For a while now I have been talking about hooking up the Emotiv headset brain computer interface (http://emotiv.com/) to the Pyrocardium and this Thursday we managed to pull it off.

What you see in this video is me in a headset actively trying to turn on the fire with my mind. You can tell I am trying to turn on the fire by when I am holding out my hand. I find using a gesture generally helps me focus on a very precise thought. The thought to turn on fire in this case was to imagine the look and sound of the fire and “will” the fire into existence using psychic energy coming out of my forehead. I realize the part about the forehead energy sounds a little flaky but the neat thing is that it works quantitatively better.

To get to this point, I trained the system to recognize the thought pattern for setting things on fire with my mind. This was done using software Emotiv distributes to developers. I took the output of this software, interpreted it some, and then wrote to the USB controller board that runs the Pyrocardium. And then fire happened!

A few other people (Wanda, Kurt, Zack, and Bash) were also able to try the headset and make fire with there thoughts. They seemed pretty pleased with the results. : )

Also, in the video is Zack spraying color fire stuff into the flame to turn it red. Neato!

Oh, and big thanks to Bash who brought the headset, software, and laptop that allowed all this to happen.

Ethan managed to turn fire a bright red color using strontium chloride salt dissolved in water. He dispensed the salt with an airbrush nozzle powered by a small air compressor and we tried on several flames including a large propane venturi flamethrower as shown in this video.

Hi-Def Version

A while back I posted about my first attempts to make a stethoscope amplifier as an input to PyroCardium.  This has worked well, though it has required some tweaking since then. First, it turned out that the resistor network used to power the electret mike and to bias the output signal to within the range of the single supply op-amp I’m using was too low resistance, and was substantially loading the relatively high impedance output of the electret. Secondly, getting a good seal between the microphone and the stethoscope tubing is critical for getting a decent signal out. Tweaking those things gave us a setup that worked pretty well for Alchemy and Maker Faire.

That said, there are some definite problems with our current stethoscope/microphone setup. First, it turns out it’s surprisingly hard to position the stethoscope on your chest to really pick up your heartbeat, especially as our current setup doesn’t have audio feedback so you can hear whether you’ve found it or not. Secondly, the stethoscope does a pretty good job of picking up ambient noise.  It works great it a quiet room but less well at a noisy, crowded event.  In fact I was able to get it to trigger on the beats of the music playing at Alchemy with just a little tweaking. This is great for some other things we have in mind for Priceless but less good for picking up your heartbeat.

So, we’d like to find a better system for Burning Man.  Ideally we’d like something where there’s no ambiguity about where to place the sensor and a sensor that robustly picks up your heartbeat, regardless of environmental noise, the fact that you’re running in place, or anything else.  So far we’ve had two ideas - use some kind of simple EKG system, or use a pulse oximeter.  This weekend I tried building this simple EKG circuit but met with no success whatsoever - all I could see at the output was 60 Hz line noise.  I’m not really sure why (the high common-mode rejection ratio of the instrumentation amp is supposed to take care of that) but I’m not sure I have the skill set to make it work.  I’m also not sure that this EKG system will be robust enough or easy enough to use on the playa, and there are also safety issues with hooking it up to a line powered system.

So, my current plan is to buy a cheap pulse oximeter and see if we can rip it apart and get it to interface to a PC.  If anyone out there reading this knows about pulse oximeters and how to do this let me know.  Also, if you’re good at analog design and want to make the ekg circuit work, that would be good too :).

When planning to fill a large enclosed chamber with hydrogen, oxygen and propane, and allow members of the public to voluntarily ignite the contents, we find it’s best to consider ways to decrease the likelihood of blasting out four foot by eight foot sheets of shattered polycarbonate at hundreds of feet per second. If you plan on igniting flammable gases for amazement and amusement (read: art), you really have to consider safety third, and certainly no lower than that.

We’ve taken a number of measures to reduce the risk of life and limb, the most important of which is the hydrogen chamber itself. There’s a lot less chance of someone burning their hand, face, neck, neck, chest or faux fur costume if they are physically separated from the fire by a plastic barrier.

But again we come back to the question: How do we avoid blowing up the chamber? The key is to understand what the maximal flow rate in cubic feet per minute (cfm) is of our bubble blowing devices, and the flash point of hydrogen. Hydrogen becomes flammable in air at concentrations of just 4%. So we built an air blower and duct system that creates positive pressure inside the chamber at roughly 50-100 times our hydrogen usage. In our case, our bubble machines will consume 3-4cfm of hydrogen. Therefore we have installed a blower system that will push around 400cfm of air into the chamber. This reduces the maximum hydrogen concentration in the chamber to less than 2%. In addition, we’ve mounted many nichrome ignitors (see Ben’s last post) to ignite any small pockets of hydrogen, either contained by or not contained by bubbles, before they can collect into a dangerous situation.

How would we know if a dangerous situation occurs before the gigantic explosion? We mount a detector on the wall of our chamber, at a level where we would want to shut down the project if gas somehow pooled from the ceiling down to the detector mount point. If the detector goes off, the project is shut down until the chamber returns to safe levels.

For PyroCardium to respond to people’s heartbeats, we need a way to electronically measure their heartbeat.   Since PyroCardium started life as a VU meter for music, the obvious way to do that was to hook up a stethoscope to a microphone.  Today I tried to do just that.

I had some cheap electret mikes leftover from an earlier project idea (a music responsive EL-wire cape that never made it off the drawing board), so I decided to do the obvious thing and jam one into the stethoscope tubing.  I then built a simple amplifier with a gain of about 100.

Hooking it up to the oscilloscope, I could see the trace move when I put the stethoscope on my chest, but I had no idea what it was responding to - if it was really my heartbeat, or if it was just noise.  It took me a while to come up with a way to figure out what was going on, but eventually I realized that I could hook it up to my stereo.  A few alligator clips later, I could hear my heartbeat booming out over my speakers.

I think our stethoscope amplifier is solved.

A couple of weeks ago, Erik did some great drawings of the hydrogen chamber in Rhino. Since then, we’ve been working to refine the design (uhh, we forgot a door… can we add one?) and build out the first level.  It was also cool to have Erik show us how he works in Rhino. The software combines the manipulation and viewing power of a 3d modeler like Maya with the accuracy and dimensioning capabilities of a CAD program. Perfect combination for this kind of work.

So far, first the six steel frames are welded, the aluminum for the surrounds is cut and drilled, and yesterday I got about half of the brackets welded on. It’s cool to see our physical work starting to look like the 3d model!

Those of you at our last few workdays will have heard me moaning about transients from our new solenoids causing the PyroCardium USB controller to trip off. The back story is that the 24V DC solenoids we had been using before turn out not to have seals compatible with propane. So I’d like to switch to some 110V AC solenoids I have which have propane compatible nitrile seals. I got 16 of these solenoids for free from my dad - they’d be about $800 new.

The problems began when I replaced the 24V solenoids with the 110V AC ones. As soon as I tried controlling them from the computer, the software crashed immediately. When you turn off a solenoid, the energy stored in the coil gives rise to a rapid increase in voltage across the inlets. This is called inductive kick. I have a varistor in place to dissipate this kick, but it doesn’t appear to be enough - my theory is that this transient voltage spike is getting coupled into the USB board and giving rise to errors. This theory is supported by the fact that adding a capacitor in parallel with the varistor, which should also help dissipate the transient, reduced the crash frequency (though it still crashes after about 5 or 10 minutes of cycling).

To fully isolate the USB board from the rest of the system, I’ve ordered some optoisolators. Optoisolators chips that electrically isolate one circuit from another by coupling them with light. They have an LED driven by the input coupled to a phototransitor that forms the output. I’ll have them on Saturday, and with them I’ll be able to fully isolate the USB board from the high power switching circuitry. Hopefully that will fix our problems.

In our latest workshop day, we got both The Hydrogen Economy V3 bubble machine prototype, and the 4-flame PyroCardium prototype working and ready for Maker’s Faire tryout day tomorrow at the Exploratorium. Then it was time for a little antics. Here’s a video of voice-controlled PyroCardium, held by Chris, exploding the bubbles or at least trying to.

The Hydrogen Bubble Machine V3 prototype is moving right along, and now we’ve attached two bubble wands to our central rotating plastic rod. I tried using different diameter wands and two needle valves to demonstrate differing flow rates. Next step Kurt’s fabbing our circuit boards for the motor controller.We had designed an ingenious little optical sensor with a veined wheel to indicate when the wands are in the up and down positions, but now we think this design can be simplified with a couple of microswitches and some stops mounted on the side panel. A screw on the axle will tap the microswitch and inform the microcontroller that the wands have reached the downward (dipping) position. The microcontroller will then reverse the motor until the other microswitch is tapped, when the wands will be in the upward (blowing) position. At this point the microcontroller will turn on the solenoid, releasing our hydrogen gas mixture and blowing bubbles for a certain (adjustable) time period. Then the microcontroller will repeat the process.I’ve mocked up this prototype using a three-way toggle switch and connecting the solenoid manually to a battery. Here’s a video showing the latest testing:

I know I should be helping get our Burning Man proposals in final shape, but I wanted to post some thoughts I’ve had on bubble ignition with lasers. For those of you who don’t know what I’m talking about, we’ve periodically discussed whether our hydrogen plus oxygen bubbles could be ignited remotely with a laser. I think it’s agreed that this would be pretty awesome. Unfortunately it turns out it’s probably quite hard.My initial thought along these lines was to use some kind of photochemical reaction along the lines of a demo I saw in my freshman chemistry class. If you shine blue light on a mixture of hydrogen and chlorine, it explodes, forming hydrochloric acid. You can see this in this video on youtube:This works because blue light photodissociates the molecular chlorine into atomic chlorine. The atomic chlorine is very reactive and reacts with the hydrogen in a chain reaction, forming HCl. Chlorine is photodissociated by UV - blue light in the 300 - 400 nm range.In principle you can do exactly the same thing with either the oxygen or hydrogen in our bubbles.Unfortunately, the wavelengths you need to photodissociate either oxygen or hydrogen are in the deep UV, and can only be generated by very expensive sources like fluorine excimer lasers. Even worse, these deep UV wavelengths are efficiently absorbed by the atmosphere. So direct photodissociation of H2 or O2 is probably out. One potential possibility is to dope our bubbles with something that can be efficiently photodissociated to spark the reaction (chlorine would work, but it’s pretty toxic) but this seems likely to be pretty complicated.There is another possibility: laser spark ignition. It turns out that if you focus enough laser power into a small spot, the resulting electric field is enough to ionize the air and create a spark. Once again, youtube has video of it (it’s amazing what you can find on youtube):This should ignite a bubble just as well as an electric spark (and in fact, there’s a lot of research into using laser spark ignition to replace spark plugs). There’s just a few problems. First, you need a lot of laser power. That video is using 200 mJ pulses that are probably a few nanoseconds long. That means the instantaneous power is on the order of 10 MW. The average power is only 4W, but you can imagine that if you happen to be in the way of one of those pulses you will be unhappy. So laser safety concerns preclude using this in an open space.Second, that laser is probably pretty expensive.Third, you have to have the beam come to a tight focus to get the spark. That means it’s not enough just to point the laser at the bubble, but you need to focus the spot into the bubble. That’s probably pretty hard to do.So, my initial ideas for laser ignition of bubbles seem not to be feasible … but I’ll keep thinking!