[ Eyebeam bio ]

Working with Hernani Dias on the Re:Farm project (see my previous post for some more info about the Re:Farm project), my first objective is to create the PCB for the Re:Farm On The Wall module, a data visualization module using an LED display to present data away from the computer screen.

[ details after the break ]

Here’s Hernani’s original sketch for displaying the soil moisture level from 6 sensors:

This design uses 7 LEDs to display the soil moisture level for each of 6 sensors (42 in all). The design concept has been formalized as seen below:

The module still has 7 LEDs for each sensor (or farm/garden), but with the addition of LEDs between the groups (for a total of 48) to create a complete circle which can be used with other templates, such as the one below, for visualizing other data such as temperature or humidity:

Ideally, the board will also incorporate a stepper motor to drive a needle (like a speedometer) for an additional data channel.

I’m going to first concentrate on the LED display. Since the board will make use of an ATMEGA microcontroller (Arduino) chip and an XBee radio (to receive the data), real estate on the approx. 3″ x 3″ board is already a bit tight, so I am currently researching Charlieplexing, an LED driving scheme which allows N*(N-1) LEDs to be directly driven by N pins of the microcontroller. 8 data pins from the microcontroller can thus be used to control the 48 LEDs since Charlieplexing with 8 pins allows control of up to 8*7=56 LEDs.

At The Next HOPE conference this past weekend (blog post about the conference forthcoming), Jimmie P. Rogers was demo’ing and selling his LoL (Lots of LEDs) Shields for the Arduino. He has a great post about his work with Charlieplexing which I’m currently consulting. Talk about serendipity!

In a way, the overlap between the needs of Hernani’s project and my own interests in lighting and internet-enabled lighting fixtures is serendipitous as well.  It’s great to be working on a project that is seemingly so diverse from my own projects, yet has so much overlap on the technical side.

It’s also been interesting to observe Hernani’s design approach/philosophy which stems from his graphic design background.  It seems like everything starts from the aesthetic: the documentation for the project has some of the nicest component diagrams I’ve seen; and even his calendar is created in Illustrator

This post assumes some knowledge of PCB etching methods and is an account of my experience with the toner transfer method. If this is new to you, you may have to get Googling on a few of things mentioned here. That said, I’ve linked to the sites I found most useful during the process.

Intro

Up until now, I’ve been hand wiring all the the circuits for my Electric Window series.

It’s a fairly straightforward process to build the circuit on a piece of protoboard, and for an Electric Window 3 circuit, it takes about two hours. The bottleneck of this process is preparing all the connecting wires (shown in white in the above photo) and ensuring all the interconnections are correct, ie, connecting the right components to the right pins of the microprocessor and the power and ground points.

While I find this process efficient/adequate for one-off circuits and prototypes, it gets monotonous when making multiple copies of the same circuit.

For our recent contribution to the Chinatown Remixed exhibition, Deb and I (working as The Latest Artists) envisioned a series of Electric Window 3 light boxes, displaying various animations along with text from Twitter via a wireless internet connection. I decied to use this as an opportunity to finally learn how to etch PCBs.

I’d previously been intimidated by the PCB etching process for similar reasons as Collin from Make:

There seemed to be a steep learning curve, as I’d yet to use electronic CAD software such as Eagle, and the actual etchning process seemed finicky and messy.

After watching Collin’s video, a few others on YouTube, and reading some on-line tutorials (found via Google), I decided to attempt etching with the toner transfer method, since I already have a laser printer and clothes iron.

[ More after the break . . . ]

Schematic capture and circuit layout

The first order of business was to electronically capture the schematic and create the board layout.

I knew I’d be creating a single-sided board for the sake of simplicity, so I created a rough layout by hand — based on my existing hand-drawn schematic (not shown) — to get an idea of where the components should best be and which microprocessor pins were best to avoid crossing wires and/or requring jumpers:

As I said, rough, but it was enough to get going.

While there are plenty of software alternatives, free and otherwise, for schematic capture and PCB layout, I decided to go with Eagle because a) it’s free to use for 3″ x 4″ boards, and b) there’s an abundance of resources/help out there for it.

I used this helpful tutorial and parts library from Sparkfun to get going with the program, and soon enough I had my first CAD schematic:

I then moved onto the PCB layout portion and used the auto-route feature to start the intimidating process of trace layout. With a bit of additional tweaking by hand, I had a layout I was happy with. I originally designed the board to be 3″ x 1.5″ but that resulted in tight traces and traces running really close to microprocessor pin connections. Being unsure of the tolerances/accuracy of the etching process to come, I decided to give the circuit more margin and ended up with a 3″ x 2″ board:

Toner transfer

Although I’d repeatedly read that glossy or photo paper was best to print the circuit for transfer onto the copper board, I didn’t have any of these fancy papers on hand, and didn’t want to make a trip to the store if I didn’t have to. So I decided to try it with regular paper.

I readied my otherwise unused clothses iron:

Naturally I didn’t turn it on until resting it on its back, and I wish more turorials were explicit about whether or not you should have water in there or not. The answer is no.

I cut my copper board to size and used a sponge to clean it with isopropyl alchohol (a 70% solution from the drug store).

I grabbed a scrap piece of wood to do the ironing on:

and it didn’t work. After several attempts to iron the circuit pattern onto the pattern from regular paper, it just wouldn’t stick. Whenever I’d move the iron around, the paper would slip and the result would be a smugged circuit pattern on the copper board.

I really didn’t want to make a run to the store for special paper, so I consulted Google instead. I found this great article explaining how you can use glossy magazine paper, the principle being that the laser toner doesn’t stick very well to it, thus making it easy to transfer onto the copper — the same principle behind using special glossy or photo paper I’d previously read about.

I set to work with the cheaper and more easily available magazine paper I already had around, trimming it to fit my printer before making a few prints of the circuit:

Back to the ironing board:

and it stuck! The paper stuck pretty much immediatly after applying heat with the iron, and I was able to move the iron around and evenly heat the entire circuit. Soon I could see the traces through the paper:

I let the board cool a bit, then put it in a sink of warm water. The magazine paper came off very easily, and the traces didn’t budge or scratch away. I’d read that you need to let the paper soak for 5-10 minutes before it’ll come off the board, but I found with the magainze paper it only needed to soak for 2 or 3.

I’d also read that you need to press really hard while ironing, and iron for 5 or more minutes. I’m not sure so much force and time are required with the magazine paper technique, as I ended up with a few smudged traces:

Specifically, there were a few pads that had shorted together on the transfer:

Using an X-Acto knife, I was easily able to scrape spaces between the pads before etching.

Etching

Now the fun part: etching with chemicals!

I decided to use ferric chloride since I didn’t want to mix my own etchant with ammonium persulphate and water, or the like.

I bought some handy etching trays with corner spouts at the local electronics store, and with some chemical resistant gloves, got set up in the utility sink:

Rubber gloves on from this point: I filled one tray with water for rinsing, and I poured about a centimetre (0.4″) of etchant into the other one. I put the circuit board, traces up (more about this later), in the etchant, and let it soak for about 15 minutes. Not much seemed to be happening, so I decided to try the scrubing technique described in Sonodrome’s great tutorial video:

It worked like a charm, the exposed copper started to come off the board, while the circuit traces held up well, even against fairly rigorous scrubbing. Caution needed to be taken at this step to avoid splashing etchant around, though!

Here’s the etched board with the toner still in tact. I think it looks silver/white from remaining fibres of the magazine paper:

Next I used isopropyl alchohol and a sponge (seperate from the one I used to scrub with etchant) to clean off the toner. Some of the pads had smugged edges, faithful to the smudged transfer image, but the board had no short or open circuits:

Drilling and assembly

I trimmed to board to length, and drilled the holes. Since I aready have a Dremel tool, I bought the Dremel drill press attachment and a package of small bits. I also bought the Dremel keyless chuck to accomodate the smaller bits.

Despite not being carbide bits, as recommended by many tutorials for drilling PCBs, they worked fine, and didn’t break or noticeably dull.

Lastly, I populated and soldered the board:

Not the most photogenic, but totally functional I wanted to tin the board, but the local shop where I got my supplies didn’t have any tinning materials.

Initial impressions

So, did I save any time!?

Let’s compare. I takes me 2 hours to make this circuit from protoboard. I’m familiar enough with the circuit that I don’t need to keep refering to the schematic, the real time taking tedium of this method is cutting all the connecting wires.

To learn Eagle and create the PCB layout, I spent about 6 hours. Since this was my first time through the process, I’m not sure how much of this time I should count against the protoboard method.

Etching and drilling, including the necessary clean-up of nasty chemicals was about an hour the first time through. Again, part of a learning process.

Soldering the circuit took only around fifteen minutes, though.

For one circuit board, this definately didn’t save any time.

Rinse and repeat

I needed 3 more boards for the Chinatown Remixed installation, so I did all these in a second etching session.

I made a few modifications to the layout, to run the traces clearer of the microprocessor pads, and to move the programming header (bottom centre) away from the LED screen connector (upper left) to make it easier to program the board when attached to the LED screen (glad I didn’t make all 4 like this!).

Here’s the 3 boards after ironing the toner transfer to the copper:

I did the 3 boards seperately instead of together on one larger boards to experiment with the timing of various steps of the process. This time around, some of my traces broke, so I repaired these with a black Sharpie marker.

All the boards etched successfully, including the portions repaired by the Sharpie. Next, I drilled them: all the small holes first, followed by large holes; I did this in case I accidentally drilled a large hole with a small bit, allowing me to enlarge it if necessary.

Next impressions

So, did I save any time across multiple boards?

The answer is yes! Although, for this circuit I only saved about 15 minutes per board compared to the protoboarding method (not including the time spent learning Eagle). However, I assume there would be more time savings with more complex boards.

That said, I do feel that I saved labour. Soldering each circuit together on a protoboard, making sure to correctly wire each connection requires considerable mental attention, in contrast to the “dumb labour” required to etch and drill the boards, followed by the comparatively easy task of populating the boards without having to handwire all the interconnections.

Another advantage to the PCB method was that Deb and I could work in parallel to solder the boards together.

I also imagine I’ll shave some time off as I become more proficient with the process, so I’ll definately continue to etch my own PCBs!

Some tips

I would highty recommend the use of glossy magazine paper for printing your circuit designs on: it’s cheap/free; you probably have some already; and it comes off in water very easily and quickly.

If you use glossy magazine paper, I found that a lot of ironing pressure wasn’t required, just be firm. I also found it only took about 3-4 minutes (for a 3″ x 2″ board) to do the transfer. More time can actually start to smudge the transfer, as discussed above. You’ll have to experient with this.

High speed steel (HSS) bits seem to work fine. I ordered some additional bits from Digikey for drilling the PCBs, since I’ve read so many times that anything other than carbide bits will dull very quickly when drilling PCBs. I’ve found the HSS bits to work well so far. Bare in mind I’ve only done about a half dozen boards. However, I have yet to break a bit and the bits are much cheaper than carbide. I just bought a bunch of bits to compensate for any dulling. For the price of one carbide bit, I was able to buy 3 HSS bits.

If you are not agitating (ie, constantly moving) the board in etchant, place the board traces up. I’ve heard that putting the traces face down enlists gravity to help remove the etched copper, but I found the ridges on the bottom of my etching tray started to leave lines on my circuit when I didn’t agitate.

The finished product

If you’ve read this far, you’re probably curious about the final product these circuit boards belong to!

They’re used to drive LED displays from Sure Electronics, for a series of Electric Window 3 light boxes in a public art installation.

The following picture shows 3 of the Sure displays cascaded, and the driver circuit with an XBee radio attached:

The XBee is used to wirelessly query a WiFi-enabled master controller (not shown) for strings of text to display, from Twitter; this post has a clear explanation of the final installation.

The black, rectangular connector on the bottom-left of the driver circuit mates directly with the LED display, as shown below:

The dangly bit on the bottom left is the power connector.

Next these units were installed in custom-made, acrylic enclosures:

Here’s some “still on the bench” shots of the finished boxes:

The deadline for this project didn’t permit time for better photos of the final boxes before installation on-site, so stay tuned for more pics and videos.

Thanks for checkin’ all this out, all the best with your circuit etching endeavors!

This piece has a colourful history

The original, static fixture shown above was built for Cube Gallery’s 2008 exhibtion, Homage, which asked participating artists to create an homage to an important influence on their art.

I chose to make a wall-mounted rendition of Dan Flavin’s fluorescent installation, “untitled (in honor of Harold Joachim)”:

For the Conjunction Collective show this fall, I decided to animated the piece . . .

I had an older, unfinished light work on the bench that I’ve slowly been pecking at for parts, based around a PIC-controlled relay board — with a little reprogramming, I’d have a quick update for the Flavin homage without having to build any new hardware.

First order of business was to make sure the old control board was still working, so I set to work with the cursed alligator clip connections:

I say cursed because I find it all too easy to short circuit something while working with the clips, but the ease and speed with which you can get components talking with one other is too tempting to pass up.

This is a common sight on my bench while I’m experimenting, and I’m sure I’m not alone on this one.

Here’s a close up of the old control board, still in its original project enclosure:

The red light is on, a good sign! The board was working fine, so I set about programming it specifically for the Flavin piece. Since everything was (temporarily) connected I figured I may as well get it behaving satisfactorily before mounting it up permanently in its new home.

I’d built this particular board before discovering Arduino, so it was based around a PIC 16F628. I like these chips because they have an internal oscillator (no external parts needed) and have enough input/output ports for smaller projects. Plus, there’s a couple free compilers out there.

I wrote up a new program in Hi-Tech C with a bunch of new functions for actuating the lights, only to discover it wouldn’t compile due to the math library needed for random numbers exceeding the available memory space on the chip.

My ideal solution would’ve been to just stick an Arduino chip on the board, but there was neither room to mount a new socket for it nor a matching pin-out (compatibility) with the existing chip.

Wanting to stick with the original plan of reusing hardware rather than building something new, I turned back to a previous (free) compiler in my toolkit: XCSB BASIC. It didn’t take too long to rewrite my C functions into BASIC, and soon enough I had the Flavin piece up and running to my liking.

I mounted the reprogrammed control board on the back of the piece:

Not the cleanest cable management, but it did the trick:

The chip still has 12 unused input/output ports, so there’s lots of room for expansion by the way of input sensors and additional outputs if need be. The Flavin piece is only using four of the eight available (yellow) relays already on the board.

Here’s a video showing it’s behavior. It has three modes which are randomly chosen for a random duration: toggle / change the state of a colour; only one colour on at a time; and any combination of colours. In each case, the colour to modify is random, as is the interval between colour changes. I like a lot of randomness

The video includes sped up and time-lapsed footage, as the actual timing of the piece varies between the speeds shown below and substantially slower; in an attempt to not bore viewers, I took a “flashier” approach to the video documentation.

[ As an aside about the music in the vid., normally for my videos I like to use some loops or beats I've already recorded, but in this case, I was coming up empty handed in my search for the right vibe amongst my existing recordings. So on went the drum machine, I picked a few sounds I'd previously tweaked, and recorded the soundtrack live into the video as a "voice over." I was quite happy with the results and perhaps it'll be the start of a new track? ]

I decided to name the piece “Flaven” after my misspelled BASIC file for the program.

Something that really became pronounced to me after animating a Flavin-like piece is the contrast between Flavin’s use of direct and reflected light. Seems obvious in writing, but the effect this new piece has when throwing light around an exhibition space is quite dramatic!

This piece and several others can be seen at the Conjunction Collective show in Toronto until November 8, 2009.

Pics and vids from the exhibition coming soon; in the meantime, check it out in person if you can!

The work of production is really reduction.

From this video:

Found via Innercityvison’s blog.

Here’s a fantastic survey of embedded computing platforms, covering everything from Arduino to plug-and-play modular systems:

http://www.partly-cloudy.com/misc

Makes for a really handy reference!

and my Conduit sculpture

I realized that experimenting with new projects with similar lights would be easier with a common development platform . . .

As part of a larger project supported by an artist grant from the Ontario Arts Council, I was able to build a prototype light sequencer, based on a PIC 16F877 microprocessor and two custom solid-state-relay output boards:

The PIC model was chosen based on its native I/O capabilities (16 digital outputs and several digital and analog inputs), and the solid-state-relays were chosen based on their high current carrying capability and low on-resistance, resulting in a negligible voltage drop across the relay (I couldn’t find an ideal transistor with a consistently low voltage drop when switching different load types such as cold cathode tube ballasts vs. EL wire ballasts, and was thus dissatisfied with the amount of voltage I was losing just to the transistor when switched on).

The PIC controls the output relays to directly switch the power from an external power supply to the loads (lights, motors, etc.); the advantage of this arrangement is that the output voltage and total available power of the sequencer is governed by this power supply which can be easily substituted based on the voltage and current requirements of a particular project.

The modular design of the system — main processor board and separate relay boards — allows me to mount the sequencer on the back of fixtures

or house it in its own enclosure:

The control board also has screw terminals for attaching input modules, such as this prototype proximity sensor based on a Sharp IR sensor:

Having a modular, easy-to-connect framework for the controller, sensor modules, and lights allows me to quickly experiment with various interactive arrangements, and has lead to several site-specific, interactive lighting installations.

Below are a few sample videos of some temporary installations built with the system. These installations used several motion sensors to control the light colour based on the direction or location of people in the exhibition space, while controlling the speed of the sequence based on the number of people passing through the space; the installations’ behaviour were thereby a reflection of the human activity within the space.

Future plans for the sequencer include migrating from a PIC processor to an Arduino (simply by changing the processor board), making more input modules, and creating a wireless network between the input and output modules.

More info can be found here.