Recently in electro-mechanical Category

Jon and I will be down south for the day--displaying a ruby programmed mini shaketable, which demonstrates the principles of our larger shake house exhibit at the academy.  The mini mars rover will also make an appearance.  Both of these units can be potentially programmed by students.  

This rover made it's debut at 2013's Robot Block Party, and as of yet, has no official name.  It's being developed in conjunction with the CAS Department of Education, and CAS AVEE.  The rover, when complete, will be 100% open-source and open-hardware.  It will hopefully be available as a California Academy of Sciences classroom kit in the fall.

Students use the NES controller to input navigational commands.  After a transmission delay, the rover enacts the commands, then transmits atmospheric conditions to the controller.  In this manner, students can then plot out their environment and crunch some numbers.  For those of you that remember dos, its kinda like logo/turtle!

The rover prototype is made of a bare-bones construction, via arduino and xbee.  The controller is 3D printed and currently has some NES guts in it.  When complete, both the rover and controller will have open-source schematics & layout--stl files as well.  

here's some coverage and a photo by James Martin of cnet. 

This is a 7 digit, 7-segmented display, utilizing several TL constant sink drivers, and the parallax propeller 32 bit, 8-core MCU.  There's a TL driver for every common anode digit; communication is handled via SPI to the cascaded shift registers.  The prototype is 100% wire wrapped, and surprisingly fast!  

Originally we looked into using the maxim 7219, which was a single IC to drive all the segments.  That particular IC is limited to 5v and something like a 60ma sink, which would never drive the 14.4 v, 100ma segments!  There is a way to do it, but the transistor array and/or auxiliary maxim IC's required felt component heavy.  One fun note on the 7219:  it works on POV!

The Propeller MCU was chosen arbitrarily; another project will require these chipsets, so it seemed like a good idea to brush up on 'em.  They are very reliable and incredibly versatile;  parallax sells a kit to make videogame systems out of them!  The compiler crashed on XP, but worked fine on windows 7.  The language itself, is a bit clunky, but the 8 "cogs", and 80mhz speed make up for it!

Lastly, this device will count waterflow, but only once the PCB is back!

Installed and operating, this device is live!  It only needs a darker tint on the glass.

Here's some ktsf video with it.

One evening a friend asked if i'd be attending a musical event.  I guffawed and said, "of course not!"  The next day, I checked the site and noticed that my name had been attached to the event as someone who would: operate a robot through the catalyst of kinect.  Well, I hadn't recalled "officially" signing up for this, but with a week to go, proceeded in a focused manner.

The first step was finding Ryan Challinor's synapse for OS X 10.6.  Synapse takes the kinect and provides UDP output/input.  This includes a cool monitor window, as well as skeleton tracking.  The only hard part part for me, was upgrading to OS X snow leopard, which I'd resisted due to the number of "wonky industrial" programs I have on my machine.  

After that, it was a short manner of using the udpsend/receive objects to port everything into MAX/msp.  With that done, the sky is the limit.  But for my purposes, I just ensured that I could track the skeleton based on hits, or by "constant tracking" (in which you must prompt for the kinect data every two seconds or so).  i'll attach the patch for those interested.

The robot itself is nothing fantastic.  Its a simple bot built off of an arduino pro 5V, sparkfun motor controller, and an 802.11b based xbee for data tx/rx, and a super cheap 2.4ghz wireless camera with audio.  It beeps too, and has had a few varieties of sensor interfaced :)

In short order, the Kinect outputs data to MAX/MSP, which then uses the serial object to access the CPU's xbee, which transmits data back and forth with the xbee modem on the rover.  The rover arduino interprets the commands, and then moves.  In this way, kinect gestures could be used to control the robot!  Also, you can see where you're going, via the rover's wireless camera.

Rather, I broke it by violating a simple rule:  ALWAYS USE A PROTECTIVE DIODE!  So with everything broken, the rover would not work for the evening.  Instead I augmented the patch so that gestures would play a midi piano; stepping right increased the scale, stepping left decreased the scale.  The nice bit was that the users got to see the skeleton version of themselves.  The sad thing, is that I found out how shotty the midi interface on my sampler is (as I tried to do this first and could only get sporadic results!)

The attached patch will take synapse data and make it more usable in the max/msp environment.  It's an earlier version, and is certainly ugly.  You'll need to program the col object with the lines of text posted after the patch. 

-------------------------------------------

joints.txt:

Based on a rough sketch, i was charged with rotating an 18" diameter disc of unknown weight, for a continuous period of time, at an indefinite RPM.  Everything was to sit within a base, like a glorified flower pot.  The sketch also indicated that the shaft was load bearing, and, that the disc was to rotate upon plastic.  These problems were solved with a lazy susan.  Oh and it rotates a snake skeleton.

Next was the motor.  I prototyped this by spinning a large spool of 18 gauge copper with a salvaged 24v DC gearmotor (taken from an old vending machine).  There was a certain amount of noise from the planetary gearbox, but nothing which was a show stopper.  Next the drivetrain shaft, couplers and a new continuous-duty motor were ordered.  At 12V, 1/100 HP, and 25 lbs of torque, this seemed like a viable, and very cheap solution for rotational force.  I added a homemade 30v motor controller for PWM speed control, based off of 555 timing circuits, and was ready to go.

Then the disc/cylinder showed up.  It was far larger and heavier than I expected, and completely acrylic.  I fabricated a flange mount by boring out a steel collar and tapping it into a heavy 4"x3/4" disc of acrylic.  This was then afixed by a peer to the disc.  Meanwhile I made a barbaric motor mount from wood and aluminum.  But this whole time, i never knew the shaft length; everything was arriving just in time, and without proper CADS!  So once all the pieces were here, the shaft was trimmed and everything was installed into the base.  I plugged it in and was driven mad by the noise.  absolutely mad.  It didn't matter that it worked.  The whiiizzz-burr sound of the planetary gearbox under load was too nerve-racking.  I stuffed the base with sound dampering foam, but, to no success.

This is when I upgraded from the cheap $40 DC gearmotor, to a $300 single-phase AC gearmotor, 15 RPM @ 1/20 HP.  Yup, a significant upgrade.  Of course by the time the motor was available for pick-up, I had SIX HOURS TO FINISH!  I grabbed the motor, wired it up, re-trimmed the shaft on the horizontal bandsaw, and put together a new fast-frame aluminum motor mount.  The equipment was installed, and, perfect (for now)!  The blue illumination was then installed via these uber nifty customizable RGB 75 LED adhesive strips.  Deadline accomplished.

But though silent, the 15 RPM motor was deemed a bit too fast.  I'd chosen it since it was reverseable, and indicated on the website that the RPMs could be throttled down via electronic speed control.  But, what did I find in the manual:  Do not use an electronic adjustable speed control device with this gearmotor.

Well, i was angry at that point, but temporarily satisfied with at least having gotten most of the job done.  The next day, upon a suggestion, we throttled the voltage down using a Variac.  At 30v, the speed slowed down.  But at 30v the starter capacitor would never charge up to kick the motor into operation.  oh well.  That is the feeling of failing.  And now I have a 90V DC, direct drive motor, with speed controller on the way.

TL;DR:  DEMAND SPECIFICATIONS, AND DON'T CHOOSE CHEAP!

Originally we were tasked with automating a DMX board, and syncing some of its functions with audio.  There are literally zillions of ways of doing this, especially if your pockets are deep.  One option was to use a Gilderfluke or Alcorn-Mcbride product-- a programmable show controller, and interfacing it to an expensive lightboard like ETC or Horizon.  However, we had the hardware lying around to change a mac mini, into a full show control system with lightboard function and gui. This is nothing groundbreaking, but it's still fun.  Most of all, we didn't have to source a lightboard.

The i/o interface was the Phidgets 0/16/16.  This was convenient since it was usb based and provided a total of 16 inputs and 16 outputs, all opto-isolated.

The DMX interface was the Entec DMX usb pro, utilizing a purchased copy of dmxusbpro by Olaf Matthes. 

Everything was then tied together in the max/msp programming environment.  The "under the hood" was covered up with a nice interface for the general user.  The prize-winning max object on this one was "col."  It essentially allows a user to program cues via text files (which is nice if you are programming 12 cues at a time to a fixture), without knowing a thing about max/msp.

In the end, for a little under $1000 we have a mini show controller with these features:

-control 1 DMX universe, 512 channels, via GUI.

-trigger events on a schedule for complete automation

-16 digital triggers

-16 digital outputs

-audio and event playback via trigger

-programmable and expandable cues/events

-diagnostic panel (power, status, connection, etc)

-nifty interface

It was mostly a rush job, and a lot of sweat was poured when nothing worked, at all.  The manufacturer's manual for one of our DMX devices, was written so that a lot of guestamation had to be used.  A contractor had also done some funny stuff with dmx to microplex adapters, ensuring that failure was imminent. In the end it was a success.  if i had more time, i would have added a log feature and terminal access.  The "exciting" bit is that the entire system can be tweaked to interface with shell scripts.