Phlegmatic prototyping

I'm a creative soft- and hardware engineer from Hamburg, Germany. These are my private projects. I'm available for contract work at www.thinventions.de

[phlegmatic-prototyping]

This is a board for soldering workshops and challenges.

It has 8 ‘levels’ with increasing difficulty. The first 4 levels are through-hole and the last 4 are SMD with decreasing size down to 0603 components. This board can also be used for soldering competitions by giving the competitors a certain amount of time (e.g. 15 minutes). Whoever has the most points wins.

You can also give a bonus point for least amount of solder used and nicest solder joints to make it a maximum of 10 points.

The points are displayed by putting a 9V battery on the board. A solid lit LED is 1 point, a blinking one is 2 points.

You can grab gerber files and BOM at CircuitMaker. It would be nice if you gave me thumbs up there.

And this is what a challenge looks like

[phlegmatic-prototyping]

Robolamp 3

This is the third iteration of my robotic lamp. The first one was a low cost version with laser cutted parts, a cheap optical sensor and an 8bit microcontroller. The second version was a lot like this one but was a possible copyright violation ;)

This one uses the CMUCAM5 vision sensor that can detect objects by their color.

I designed the lamp in Inventor and had it print at shapeways.

The LED board was made at Dirt Cheap Dirty Boards.

The software is the original CMUCAM firmware. Here’s the lamp in action:

[phlegmatic-prototyping]

Micro JTAG adapter

I was annoyed with bulky and expensive JTAG adapters for ARM Cortex-M development. So I built one that’s small, cheap and works with all the IDEs I use (LCPExpresso, Crossworks & Keil).

The Micro JTAG has an NXP LPC11U35 microcontroller and implements ARMs CMSIS-DAP protocol - I took huge parts of this code by micromint. It’s also a USB-Serial adapter. The board was made by OSH-Park. Three of them were just under 3$ including shipping, because they’re so small.

[phlegmatic-prototyping]

A failed attempt to create wearable technology

This is something I made lately. It is an armband that you can wear on your upper arm. It contains a battery, a Nordic Semiconductor bluetooth LE chip (NRF51822), a battery, a 9 degree motion processing unit, a lithium polymer battery and 6 vibration motors. It is supposed to give you a constant sensory input about which direction is north.

I read a paper about this some time ago and according to the research the human brain can adapt to new sensory input very fast. So in theory someone who receives constant input of an absolute reference point for 'north' via the vibration motors that are arranged around the arm should get a better sense of orientation without really thinking about it.

The only downside is that it does not work the way I designed it...

Some details first: I used a module with the nordic chip (8$ on Aliexpress) to create a quick prototype. So the board is pretty simple:

I contains only the NRF51822 module and a MPU9150 nine axis gyro/accelerometer/magnetometer soc. There's also some stuff to load the battery via a micro usb port on it.

The vibrations motors are from an iphone 4s. These can be operated with a pwm signal and are available for a few cent on aliexpress.

I made some carrier boards for them as well. The boards are from smartprotoyping.com. They provide a great service but I should have chosen a different pcb color than yellow. This is more like orange.

This is how the pcbs look assembled:

I measured my arm and made a quick design in Freecad.

I planned to 3d print this with flexible PLA but had some problems with the process so I made the first one from hard plastic. Put the battery and the boards in it and went on to programming (a.k.a. 'The boring part' :) )

BTW: It feels kind of weird to debug something that is attached to your body.

So I wanted to have two modes on this thing. In 'standalone mode' it should just constantly tell you where north is by vibrating the motor that is closest to that direction. The second function is a 'navigation mode' in which the device sends the direction data from the MPU to a smartphone via bluetooth LE and a navigation software on the phone controls the vibration motors so it can direct you when you're walking or biking through the city.

Everything worked as expected, except for one thing: The skin on the upper arm is too insensitive to tell the different vibrations motors apart - I should really have tested this before :). Now I have to figure out a different way of getting that data into the brain. Any ideas?

[phlegmatic-prototyping]

My current project: A robotic camera dolly that can turn on a dime. You can 'record' movements by driving it with your hands and than replay these at an adjustable speed. Controllable via smartphone (bluetooth le).

I built the robot last summer because I wanted to shoot time-lapse videos with movement. Because it would be too complicated to program the path for complex movements I decided to give it a recording function. So you should be able to put your camera on it and guide the bot manually while watching the image on the cameras screen. Then the bot will replay these movements back and forth at a configurable speed. It would also ease in and out to suppress jerky movements movements. For the time-lapse the bot would move a little distance, the camera takes a shot it moves a bit more along the path, another shot and so on.

I bought some high-resolution industrial rotary encoders from Ali-Express and combined them with large stepper motors. I used two Pololu A4983 stepper motor drivers and a Teensy 3.0 arm-board as a controller. I also added an SD-card reader for recording the movements and a bluetooth LE module from Texas Instruments for remote control. The power is supplied by a 3 cell 2000mAh lithium battery. As an input for the settings I’m using a small rotary encoder knob with illumination from spark fun (https://www.sparkfun.com/products/10596). I also added 3 white status LEDs. Some initial tests with these components on a breadboard were successful so I went on with designing the chassis.

It took some time to figure out how to make the robot turn in place without adding a steered wheel. An additional turning wheel would make the movement recording part difficult, because it would swivel around when changing the direction and add unwanted side-movement. After some research I found an Australian company called ‚Rotacaster‘ that makes omni-directional wheels that are soft enough to give a smooth ride in all directions (http://www.rotacaster.com.au). I ordered some and retrofitted them with high quality ball bearings.

For the driven wheels I used large wheels from inline skates. I made some special hubs on the laser cutter at the local hackerspace „Attraktor“ (http://attraktor.org) in order to mount them directly to the stepper motors. The things that hold the motor and the encoder I made of aluminum profile by with a saw and a drill. The base plate was made from a piece of "Aluminium Composite Panel“ that I found at the hackerspace. I used the spaces CNC-mill to cut it. Then I mounted the wheels, motors and encoders on it.

The circuit board for the electronics was made on the pcb-mill at the hackerspace. It’s just a large board with pin headers on it. All the components are plugged into that board which allows for quick replacements if something should brake.

When it was done it looked quite cool but also very industrial because the top showed a lot of screw heads. Someone at the space had the idea of putting a panel of semi-transparent acrylics on top of it. So I went back to the laser cutter and made a grid from wood to put on the top in order to get a level surface for the acrylics (You can see it shining through a bit). Then I cut the acrylics and glued it on top. It’s really cool that the status less shine through it.

I added a quick-release camera head and the hardware was done. So I started on the Firmware. That went pretty quick, because I could use large pieces of code from a project called ‚grbl‘ that is used to drive cmc machines (https://github.com/grbl/grbl). It’s not perfect yet, but it works quite well.

Here’s the first Video I shot with it: https://www.youtube.com/watch?v=b6aIduEKIyE

This year I would like to switch from the stepper motors to brushless motors to make it quiet, but there’s some serious work involved in controlling these motors slow an precise.

[phlegmatic-prototyping]

I was playing with a laser cutter lately and this is the outcome. A robotic lamp that follows your movements with an optical flow sensor from a computer mouse and an ATMega 328.

I'll release the files and schematics when I've fixed a few mechanical bugs.

Heres a video: https://vine.co/v/bUzYzEmiKLK

[phlegmatic-prototyping]

I built a lamp from a block of nice wood.

Here's a video: http://vine.co/v/b1K1LqqBAYD

[phlegmatic-prototyping]

The Count-O-Matic

 A nice little appliance that displays up to 8 different metrics. Powered by a Raspberry Pi. We build this because I needed something that looks nice and displays large numbers from any web api. The case was designed and built by [Jochen] from the local makerspace. It contains a chinese tm1638 segment display with buttons and a Raspberry Pi. I wrote a ruby wrapper for Martin Oldfields [c library for that display](http://mjo.tc/atelier/2012/08/pi-tm1638.html), so it's easy to read data from sources like google analytics and display them. The buttons switch the displayed number. If you're interested in getting some of these made contact me on twitter (@r1ckp).

[phlegmatic-prototyping]

The OLEDuino

 This is a cheap electronic name badge that can be programmed with the arduino ide. I built it for events like barcamps at the attraktor makerspace.  It shows your name and info and if you get bored you can play a game of breakout on it or program it with your own game. It's basically a minimal arduino that can be programmed with the arduino ide and an FTDI adapter. The battery can be charged with a standard lithium-polymer charger for RC helicopters. Due to the energy efficient OLED it runs about 10 hours with a small battery. If you'd like to build one get the schematics here: https://github.com/RickP/oleduino - The circuit board is single-sided, so it's possible to make it yourself. There are some hints in the readme on where to source the components.

[phlegmatic-prototyping]

CNC Music.

[phlegmatic-prototyping]

A case for my Raspberry Pi. Made from recycled teak wood.

[phlegmatic-prototyping]

The first design of universal smartphone holder for my motorbike.

[phlegmatic-prototyping]

Great work with the RFM12B, looking forward to trying this out on the RaspberryPi. Are you aware that the SPI bus now has drivers on Pi ? I assume this will help bit rate ?

Yes, I got my Pi last week and I'm working on it. SPI works great but I'm still missing interrupt support for the gpio pins. I guess it will be in the stock kernel when I'm done - I want to add a web interface for configuring the nodes and sending data to/from them and that will take some time :)

[phlegmatic-prototyping]

Printing a molecule - Part 2

I finally found the patience to finish this.

I printed the molecule model on the attraktor 3d printer - a custom build printer with a large cnc table and two makerbot mk6 extruders that is controlled by a computer running linuxcnc .

Because of the very unregular stucture i had to added some support structure to the print. That means that the overhanging parts are supported by a grid of extruded plastic that is extruded at lower speeds.

The printing process:

All the parts with a grid-like structure are support material.

The print took about 8 hours. Here’s the result:

After breaking away the support:

The surface was pretty rough, so I used a small sandblaster with my airbrush compressor to prepare the surface for the next step:

After that I applied about 10 layers of white paint primer. The result was a very smooth surface and I got rid of the visible "steps" from the printing process. I finished it with a layer of glossy varnish because I had to hand-paint it afterwards and that way I was able to wipe away paint with a q-tip when something went wrong.

Then I painted yellow, red and blue dots the surface (Red = Oxygen, Blue = Nitrogen, Yellow = Sulfur). I used a toothpick and airbrush color for that. That wasn’t fun at all and I won’t do that again :)

Then I put some layers of matte varnish on it to protect the paint. Here’s a video comparing the computer model to the printed model:

[phlegmatic-prototyping]

Connecting microcontroller projects to the internet - the cheap way

 There are a few options to connect an arduino to the internet. Unfortunately most of them are pretty expensive. You can get an arduino board with built-in ethernet or attach an ethernet shield to a standard arduino board. These will cost about 40 euros and you'll have ethernet cables lying around. There are also WiFi shields but these start at about 60 euros for the shield alone. Also these add significantly to the project's power consumption A cheap and easy solution for adding wireless communication to an arduino is the 868Mhz [HopeRF RFM12 wireless radio module](http://www.hoperf.com/pro/rf/cob/rfm12b.htm). These transceivers are about 4 euros and there are very nice arduino clones that have them built in from [JeeLabs](http://jeelabs.net/projects/hardware/wiki/jn6). The range is about 100 meters indoors and up to 300 meters outdoors.  I got one of these and some of the smaller [JeeNode Micros](http://jeelabs.net/projects/hardware/wiki/JeeNode_Micro) that use a ATTiny 84 instead of the ATMega 328 - These are great for small battery powered sensor applications.  You can get USB modules with an RFM12 on them and attach them so a computer to create a bridge from them to the internet but that means that you'll have to run a computer all the time just to forward things to the net. I had a couple of old 'La Fonera' routers available. These were given away for free some years ago by a company called [Fon](http://www.fon.com/) and there was a box full of them at the local [hackerspace](http://wiki.attraktor.org). I decided to create a gateway from the RFM12 transceivers to the internet with these routers.  The main use case for this are battery powered sensors/actors that are in sleep mode most of the time and wake up in certein intervals to either send some sensor values to the net (e.g. Twitter) or get some setting from the net and switch something on/off. Initially I wanted to send information about the humidity of my plants on the balcony to a website. The Fonera routers have 5 gpio pins that can be accessed on the board and used to create a [bitbanged](http://en.wikipedia.org/wiki/Bit-banging) SPI port to talk to the RFM12. They also run [Openwrt](https://openwrt.org/) - a highly customizable embedded linux for routers. With the added RFM12 chip you can just put one in WiFi client mode and have it sitting on a shelf. Up to 30 arduinos/JeeNodes can send data to it and the little device routes them to the internet. The gpio pins are described on the [openwrt fonera page](http://wiki.openwrt.org/toh/fon/fonera). Four of them are in a row next to the ethernet port:  And the fifth is on the back of the board on TP3:  In order to use these pins you have to remove the four capacitors for these pins. Then I put a pin header on the board and soldered a wire to TP3  and connected the pins to the rfm12 module in the following way: Fonera || rfm12 GPIO1 -> sck GPIO3 -> sdi GPIO4 -> sdo GPIO7 -> sel GPIO0 -> irq  I had to patch the openwrt linux kernel so it creates an spi port on these pins and uses the spidev driver to make the port available to userspace programs. The patch is on my github repository. Then I ported the C++ jeelabs RFM12 driver for atmel to C and made it use spidev. Everything is available on gitub: https://github.com/RickP/rfm12_gateway - There is also a ready to use openwrt image for fonera in the Fonera folder - so as long as you don't want to do this with a different router you don't have to compile it yourself. The server daemon on the openwrt device waits for messages in a certain format from the RFM12 nodes and then calls a lua script in /etc/rfm12.d. The scriptname is given to it by the arduino. So you can add your own lua script (the length of the name has a maximum of 5 letters) with a function 'process'. That function gets passed the message from the arduino. It can then route the request to the internet. When the router receives a message it sends back an ACK signal to the sender so it can retry until it knows the message gets through. The return value of the lua process function is also returned to the sender so you can fetch data from the net. There's a sample script for posting sensor values to cosm (formerly pachube) included. Just edit it to add your api key and feed id to it. You'll find an arduino sketch for sending values to the cosm script in the arduino folder. On the arduino side you need a slightly changed JeeLabs rfm12 library that uses a lower bitrate due to the bitbanged spi on the Fonera. Get it here: https://github.com/RickP/jeelib I'll try to port this to raspberry pi as soon as I get one, so this can be done with hardware that is still available. Also the faster processor would allow me to exchange the lua scripts for nodejs or ruby/eventmachine scripts. Also a kind of web interface where you can control ports or add scripts with a kind of wizard would be nice.

[phlegmatic-prototyping]

Printing a molecule - Part 1

Martin, a friend of mine, is currently working on his Ph.D. in chemistry. He is researching an inhibitor for a proteine that is involved in the alzheimer desease. I talked to him last weekend about his research and he mentioned that he's like to have a model of that protein.

At first I imagined something like a hard to print stick-and-ball model most people associate with molecular models but he said he needs a different model that just shows the surface of the molecule. I told him that I have access to a 3D printer at the attraktor makerspace in Hamburg and that I'd like to try to print that in plastic.

If you never heard of a 3D printer read this.

He sent me a file with the extension mol2 that is basically a text file that describes the position of all the atoms in the molecule and the bindings between them. This is pretty useless for a 3D model.

I searched the internet for an application that displays these files. At first I found a free mac software called FieldView that displayed the molecule and let me rotate it.

It's nice I don't have to create a stick-and-ball (that's the actual scientific name). This protein consists of 5215 atoms. Unfortunately FieldView has no option to export the model as an STL file.

STL is the most used file format for CAM (Computer Aided Manufaturing). STL files describe 3-dimensional objects with lots of triangles that make up surfaces. CAM tools can generate the paths for a CNC mill or a 3D printer from these files as G-code.

I tried BallView, an Open Source molecular modeling and visualization tool that seemed perfect for this tasks. Unfortunately the OSX version crashed immediately when I opened the molecule file. But it worked in my vmware ubuntu.

Loaded the mol2 file, set the display model to SES. Thats the type of surface model I need. Unfortunately I know too little about chemistry to elaborate on that. Then I exported the model as an stl file.

There's another open source tool called 'pymol' but ballview seems to have more features when it comes to graphics.

I checked the exported stl file in MeshLab and it looked OK. I think I have to put the colors for the atoms on it afterwards with the airbrush.

Now that I have an stl file I can finally generate the gcode for the 3d printer. I'm using replicatorG for that with a profile for the attraktor's printer.

I loaded the stl and positioned it. Wasn't easy to find the flattest surface :)

Then I let it calculate how to spread the plastic layers. I got a lot of warnings about the meshes in the model. Tried to fix them in MeshLab but there are way too many options in that tool. A web search brought up the free netfabb cloud service. I uploaded the stl to them and they generated a cleaned up model.

ReplicatorG wasn't complaining about the cleaned up stl file and displayed some cool looking toolpaths for each the 180 layers of plastic it will lay down:

If you're courious - this is how the gcode looks like (some thousand of these lines):

G1 X23.48 Y-1.68 Z0.89 F1800.0

This means 'move in a straight line' (G1) to the following XYZ coordinates with a speed of 1800mm per minute:

I'll load it into the attraktor printer as soon as I get some private time with it.

Part 2 here

[phlegmatic-prototyping]

iPhone geiger counter

 I built the first prototype of an iPhone geiger counter. It uses a russian SI-29BG geiger tube that I got from ebay for 20 Euros:  These tubes need a current of 400 volts to work, so I looked for a circuit that accomplishes that on a small area and uses standard components. I found this: [Geiger circuit](http://sites.google.com/site/diygeigercounter/circuit-descript) After adapting it a bit, so it can output the signal to the iPhones audio jack I tested it on a breadboard:  It worked. Then I spent a long time fitting it on a 25x55mm board with a one-layer routing I milled a pcb from that.  Next thing was the case. I designed it in FreeCAD:  Milled that one from a POM plastic block on the cnc mill of the [attraktor hackerspace](http://attraktor.org) in hamburg (best place in the world!), turned out OK:  Then I put my stuff in it:  The app is called 'geiger bot' and it's available for free on the appstore. I'm currently working on Version2.0 that won't have a physical switch, a smaller battery and is 5mm shorter. When that is done I'll release everything that's needed to build one yourself here. I could also make a small series of cases if enough people want one.