Cnidaria Hallucigenia

This is the first version of a jellyfish, with each of the colors lit. Since I made this one, I have finished one more, completed one wiring harness, and stripped the wires for two more. There will be five C. Hallucigenia when I'm done. I ordered six umbrellas, but the sixth one will be used as a sacrificial test subject to determine if there is an acceptable method to make them slighly translucent. I would prefer that the bodies of the jellyfish be translucent in order to diffuse light better. So far, the possibilities for doing this are window film that has a translucent design on it (so they can have a frost pattern or something on them) or matte-finish spray varnish. The film doesn't have adhesive, though, and the spray might dry up and flake off, hence the use of a sacrifical tester before full deployment.

Assuming all goes well this evening, I will have the LED wiring and some of the installation done by the middle of the week, and be all set to install the controller boards before the weekend. I anticipate that the jellyfish will not draw more than 1.5A at their maximum current, so the 7.2V 2200mAh RC car batteries will be more than enough to run them overnight. I'm tempted to build in an 8-hour cutoff, so that I can avoid having them run during the day and just go around turning them on at dusk.

Of course, if I do that, I'll want a costume and a magic wand to do it with, and that means building a reed switch and a micropower sleep mode into the system, and making a magic wand with a magnet, and so forth and so on until I'm standing on a mountain in Nepal with a yak leash in one hand and a can of shaving cream in the other.

I also got a slingshot and fishing reel. The slingshot will be used to shoot a weight tied to a fishing line over branches that I want to use to hang a jellyfish. It may be that fishing line is not sufficient for this task, but I think a high-test line will certainly hold. I intend to test the deployment system before heading into the woods, so I'll have a solid way of mounting them by then.

For the assembly of the other four jellyfish, I am going to assemble the light wiring and then attach it to the frame of the umbrella, rather than attaching the LED/resistor groups and then wiring them.

The picture to the left shows the individual LED boards. Each board has a RGB LED, current-limiting resistors, and enough spare holes to mount them to the umbrella.

The rightmost picture shows the umbrella opened, with 2/3 of the LEDs installed. The remaining 1/3, 8 LEDs, will be around the bottom edge of the umbrella with tendrils of fiber-optic cable trailing down from them. I made the tendrils a couple of nights ago, and will be putting them on the umbrella tonight.

Once the tendrils are in place, I'm going to wire them up with very fine wire-wrapping wire and power it up with one of the prototype boards. If that looks good, I'll be all set for mass producing the jellyfish bodies.

The cheap recorder modules took forever to arrive, and make a beeping sound when they start recording. This isn't acceptable for my purposes, so I got some slightly more expensive, but silent, recorders. The other ones can end up in something that only needs recording once, but then plays back multiple times.

I'm already having ideas for using the other modules. Hooking them to motion detectors would be amusing. I could have a simple story told in parts, and have each part trigger as a viewer passes the module. If the playback system is properly camouflaged, the story would be an ambient thing, without an obvious technological manifestation. Perhaps I'll call it "Rumors".

It looks like using small bits of stripboard to attach the LEDs to the umbrella ribs is the way and the light, a truth for the ages. I need another piece of stripboard, as I've run out, but the attachement goes fast. Pictures soon.

I'm planning to have falls of fiber optic tentacle at each rib end, so the overall form will be closer to a box jellyfish than a sea nettle. To my mind's eye, that still reads as a jellyfish. If the finished form doesn't suit, I can add a fringe of vinyl tentacles around the rim easily.

In the service of the fiber tentacles, I'm making a jig to strip the fiber. The plan is to finish the jig tomorrow, and finish the LED mounting by early next week.

I still haven't received the mass order of recorder modules. If they don't show up by this weekend, I'm going to have to get in touch with the shipper. Some delay is to be expected, as they are coming from China, but I need to verify the amp design as soon as possible.

PDF datasheet for the ISD1510, a common voice recording chip. Simple non-inverting amps in place of C4 and C5 in the schematic on page 10 of the datasheet will probably help. I was using a single amp with gain of 10^5, but I think a gain of 10 or 100 will probably be sufficient, as it is multiplied by the gain of the internal preamp.

Another possibility is to use a single-ended mic, ground the mic ref pin, and amplify the mic so that a normal voice at 5 feet has a 100mV peak-to-peak signal. This will need some mic output limiting.

One approach may be a logrithmic amplifier with a bipolar transdiode configuration. This will have lots of gain at low levels, but not so much at high levels, so it will pick up a whisper, but not amplify a shout too much. There is a schematic here. Gain of 10 is probably sufficient, as the mic capsule tends to have a 0.1-10mV output.

I got home yesterday (5/13) to find that parts had arrived. I have finals early next week, but by the middle of next week, I should have a final version body with working lights.

The audio circuit is still giving me a little trouble, but I think I have solutions. The voice recorder uses a balanced mic input for noise suppression and an automatic gain control to adjust for input volume. In order to use a high-gain mic preamp, I may have to either:

A) Adjust the AGC to be more sensitive to quieter sounds, thus allowing recording over a greater distance. The available parameters are the attack and decay times of the gain envelope, set by an RC circut outside the chip. I think I would want fast attack and slow decay. The datasheet for the IC doesn't really describe the effects of different values well.

B) Design a balanced mic preamp. This would be a little harder, but I suspect it's one of those two-op-amps-and-a-pinch-of-passives deals that a bit of googling will turn up. It won't need much gain, as the IC expects a mic signal in the millivolt range.

My first attempt at making the jellyfish body used vinyl shower curtains to make the bell, over a normal umbrella frame. I do not like this design. It's not deeply curved enough to really look enough like a jellyfish to me. The reason for the lack of curve is that the normal skin of an umbrella applies tension to pull it into a more deeply curved shape. The vinyl is heat-welded with a soldering iron, which I haven't gotten very good at yet, so the process yeilds a somewhat weak bond, which cannot apply the tension needed to curve the umbrella ribs properly.

Instead of making my own umbrella skin, I have elected to use clear "bubble" umbrellas. If they need to be translucent, I can apply privacy film for windows to them, which will help diffuse the light, and can be applied in patterns.

The glowing cube is a working prototype of the light control circuit. It was built as a decoration for my girlfriend's birthday party. The cube does not record and play voices, but it does use the same hardware and software as a earlier device that did.

These are prototypes of the boards that will go into the Cnidaria Hallucigenia. As is the way of prototypes, they have some errors

The one on the right has the PCB art applied backwards, so the parts had to be put on the trace side of the board for testing. It also used the wrong footprint for one of the driver chips, so it really didn't work at all. However, the voltage regulator and layout around the microcontroller worked just fine.

The one on the left is closer. The location of the PCB art and the footprint of the driver chip are correct. However, the driver is a current-sink driver, and the LEDs are common cathode, not common anode, so they really need a current-source driver. I ordered some common anode LEDs, but then decided that it would be simpler to roll my own current-source drivers using beefy power transistors as supply-side switches.

Being able to prototype my own boards means I can catch errors like this fairly quickly, instead of waiting on turnaround from a board fab.