What has Xander been up to?

Thu 02 August, 2007

I've requested samples Sipex on a number of parts that may be useful for our project, specifically dealing with the power requirements of the LED dish.

Part No.	Description
SP7655	Wide Input Voltage Range 8A, 300kHz, Buck Regulator
SPX29502T5	5A Low Dropout Voltage Regulator

*keeps fingers crossed*

Also, I've gotten the green LED's wired up. They're bright.

(I currently don't have a power supply beefy enough to drive these at full blast. This picture was taken with the LED's being driven at 50%)

Wed 01 August, 2007

Who knew gluing LED's on a plate of metal could take so long? Jesus.

Anyways, the gluing is done. Zoe's dish is ready for wiring.

Mon 30 July, 2007

12 of 27 LED's have been epoxied onto the dish. It's slow going, I'm only able to glue 3 of them on at a time before waiting 30 minutes to glue more.

Sat 28 July, 2007

I finished machining Zoe's light dish tonight. Now I need to epoxy on the LED's and do some wiring.

Fri 27 July, 2007

Obsolescence sucks. Check out this new product that has come out, a RGB Luxeon Rebel Star.

Depending on what part # Rebels they used, we could have gotten away with using only 11 of these (versus the 24 RGB's we have now) to get the same amount of light (with better color blending) and paid about 25% less (or we could have needed 20 and paid 36% more, again, depending on what particular Rebel they used (they don't say)).

Oh well, we already have our LED's so it's too late now.

Speaking of, I'm nearly done machining a block of aluminum into a round dish to use for mounting our LED's on. I hope to finish it on saturday.

Thur 12 July, 2007

sigh I really need to get better about looking for off-the-shelf solutions to problems before trying to roll my own. Case in point, this lovely solenoid driver board which drives 8 solenoids from 5V inputs from a microcontroller (just like mine), and only costs $30.

DC Opto-Isolated Output Board

Mind you, my board drives 9 solenoids, not 8 and thus this wouldn't be a complete drop in replacement. However, yeah...

oh well.

In that vein, we might well consider this cheap little switching 5V powersupply board from the same vendor:

Power Supply Mini Board

Costs about $6 and can supply 1A of current. Might be enough for our main logic cabinet.

Thur 14 June, 2007

I evaluated the Soekris net4521 for the role of central Zoe brain and believe it will be sufficient to handle the job. The chief win for the net4521 is that I have a spare one of them lying around, and also it can drive a wireless networking card as well as use a compact flash card for disk storage. The compact flash is important since I want to be able to have a full compiler on-board as well as perl/python, etc in case software patches out the field are needed.

Routerboard, which I had been considering earlier, doesn't have compact flash storage on their lower-end units, meaning it would be painful trying to fit a compiler on to the OS installation. But again, the chief win with the net4521 is that I already have one.

Also I did some exploratory surgery on the net4521 and found that it's voltage regulator provides a +5V line rated for up to 2 Amps, and the rest of the board only draws around 300mA (without a wireless PCI card installed, which ups the draw significantly I hear). Thus it's possible that we can pull our 5V supply line, at least in the main cabinet, off of the net4521.

I had a mishap earlier today where my STK500 board (the board I program the microcontroller chips with) stopped working. I traced it down to a rouge blob of solder that had flown off earlier and splashed over the STK500's circuitry, creating a short between VTARG and GND. Once removing the blob and after a significant amount of coaxing, I was able to calm the STK500 down again and get it to work. I have to say I'm very impressed with the engineerng on the STK500.

I made significant software progress today. The caustics generator controller is now able to generate 9 independent and arbitrary pulse waveforms meaning I can basically set seperate firing frequencies for each ripple generator. Also I wrote a rudimentary GUI interface in Perl/Tk (my first Perl/Tk app) to control the solenoid frequencies and got the serial communications between the controller and my laptop to work. It's ugly, but it works.

If the hardware mounts for the solenoids were done, we'd be able to start playing with ripple effects now.

holy jesus am I tired.

Wed 13 June, 2007

I woke up (without an alarm!) at 7am this morning and was soldering by 8am. I finished the causgen controller board and have started working on software. I'm kinda tired right now so I'll keep this note short.

Tue 12 June, 20007

Xander's week of Zoe continues today with a full day's worth of labors from wake to crash. I didn't quite make it to my goal of finishing the caustics controller board however I came very close. I have the board etched, inked, drilled and sealed and all that's left is to solder on the components which has to wait until the sealant dries. I expect to have the board completed and under test by lunchtime tomorrow.

Adam sez

Are you using photosensitive boards or how do you get the lines so pritty?

Xander sez

Nope, I'm using the toner transfer method. I went off of this guys guide for making PCB's and followed it pretty much to the letter, using the same Staples brand "Photo Basic Gloss" paper and a laser printer. I have to say, the results have been stunning. I've since read posts from other people trying different methods or different paper and I get the sense that the results I've been getting are way above the grade. I've so far etched a total of three boards with this method and each one has been essentially flawless. However, I do have a bit of trouble getting a good transfer of artwork onto the component side of the board, however that's purely a cosmetic issue.

(update 07/27/2007)

For more information about the toner-transfer technique I used to make the board, check out this podcast on Make Magazine's website, featuring your's truly

Mon 11 June, 20007

I've taken this entire week off from work in order to focus on Psyphozoa. In particular I'll be focusing on the caustics generation system and hope to have the hardware side of that mostly done by the end of the week.

I'm off to a good start. I've already finished building the caustics generator driver board and so far it seems to test ok. The causgen driver board is the muscle part of the caustics generator subsystem while the "caustics generator controller board" is the brains. I have the design for the controller board pretty much done but I want to spend a little more quality time dotting the 'i's and crossing the 't's before I commit it to silicon.

Check out the driver board and note that I didn't skimp out on the blinky. It wouldn't be burningman without the blinky.

Now I need to run off to Raido Shack and pick up more copper-clad PCB material.

Fri 8 June, 2007

Ok, it's been a while since I've updated but a bit of stuff has been going on all the same. Here's a nutshell update.

The caustics generator control board is nearly complete. I simply need to put some final touches and I'll be ready to etch a (hopefully) working prototype board. I redid the board to be single-layer which will save me the hassle of attempting to hand-etch two layers of a board at one time and hoping they align properly. One question that my brain is still chewing on is how do we interface with the CAN boards? Do we simply mount them seperately and run 7-pin molex-connected cables between them or do we instead perhaps do some sort of daughter-board type thing where the CAN module actually plugs in on top of another board?

The caustics driver board is also coming along nicely.

I've spent a bit of time dealing with connectors. What connector do we use to plug the solenoids into the caustics controller board? What connector to supply 5V power to the board? What about 12V? 12V connector should be incompatible with 5V connector otherwise someone might mistakingly plug 12V into a 5V board and probably fry it instantly. Etc. I've spent a bit of time in the molex online catalog going back and forth trying to find solutions that DigiKey? actually sells enough parts to make workable.

Of course, that's when I realized something that I shamefully should have never forgotten: recycling! A trip to Re-PC netted me a box of sketchy 'as is' old computer hardware, most of it costing $1, and all of it loaded with useful connectors, which I've since harvested. Thus most of my connector angst has now been resolved.

On the lighting side of things, I went to order LED's for the main emitter array and my eyeballs jumped out of my head. I had spec'd out 6 Luxeon III stars and 18 Luxeon V stars, only to discover that the Luxeon V stars cost like $20 each! Jesus christ on a stick! Using Luxeon V stars we were looking at LED prices for the emitter array in excess of $400. No good. I had originally spec'd out Luxeon K2's but wasn't able to use them due to the fact that Lumileds doesn't currently sell them as stars (i.e. with heatsinking circuitboards pre-attached). With that wake up call, I went back to my spreadsheet and my spec sheets and re-ran my cost-benefit calculation to see what our options were. I also investigated the LED offerings from Cree, whom the flashlight nerds on candlepowerforums seem all a-gush about. Also, I found another vendor, LedSupply which sells Luxeon K2 stars for a reasonable price. I put all the possibilities together and searched for the solution with the highest lumens/$ ratio. The final solution was using 6 Luxeon III Red stars, 12 Luxeon K2 Green stars, and 6 Luxeon K2 Blue stars. The Cree units were pretty cool and typically a little more efficient, but Luxeon won out on cost. I've also added 3 Luxeon III Royal Blue stars into the specs. The Royal Blues are fairly similar to what other vendors call UV LED's and apparently they'll make some things floresce much like a blacklight would. I think this would be a cool effect for Zoe to have and add to the ethereal surrealism. (mind you with only 3 LED's, the effect probably won't be very strong but hopefully it'll add a sprinkle of fairy dust to the effect).

I placed orders for all of these LED's today. Grand total, about $220.

Also, with the new LED specs, I've calculated that we're going to need 22 TI TLC5940 chips to drive them. At roughly $5 a chip, that's going to add another $100 to the costs. However, TI will send you 5 of the chips for free (as a sample) to basically anybody who asks, and I registered with their samples site and ordered my 5 free chips online today. Also, if I drive the greens as 6 pairs of 2, that will cut down on the number of TLC5940's needed by 5 chips, which I may end up doing.

gads, it's 3am. Bedtime.

Tues 29 May, 2007

Another attempt. This design breaks out the control part of the board with the driver part. The driver part will be a seperate board.

This is just the control board.

Thur 24 May, 2007

Well, it's a start. EagleCAD is somewhat of a bitch to learn but I'm making headway. Here's a first attempt at the caustics generator board:

Circuits are pretty.

(note this board is no where near done, and already I see several problems with it. This is simply a snapshot of where I stopped tonight).

I still need to write out a response to Adam's email and also write up specs that Art requested, however I won't be able to do that tonight.

Sun 13 May, 2007

Some notes on equipment.

The 12VDC power supply arrived in the mail this week. It looks well used but I've hooked it up and it seems to work fine. I haven't figured out how to test it at its rated capacity since creating a 1500 Watt load is fairly perilous. I tried hooking it up to some heater coils but since those coils were designed to run at 120V instead of 12V, not a lot happened.

I also received the water pump in the mail but it appears that the dark side of shopping on eBay asserted itself and the thing was dead on arrival. It'll spin up when hooked to power but it won't move a drop of water. Hopefully the ebay person I bought it for will let me refund or exchange it or something but I have a feeling they're going to screw me on this one.

In the category of insult to injury, and still talking about the water pump, I went down to Harbor Freight and they had the exact same model of pump, new, on sale, for like 7 bucks more than I paid for the used broken one. Guh. I felt a bit retarded for having bothered with eBay at all but I bought a new pump and brought it home.

The new pump works fine. I bought some tubing and hooked it all up and was able to verify that the pump would move water more than 16 ft off the ground. I wasn't able to reach higher so that's the as far as I could test but I don't forsee a problem. The pump consumes about 65 watts of power and I used it to test the DC power supply, which powered it without difficulty. Also, I found about 20ft of thin tubing for free at Seattle University's surplus sale so I brought that home. It's a little stiff and the inner diameter is so small that water flow is quite low but we can potentially use this to run up Zoe's leg and into the water dish.

Fri 05 May, 2007

I've ordered a batch of ATMega168's and Darlington arrays from DigiKey. The Darlingtons will be used to drive the solenoids and the ATMega168's will be used as the standadrd MCU of choice wherever possible since that's what Adam is targeting with his CAN library.

In other news, I've bought one of the Cosel PBA1500F-12 1500Watt 12V power supplies from EBay. This is what will convert the power from the generator to 12V we need for our electronics. Total cost was about $88 after shipping and it should arrive sometime next week.

I've also eBayed us a water pump for $23 which should do the trick. It's a used one of these. They're like $40 new not including tax or shipping (didn't see one of these at Harbor Freight in town last time I went to look) so that's not a bad deal. I doubt it'll be heavily used.

Thu 03 May, 2007

For some time there was something nagging at the back of my head about the design of the main LED array. It had so very much blue.. and yet.. arrays I've seen other people design... not so much. Why did mine have so much? Well, something finally clicked in my head today and I went back over my old calculations, challenging some of the assumptions I had made. Sure enough, I found an error, and sure enough, it was with the blue.

So I re-ran all my calcs and layout programs and, lo' and behold, I had grossly overprovisioned on the blue. The new LED count is now 29, down from 47. w00t. That should make things a bit easier.

I'll be updating the array page shortly.

Sat 18 Mar, 2007

I put together a revised Zoe mockup with the helix-like leg design. Don't ask how long it took :P. Fortunately however, I'm posing the sketchup source so anybody can re-use it as they please. I made the leg into a component so it should be quite easy to re-use.

(I just now realized that I didn't put the cross-beam supports in. oops. oh well)

Also, Bez, you totally called it... I was reading the solenoid spec sheet wrong and indeed the solenoid is its strongest when the push-side of the shaft is fully extended. Good call.

I'd like to proceed with getting some caustics generator mounts put together but I'm a little hung up on how the top part of the dish is going to look (which has a direct impact on how we mount the caustics generators)

Wed 7 Mar, 2007

Solenoids are here!!@#$!@

The solenoids came today and I have to admit that they're a little smaller than I imagined, which is funny considering that I've been working with scale models of them. Still, I had a little time to hook one up and they're jammin' little punchers that kick like tiny mules. By holding one by hand over a caustics generator I was able to get some rather solid ripples at speeds of upwards of 30hz, which is quite a bit better than the motor+cam prototype fared. And this is not factoring in the fact that most of the buggers force was going into my hand and not the pingpong ball. I think with some proper solid mounts and better springs, we're going to be able to get a very workable range of ripple frequencies out of these.

I did notice that they had a tendency towards getting hot rather quickly, however I attribute that to the fact that I really wasn't using them properly or had proper mounts for them which caused them to stall quite a bit (stalling == muchos heat).

Demo at this thursday's meeting!

Sun 4 Mar, 2007

Ok, yeah, Google Sketch-Up is cool. I spent a fair chunk of the day trying to figure out a design for mounting the caustics generators in the dish. Came up with a few basic ideas, could really use some feedback.

Sat 3 Mar, 2007

Ok, so there was a snaffu with the the solenoid order however it should be cleared up now and the solenoids should ship on Monday. See the CausticsGenerators page for more info.

Also, Google SketchUp? fucking rocks. This diagram only took me a couple hours with no prior experience.

I've been doing a lot of work trying to calculate the strength and type of return spring we need to balance the solenoids.

Also, I've been looking around for where to buy springs and have found that they can be suprisingly espensive. A number of places will sell them for like $5-$6 for a little 0.5" spring. Yeah, just one. Also I haven't been finding too many springs that need less than a pound of force to compress them an inch.

Sun 25 Feb, 2007

Got 8 ping-pong balls center-drilled. Took me a little while to figure out how drill straight through the center of a sphere but I figured it out and made a little wood template that will make it easier to do in the future.

Hacked together a circuit to drive the motor at voltages up to 24V using only 5 V of control. This is essential for controlling the motor via a microcontroller. I almost hooked the motor up to the microcontroller but ended up using my function generator to emulate PWM to drive the motor. It's faster to turn knobs than it is to write code.

Playing around with this drive circuit at different PWM rates, I was able to get the wave-generation speed pretty low. However, I did have some significant problems doing this reliably and avoiding stalls. The motor needs a lot more juice to get going initially than it does to keep moving. One of the problems I think is that all of the springs I have been using are much stronger than they need to be, and this is forcing the motor to do a lot more work than it otherwise needs to. I tried heating, stretching, and cutting the springs to soften them up a bit and, while that worked, heating oxidized the fuck out of the springs and made them brittle. I don't think that's the way to go (and researching around on the internet supports that conclusion). I think I'm going to need to find some volute (conical-shaped) springs which are softer than any of the various cylindrical springs I've found in the 5 or 6 different hardware stores I've been to looking for springs. I'm thinking mailorder... soon as I can find out who in hell the amazon.com of springs is.

In general I think my experiences today have somewhat weakened my sense that the cam-design is the way to go. I've still been shopping around for solenoids and still the only ones that I've found that seem to fit the bill are going to run $20-$30 a pop. 16 of those are going to cost about as much as the water lens. The most promising place for solenoids I've seen so far has been Planet Engineers.

However, one thing I should keep in mind is that I'm using el-cheapo motors. Better motors with more torque and lower max RPMs might work much better.

---

Jesus H. Christ in a chicken basket Allelectronics.com has just gotten in some minature 12V push-type solenoids which sound damn near close to perfect from what I can tell and they're charging only $3.25 for them. Having no idea how long they're going to have these in stock, I've just ordered 18 of them, 2nd day air.

w00t

Sat 24 Feb, 2007

Went back to the machine shop again today. Got some time on the drill press and drilled holes for another 4 or so caustics generator bases (usable for either motor or solenoid-based design). Also spent some time on the lathe trying to de-thread sections of threaded #6-32 rod however this wasn't met with a lot of success. I had considerable difficulty keeping the short, thin rod in place enough for the cutting bit to actually cut into it and not simply push it away. Ultimately I gave up using the cutting bit and just resorted to using a file to file down the threads around the areas on the shaft that will be making contact with the support frame. I might of had more success if I wasn't a total n00b when it comes to lathes and a significant amount of time was spent figuring out how to use the thing.

Everybody else in the group was busy working on a weather-balloon project where they're going to send up a helium filled balloon into the upper atmosphere equipped with a microcontroller board, digital camera, GPS unit, and packet radio to aid in recovery. Once the balloon finally pops, they'll recover the camera and check out the pretty pictures on the compact flash card. It's destined for Make magazine I'm sure seeing as one of the Make staff is driving it. Cool crowd.

Once back home I played with a now more refined caustics generator prototype (motor based) some more. The refinements I made now let the motor run at a slower speed which greatly improves the quality of the waves. On that subject, I've found something that's fairly obvious when you think about it but deserves mention: when it comes to making waves on the surface of water, the maximum frequency you can create waves is inversely proportional to the amplitude of the waves. In other words, you can make big waves but you can only do it slowly, or you can make extremely fast feathery type waves but they have to be extremely small. Water will only let you push it so far. Push it beyond its limits and, for instance, try to make too big of waves too quickly, and you'll notice undesirable effects such as your wave generator pulling water up and throwing it out, muddying your once clean waves and making a mess.

On that subject I've had a thought that relates to the solenoid vs motor based designs. With motor based, you are limited to a fixed wave amplitude. How you build the cam determines the amplitude of your waves and that's the end of the story. With a solenoid however, there's the potential that you can influence the depth of your waves provided you have a well-balanced spring setup. Basically the longer the current pulse you send to the solenoid, the further the plunger will extend and the deeper the wave. If you limit the duration of your pulses (in software), you can limit the size of your waves, thus allowing you more flexibility. If this is a desirable trait when dealing with a network of generators all interfering with each other, I'm not sure.

Also I hacked together a very quick-dirty solution to measure the frequency of the caustics generator prototype and found that the sweet-spot in terms of cool, clean ripply waves was around the neighborhood of 10hz.

Sun 18 Feb, 2007

I notice that it's been a while since I've written anything on the wiki or have been vocal in email. I don't want to give anyone the sense that I've checked out of this project or have been too busy with other things to take it seriously. Quite the opposite. Well, I have been busy with other things, this is true, but the obsessive psyphozoa fires have been burning as bright as ever and I've been devoting a lot of so-far unreported energy to the project. Here's a quick summary of things I've been playing with:

LED Calculations

I've been refining the programs I've written for calculating the number and type of LED's that we need as well as placing them in a pattern that optimizes color coverage while minimizing diameter. Both of these things have come a long way from the data I posted to the site a few weeks ago. However, I do have to be honest with myself and admit that I've passed the point of diminishing returns and moved on into the realm of nerdish obsession with respect to trying to get them 'perfect' and really need to just pump out some final results and move on to other things.

LEDs Purchased

I've purchased and received a batch of Luxeon LED's of various types and colors. I got 4 each of Red, Green, Blue K2's and some Luxeon III's as well. This will let me start work on desiging a high-current control solution to drive them. The K2's were packaged with some rather extreme anti-humidity measures which have me scratching my head and wondering if they are going to to work in a relatively-high humidity environment. I'm going to have to research that. The luxeon III's on the other hand were packaged normally.

Machine Shop

Now this is big news, I've managed to get access to a machine shop on saturdays and have spent the last two saturday evenings in this shop tinkering. The shop is part of a community effort of likeminded hax0rs, geeks and nerds (some burners) who are pretty hardcore (previous projects of theirs include things like RFID scanners, self-balancing robots, a CNC lathe, etc) and a number of their projects have been written up in Make magazine I believe. They're calling it the "Public Nerd Area" and they've been pretty welcoming of me. This place/community has an impressive array of tools and expertise. On the tools front, some of the things they have inclue: drill press, bandsaw, chop saw, metal lathe (with homebrew CNC), compressor, decent array of electronics tools, huge array of hand tools. Anyways, I could go on but not without completely fawning all over myself. Suffice to say, I've long dreamed of exactly this sort of tool co-op community and long assumed that the only way I was going to be a part of one was to start it myself, and now I've found one.

Caustics Generator

I've got a working prototype of the motor-and-cam design for the caustics generator. Here's a pic:

• Initial Prototype of Cam-based Caustics Generator:
  

The cam is a disk of Delrin plastic and is attached to the motor shaft by friction alone (this works suprisingly well). The cam drives a shaft (a bit of 6-32 threaded rod) through a couple of steel angle brackets (found at home depot) and into the ping pong ball (drilled and attached to the rod using nylon nuts and washers). A spring supplies a return force for the shaft.

This is a very early prototype and is horribly machined (the shaft holes aren't even aligned!) however it does seem to get the job done and, once refined, promises to be a cheap as well as fast/easy to build causics solution. However it still remains to be seen if the motor speed can be reliably controlled enough to get the range of patterns we want. If it can't then we'll probably need to go the solenoid route.

One gotcha I've already found is that, if you run the motor too fast, the surface of the ping-pong ball will actually pick up the water and throw it out in the form of little droplets. These droplets then land and mess up the caustics pattern somewhat.

I'll have a simple demo ready for this Thursday's meeting.

Ok, i'm outta gas. I'm having my ass kicked by a cold and need to get some rest.