Weirdscience.us

We are now actively seeking investment. Stay tuned.

Our plastic recycling unit currently requires 3 steps, which we hope to reduce to 2 steps by production time. Static electricity poses major dangers and difficulties, so we had to design, build, test and integrate an electrostatic eliminator. We also need to reduce the size of the unit for home use. Still on target to present to investors in late February.

I completed construction of the recycling unit prototype and it is currently under test. I will be looking closely at the results vs a multitude of design parameters so I can optimize the process.

I have begun construction of a thermoplastic recycling unit. Not only do I need this for or own experimentation, but I also plan to offerer it as an accessory to the 3D printer. Of course, this will allow people to process recyclable household items in addition to previously printed objects.

With a fully functional (although totally hacked together) CNC under my belt, I am ready to announce a project that has been 3 years in the making: a personal, professional-grade 3D printer for the average person. Building my own CNC from scratch has not only given me the necessary skill set for this project, it has also given me a working tool in which to machine parts for the prototype.

Specs (basic series):
Work area: 8.5″ x 11″ x 6″
Work media: A large variety of thermoplastics, including PP, PE, HDPE, UHMW, PVC, ABS, PLA.
Spatial Resolution: 40 microns
Supporting structures/material: Unnecessary.
Surface finish: Sandpapered/powdery feel
Material Strength: Near 100% density, suitable for end product with no tendency to delaminate.

Price: $600

First prototype for potential investors will be ready in Feb 2013.

The modified drill press vise slides had to go, which were replaced with keyboard drawer slides mounted on steel plate back-planes. I decided not to go with dovetail slides due to expense involved. The drives nut were made from HDPE, tapped with a 5/16″ – 18 thread.  The drive screws are simply 5/16″ – 18 threaded rods, fixed to HDPE end-bearing blocks with ABEC-7 roller blade bearings serving as the actual thrust bearings. The stepper motors (Minebea PM55l-048-HP69) were harvested from old, discarded HP Deskjet printers and modified to run in bipolar mode. They are coupled to the drive screw by home-made helical flexure shaft couplers made out of nylon spacers. I also opted to use closed loop servo control (instead of open loop control) using the optical encoder assemblies harvested from the same printers used for the stepper motors.

There are many things I would have done differently with this CNC. It has proven to be somewhat high maintenance. The keyboard drawer slide bearings and optical encoders are exposed and need periodic cleaning. Sometimes this means having to completely dismantle and reassemble the slides. I have already gone through a few shaft couplers on the x-axis due to the high degree of misalignment of the motor and shaft. Although the machine is fairly rigid, it is only capable of milling wood, plastic and soft metals such as Aluminum. On the bright side, the closed loop servo system is able to maintain around 0.005″ tolerance throughout its workspace.

The next CNC mill I make will employ parallel kinematics for greater rigidity and flexure bearings for virtually maintenance free operation. By parallel kinematics, I don’t mean the typical Delta or Stewart configuration- there will be flexures and lever reduction that will increase tolerance by nearly an order of magnitude. I will also give it an actual spindle that accepts standard collets, as opposed to using a dremel tool. Workspace will be about the same, 6″ x 6″ x 3.5″

More info to follow.

I found a minor flaw in the CNC’s control board. After a little troubleshooting, I found that I accidentally swapped pins 16 (VCC1) and 8 (VCC2) for each axis. It’s going to require a bit of work to desolder and resolder the affected areas of the protoboard.

One time pads are perfectly secure in theory; however in practical use, simple mistakes can render a one time pad completely insecure. I asked a friend to send me a few short messages encrypted with the same one time pad for me to attempt to decode.

ETEMKSASXTNQVBTFTY
TMIUZRSDJJBRATIYRM
MQMBOJMTCTNQRLOINM

I wrote a genetic algorithm to mount a dictionary attack against the cipher, giving the following results in about 10 minutes:

HOW LONG DID THIS TAKE
WHAT DM YOU THINK IT IS
PLEASE SEND THE CODES

And apparently the one time pad was not random either (WE HAVE TO OPTIMIZE IT)

The genetic algorithm was kind of a sledgehammer approach to solving this problem. When I find the time, I would love to delve deeper into the mathematics necessary for a faster, more analytical approach.

1. CNC Mill version 2:
Dovetail slides, Open loop control, Digital Read Out

2. Hardware random number generator. (probably shot noise based)

3. Software Defined Radio with Integrated GPS for time stamping samples from the ADC.

4. Total website overhaul.

Work on the CNC mill has ground to a slow crawl, but it has basic functionality.  The next step is to improve the quality of the linear slides, as the modified drill press slides really are not up for any other task except perhaps making parts for the next CNC mill in the works.  I am really disappointed with the results, but it was a pretty good learning experience.  I can definitely see why most mills use things like dovetail slides, open-loop control and stepper motors.