, but on the back of some pieces, the pattern of the honeycomb bed can be seen in relief. It's particularly noticeable when cutting black acrylic:

After removing the protective film, small pits can be seen on the back and on the edges of the piece:

Yet the "front" face of each piece is cut perfectly, with a smooth, straight polished edge:

We've had conflicting advice on how to correct this - some people say a faster cut with more power gives a better finish. Some people have told us that we need to turn down the power and cut more slowly. In our experience, both approaches still create this "crazing" on the back. Maybe it's something to do with the bed. Maybe it's time to put in another appearance on the LaserScript forums to see what advice any other users can give. Those guys have been so helpful and supportive getting us up and running - we can't be the only people to have experienced this problem. Other than this little niggle, we're absolutely thrilled with the laser cutter and can't wait to start making "proper" stuff with it!

Arduino and general geek-kits specialists sent a surprise - and very welcome - package to Nerd Towers this morning. It follows a discussion with Aaron at Oomlout (does it look weird with the first "o" capitalised?) which in turn came about after demonstrating a few project ideas from these very pages to Chris at HPC Laser (where we recently bought our spanky LS3020 machine)

Oomlout have been providing Arduino type starter kits for years and have excellent facilities up in Halifax. In fact, getting your stuff from them means you're dealing with likeminded nerds and geeks - these guys don't just put components into boxes and sell them on, they actually use and build stuff with them too!
A lot of the machinery at Oomlout has been designed and built by Aaron and the guys - cogs made from laser cut acrylic, arms moving on servos, all connected to homebrew control boards: if ever you need advice on what to get for your own projects, at least you know you're dealing with people who know exactly what you're going through!

Enough blather about what a cool bunch of guys these are, let's have a look at what we've got:

I'm personally no great lover of Arduino, but this starter kit has got me genuinely intrigued. As well as the main Arduino control board, there's everything you'd expect to find to get you started - breadboard, jumper wires, LEDs, resistors and so on - and some slightly more exotic components for the more adventurous, including a motor, pushbuttons, micro potentiometer (to make rotary dials), transistors, LDR (light dependent resistor), a shift register, a relay and a piezo sounder/buzzer.
All in all, some pretty interesting components - maybe not enough to make one big uber project with everything in it, but certainly enough to learn and understand about each one and what it could be used for.
As well as this little lot, we also got some micro servos (we're already familiar with what to do with these!) a big beefy continuous rotation servo - can be used as a high-torque motor for driving things around the place - and one of the most intriguing parts in the box, an ethernet shield.

Already we've plans for the ethernet shield, following our failed mBed HTTP client project. We'd love to find out how robust the Arduino-to-web communication is and whether that would be a more suitable for our SMS-to-web and switching unit projects.

However, the thing that is really impressive about the Oomlout starter kit is the documentation.

Each component in the starter set is introduced, with a description of what it is, what is does and how it can be used.

These are coupled with try-it-yourself type examples with real world images of what the project will look like on the board. Having run a number of workshops for BuildBrighton, I know how useful these can be. Working from simple schematics and photos off the internet is a very worthwhile exercise, but to new users, it can be confusing. Until you're comfortable with understanding how a circuit diagram translates into which wire goes where on the circuit board, a diagram showing what the end result will look like is an invaluable aid. These diagrams are also slightly "exploded" so you can see exactly what goes where - something which can be tricky if you're following a project from a website and it has only one or two photos of the finished project, taken from an obscured camera angle.

For me, the very best bit of the starter kit is the simplest of ideas.
Each project has a paper overlay which exactly matches the breadboard layout, and has the components used printed at actual size. The end result is a sheet which you can stick to your breadboard and simply poke the components through, ensuring that every leg and every wire goes into exactly the right place.

It's such a simple idea, I'm amazed it's not more popular.
I've never seen this approach before, and it's as brilliant as it is simple.

So even if you're an old hand, whether you're starting out, or just looking for something different to get out of a rut, check out the Oomlout website. If their level of attention to detail on the other products is as high as it is for the starter kit, you're sure of some great help and advice along the way - and that in itself is worth ten times the value of the components in the kit!

Sunday 24 July 2011

First shapes off the laser cutter

While waiting for a delivery of acrylic sheets (ok, it was only ordered a few hours ago and it's Sunday evening!) we've been playing about with the few sheets of clear plastic we got with the cutter, to find out which settings work best for us.

We found that cutting at 15mA, 20mm/sec works really well for 3mm acrylic.
Any slower and the beam reflects off the honeycomb bed and pits the reverse of the plastic. Any faster and the beam doesn't always cut all the way through on the tight corners of intricate shapes.

Here are some example logos for a new shop opening soon; it's called NIX and offers an outlet for cottage industries and artisans. The idea is really simple. For a fixed monthly fee, you get a shelf on which you can display and sell your wares.

Obviously these would look much better in coloured and/or opaque acrylic, but they're just test pieces to see what the laser cutter can do. And so far, the results are pretty impressive!

line drawing exported as dxf for importing into NewlyDraw

Here's the original business logo, to compare the results to:

Cutting at last!

After a few false starts, and a nervous first try, we're finally cutting stuff with our new LS3020 laser cutter! The tiny bubbles problem apparently is quite common with new tubes, and a few capfuls of Milton sterilising fluid soon sorted them out (thanks guys on the LaserScript forums for the tip!)

So now we're safely(?!) cutting acrylic with the laser, it's time to get familiar with the NewlyDraw software and find out which power and speed settings work best with which materials.

At the minute, we're trying on clear 3mm acrylic.
The laser cutter is whizzing little 10mm circles out no problem, at about 12mA. Now we just need to find out what all these settings in the software actually do!

After selecting the items to cut out and from the file menu, hitting ENGRAVE, the following dialogue appears. From here you can jog the head around or set it to it's origin (we found that -80,0 put us in a good starting point, at the top left of the sheet of acrylic) and even choose optimum cutting paths (more on that later, we're still trying to walk, let alone sprint off into the distance!)

After hitting "start engrave" a second dialogue appears

This has more of the same - a second chance to jog the head around - and some speed/power settings. We're not quite sure which of these relates to what: whether the speed setting is for jogging the head, or for cutting. To date, changing these values doesn't really make much difference: the laser head goes off on it's merry way, making short work of the 3mm acrylic, no matter what speed/power settings we put into this second screen. We're guessing that these need to be changed somewhere else, and we're probably giving it too much juice or not moving as quickly as we could, for the given power level or something.
Either way, it's another visit to the LaserScript forums to double-check we're doing everything right. It's exciting having a laser cutter. And a little bit scary!

Laser cutter from HPC

It's been a busy couple of weeks, with "real life" taking over for a while (always the case, as summer approaches, family holidays are arranged, lots of doubling up at work covering for colleagues who are lucky enough to take their holidays before the schools break up and so on).

As work on the touch-sensitive range of instruments stalled, due to cutting problems, we decided to take the plunge and make a heavy investment. In fact, we're making quite a few BIG investments at the minute, the least of which is taking out a lease on a shop to sell homewares, home-made goods and generally supporting the whole maker/cottage industries that are slowly but surely growing throughout the UK. As well as buying stock in, I've got a crazy idea about selling make-it-yourself kits and pre-assembled "geeky gadgets". And the first range of products? Yup. Miniature instruments.

So making them from MDF seems difficult due to the limitations of the CNC, what's the alternative? Acrylic of course! You can see where this is heading....

...HPC in Halifax supply laser cutters. And we're the proud owners of a shiny new LS3020 machine. We spent the day up there recently and saw one in action. It was amazing - blasting through 6mm acrylic in a single pass with no problems, cutting mdf, laser ply, and even solid maple, all 6mm thick, in a single pass at varying speeds (the slowest at around 8mm/sec cut 4 layers for an entire guitar in just a few minutes). Exciting stuff!

After getting everything set up and following the instructions and advice very closely, we're almost ready to cut. The only thing is, this is an expensive bit of kit, and very easy to get wrong (and break £150 CO2 tubes apparently). So we're a bit nervous of actually firing the thing up on our own!
The thing is, the CO2 tube had some air bubbles in it even after the machine had been switched on (with the water pump running) for about half an hour. We got rid of the big bubbles that had collected at either end, by tipping the machine then pinching the water inlet pipe for a second or two and releasing (returning the machine to horizontal). This worked really well for getting rid of the "air locks" in the tube.
completely air bubble free water jacket around the CO2 tube

The worry thing is that the main part of the tube has a lot of little tiny weeny bubbles along its length.

smaller than the bubbles in a Wispa bar, are these anything to worry about?

We've tried all manner of stopping and starting the water flow, tipping the machine, leaving it to rest, turning it off and going for a walk, switching it on and moving it around. It doesn't matter what we do, those little tiny bubbles don't want to go! The question is - can we run the machine with these little bubbles? Do they matter that much? Not knowing enough about it, we're frightened to fire the machine up fully and try cutting, just in case the tube gets busted. That'd be a real bummer. The machine's only a few days old; we don't want to break it before we get started!

Friday 8 July 2011

Another Thursday, another partial success

The touch-sensitive posts last week were leading up to making some cool stuff but we've been waylayed with other stuff and beset by problems. Not least of all, getting our CNC machine to cut at a reasonable speed.

Another night at BuildBrighton and I arrived early to get set up and get cracking on with things. In the last three previous weeks, I'd had problems with the PC running Mach3 (the milling software that actually controls the CNC) and ended up having to do a fresh Windows install each time, just to get the PC working!
This week I was determined I wasn't going to have to install Windows yet again. The PC, however, had other ideas. After plugging everything in and switching on at the wall, the HP Compaq base unit gave a little cough, popped, fizzed, at started to smell really bad. On further inspection, the entire power supply had frazzled. Not only did I need a fresh install of Windows, I needed a fresh PC!

Despite this little hiccup, Robot Steve came to the rescue with a little tiny base unit that was already pre-installed with Mach3, from when he ran his own home-made CNC Machine. Which was great. Until I tried to import some dxf drawings. That's when Mach3 started moaning about invalid parameters, unit scale lengths being all wrong, tool paths went haywire - basically nothing worked as it should!

It took nearly three hours to get the CNC reproducing the 0.1" pitch grid that we successfully got working last week - but eventually we were up and running!

I installed a routing bit into the Dremel clone that was used as a router, got a simple dxf drawing to convert to gcode (and draw at the correct scale) and set it cutting.

Things went well (except for the horrendous noise and thick smoke billowing out of the MDF) until about half-way around the shape, when the Dremel stalled and the CNC machine stopped moving, despite the PC still sending movement commands.
It seems that when trying to cut some 6mm MDF in a single pass, the router bit was biting into the wood and being pulled down, out of the collet. The end result being that the cut was getting deeper and deeper - eventually stalling the machine once it hit the metal cutting bed. Ooooops!

Undeterred, I set about cutting a new piece of MDF - specifically adding some feed-rate commands to slow the machine down on the sharper corners, and setting the cutting height to about half the wood's thickness (the idea being to cut the shape out in two passes). This worked to a degree - except that on the sharper corners, the smoke really did billow out from the cutting head - and on the straight bits, the router jarred and bounced around. In the end I had to go for multiple, shallow cuts, to get the shape to cut out at all.

the banding on the edge of the piece was caused by setting the router to different depths and making multiple passes

The idea behind using a router rather than laser cutting was speed - the router was supposed to cut through the wood in a single pass, whereas the laser cutter would take three or four passes to cut the shape out. However, it seems that laser cutting 6mm (or even 9mm) MDF over three or four passes might actually be quicker than cutting the same shapes out with the CNC machine (which after testing tonight, seems to take three or four passes itself to cut through relatively dense material).

It might be that the Dremel router isn't powerful enough.
Or maybe I tried forcing the cutting head to quickly through the wood.
Or maybe not quickly enough.

In short, I don't know why, but routing MDF to cut out shapes was only partially successful. IT worked, and was repeatable (the cutting head went to exactly the same start place and followed exactly the same line every time). But given that it took so long to get a single shape out of the machine, it might not be the quickest way to produce shapes from MDF.

Different coloured bands caused by running the router at different speeds to determine the best cut. Darker bands caused by running the router very quickly, moving the CNC machine slowly: it created lots of smoke!

But on a brighter note, after getting it to cut some wood, I decided to try a drilling pattern, in preparation for next time; I'm hoping to pre-cut some copper boards before applying a PCB design with press-n-peel and etching. If this is half as successful as tonight, it will save hours when making circuit boards for the up-and-coming projects on this blog!