The text of the logo is recognizable even in water. With glycerin I think it will be even nicer looking.
I’m actually cheating by rendering text as a bitmap image and then using software to load the bitmap into the arduino. So really, this is a video that demonstrates my ability to display arbitrary graphics, not actually generate text. (But I do plan on embedding a simple font table in the arduino so that the arduino can just render generic text.)
Now that I have the electro-mechanical systems working, I am focusing on finishing the trim pieces for the frame that will hide the electronics and top of the tubes.
This mostly consisted of notching a few slots in the pieces of plywood for the aluminum bottom pieces and cables to go through, and a day’s worth of work with a spray gun.
I had initially planned on using bright aluminum window screen mesh in a half tube shape over the top of the upright acrylic tubes, both to keep bugs out, as well as to give a nice visual appearance. But once I tried it out, I didn’t like the appearance, and couldn’t think of a good way to mount the ends of the screen cleanly. I am now considering using the old standby PVC pipe (probably painted some solid color with plastic spray paint). The other option is to buy a large clear acrylic tube (square or round) and use that in the same place. This would allow people to see the tops of the tubes (perhaps too well…it’s too bad McMaster doesn’t sell frosted acrylic tubes…) This is my current prototype:
Because the air hose is connected to the center of each square acrylic tube, the 60 LED lights on the bubble display are mounted almost between tubes, very close to the left edge of each tube. This means that a particular LED contributes light to both it’s own tube, and the one to it’s left. On the far right side of the display this makes mounting the first LED easy, as it is well away from the outside aluminum support upright. However, on the far left side of the array, the hole for the LED is right up against the aluminum support upright. The LED’s have a square circuit board below them that makes the entire unit wider than just the LED, and it won’t fit.
I knew this would be an issue from the beginning, and had designed my way around it using a bit of fiber optics. Of course, the far left tube is slightly dimmer than the others…but then again, so is the far right tube (which only has one LED).
Yes, I finally have all sixty tubes mounted and working. I still need to swap out two or three of the under-performing motors, and calibrate all motors, and do a lot of cosmetic finishing work (cover pieces for the top and bottom), and some software tweaks….but it’s finished enough to actually work!
Here is a video of 22 tubes running with water:
Now for the bad news….22 tubes actually measure 22 1/8″ wide. (apparently each tube is actually 0.00568 inches larger than 1″) This means my entire display would be 3/8″ too narrow, which just won’t work with the tolerances I have designed into my tube holder. That is why I only have 22 tubes in this video….the 23rd tube was just far enough off to not fit into it’s proper hole.
I fixed it by cutting the tube holder into 3 pieces and mounting them with a 1/4″ gaps between them.
I also had to space the end pieces of aluminum farther out, which meant I had to drill 4 new holes to mount the side brackets to the bottom board 1/2″ farther apart. Luckily the T-Sloted aluminum mounting system allowed me to easily move my L-Brackets up slightly to hit the new holes.
Unfortunately, I mounted the Shift MOSFET circuit boards on TOP of the screw heads holding the tube support board, so I had to take them all off to re-position the tube support board. Annoying, but not the end of the world.
So, after a bit of additional work, I now have all 60 tubes working.
All sixty RGB LED’s and air pumps working under computational control. Next up, physical assembly.
I use a scribbler robot with an IPRE fluke for my classes. On Ubuntu, I have a funky situation where the fluke does not work correctly the very first time I connect to it. Dan Walker, of BetterBots has captured all of the bluetooth data and says that Ubuntu is sending “AT+GCAP” data (AT command for a modem to “Request Complete Capabilities List”) to the fluke when first connected. This command gets the fluke into an unstable state and make it not work until rebooted (a possible problem with the fluke…but really, should custom hardware have to anticipate receiving random AT commands?)
The extra data doesn’t happen on subsequent connections to the same /dev/rfcomm* port until the laptop is rebooted. We suspect that the network manager may be the culprit. (I stay logged in until rebooting, so perhaps if a user logged out that would also trigger the network manager to re-probe?)
Thanks to Doug Harms for the solution…this is the ModemManager service probing all bluetooth serial ports. To disable it you can issue the following command (all on one line…):
sudo mv /usr/share/dbus-1/system-services/org.freedesktop.ModemManager.service
/usr/share/dbus-1/system-services/org.freedesktop.ModemManager.service.disabled
Here is a video of the bubble display base with 24 (of 60 total) motors connected running a test pattern. Next up….lots more soldering. I’m very glad I was able to find strings of 20 RGB leds already connected!
If you have cast iron drain (waste) pipes they are likely joined with a lead & Oakum seal. Oakum is a hemp material coated in tar, it is placed into the joint to seal it (Oakum expands when it gets wet to seal the joint). The Oakum is held in place (and the pipe joint is given some strength) by casting a bead of lead about 1″ thick around the pipe inside the hub/socket. In this picture the cast iron pipe hub is on the left.
It is possible to remove the lead and oakum. After you have gotten all the lead out and cleaned up the hub, you can place a donut (also known as a compression joint seal) in the hub of the cast iron pipe to adapt it to PVC. (Note that donuts are only usable on non-pressurized pipes, such as waste/drain lines.)
I have manually cleaned out a lead & oakum cast iron joint. It sucks. My recommendation is to cut the pipe somewhere other than at the lead & oakum joint and then use a Flexible Coupling to join it to your PVC. Yes, this introduces discontinuities in the waste water flow, and gunk may collect and build up at these edges, possibly leading to plugged pipes in the future….but that is small price to pay for avoiding having to manually unseal a lead & Oakum joint. For the gory details of how to manually disconnect and clean out a lead & Oakum joint, read on…
Continue reading
Just because the black iron pipe is near the plumbing section at the hardware store doesn’t mean that you should substitute it for a bronze pipe nipple that attaches to a gate valve for a dishwasher. Iron and water don’t mix. Well, actually, they do mix very well, leading to rust and a leaking pipe, and rotting wood. Below is a picture of the before (incorrect iron pipe) and after (shiny new bronze pipe nipple). Click to zoom into the corroded mess that used to be the bottom of the pipe….
As it turns out, I could have actually re-used the dishwasher connector hose, but that corrosion looked so bad I just bought an entire new kit.
Of course, if I had seen this gem BEFORE finding the leak and having to move the dishwasher, I probably would have double-checked all of the water fittings at that point. Click to zoom in on the sacrificial-vice-grips-as-pipe-anchor in all it’s glory. (I actually couldn’t find a good practical reason to replace this….as it’s a righteous hack…but a hack it is…)
