Watching closely, it is not a single LED, but twenty single 1-W dies tied in series and parallel in a single package. The seller offers also a 50W model arranged in the same fashion

The setup i rigged up for a quick and dirty test of the 20W led

These are the parameters of the led from the vendor; there is no official datasheet, even if the page of the seller cites the luxeon brand. The following information however is all I needed to make a quick try:

Lens Color: Water Clear
Emitted Color : Warm White
DC Forward Voltage (VF): 13.5 ~ 15.0 VDC
Forward Current (IF): 1400mA ~ 1800mA
Viewing Angle: 140 Degrees
Color Temperature: 5500K
Intensity Luminous (Iv): 1000LM

After some rough calculations I found that a 18V power supply and a couple of 1 ohm resistors (what I had lying around) was just about right to turn the led on and keeping inside spec. The heatsink (coming from an AMD processor box) keeps the led cool, and prevents it from frying up in a few seconds: this little beast puts out a lot of heat.

See for yourself the comparison pictures between a conventional light and the led ligt from the test setup:

My room lit by my every day lighting (six fluorescent light bulbs)

my room lit by the 20W led

As you can see a single led can replace a whole chandelier of six light bulbs, giving comparable results and a clearer light. This is definitely going to be the lighting system of the future, because duration and consumption are far more efficient than conventional systems. I wonder what the 50W model could do.

A final word of recommendation: if you decide to try one of these leds, make sure you don’t watch it directly… it is like when you watch the sun with your bare eyes. Always wear sunglasses when you’re tinkering with these things.

Have you ever found yourself missing the lyrics for a song in your iPod, but you don’t want to go search and add lyrics manually to each song in your iTunes library? Then this is the app for you!

It can update the lyrics field in iTunes for the currently selected songs, automatically. The lyrics are fetched from the lyricwiki.org website, which has a pretty good library of songs, even though some of them have been obscured by a copyright claim.

This is how to use it:

Step 1: select some songs in iTunes

Select some songs in iTunes. (you can use command-a to select all songs on screen)

Step 2: open MassiveLyrics and hit the Get button

The lyrics have been looked up

The app will start looking up songs on the internet. If the lyrics are not available it will be shown, otherwise you’ll see ‘lyrics found’

Step 3: push Write all found lyrics to iTunes to update the songs in your library

Just the songs in the ‘lyrics found’ state will be updated in your iTunes library. Song state will be updated to lyrics written.

Each of the previous three steps will take some time, depending on the size of the selection you’ve made. Please be patient while it loads

Step 4: enjoy!

You will find the newly added lyrics in the get-info dialog of your iTunes songs, under the lyrics tab, and next time you’ll sync your music player they will be copied and available also on the go.

Download massivelyrics v1.1

I have tested the app for some time on my mac, and it works fine.
However it is provided AS-IS and I cannot be held responsible for any damage it may make.
If you download lots of lyrics with this app, please consider buying me a coffee donating a small amount through the button in the app. Thank you!

Here is the schematic [click to enlarge]:

The top left connector is for input from a microcontroller, which determines which coils are turned on. The rest is divided in four identical sections, each having a BC337 NPN transistor driving a TIP122 power darlington pair, which in turn is attached to the motor coil through the bottom right connector. Collectors of the power transistors are connected to a sense power resistor which allows fixed current to flow in the motor, independently of the voltage.
As for now the motor I am using is underpowered: its spec says 24v but I am giving this circuit a 18v supply, so speed and torque may be affected. For now I can reach a speed of about 10 revolutions per second of the stepper.

Here is how it came out on a breadboard, attached to an arduino and to a stepper motor [click to enlarge]:

The little trimmer in the picture has been added for controlling a delay in the firmware, so to test different speeds of the motor. In fact the speed can be raised until the motor reaches resonance, and just a tad before that point the motor is going at full speed. This is the code:

void setup()   {
 DDRB = DDRB | B11111111;
}

int sensorPin = 0;
int sensorValue = 10;

void loop()
{
  sensorValue = analogRead(sensorPin);

  PORTB = B00000001;
  delayMicroseconds(sensorValue*10);
  PORTB = B00000011;
  delayMicroseconds(sensorValue*10);
  PORTB = B00000010;
  delayMicroseconds(sensorValue*10);
  PORTB = B00000110;
  delayMicroseconds(sensorValue*10);
  PORTB = B00000100;
  delayMicroseconds(sensorValue*10);
  PORTB = B00001100;
  delayMicroseconds(sensorValue*10);
  PORTB = B00001000;
  delayMicroseconds(sensorValue*10);
  PORTB = B00001001;
  delayMicroseconds(sensorValue*10);
}

As you can see the code is very simple: it just does a full step sequence delayed by the value read from the trimmer. It is good for testing different speeds and step sequences, to see which one is best for achieving maximum speed.

The next steps for beta version will be:

• add PWM to software
• port software to avr-gcc and optimize timings
• determine if BC337 are really necessary or not
• add filtering caps for PWM
• add diodes for coil protection

Apple did not do a fair thing here: I watched in horror the screen turn progressively white, while hastily trying to power off the device. I couldn’t make use of the software switch and so I had to watch with even more horror the phone get warm and stay on until I came home from work (50 minutes).

Damn… there should be a hardware switch to turn it off… and iPhones should be waterproof too. I mean what is the use of having a compass inside? Make it waterproof instead. These would be killer features I think…

However to turn it off I picked up one of the many disassembly videos and followed the instructions carefully. The phone eventually did turn off, when I unlatched some cable inside, and this nice picture is what I ended up with.

my iPhone dissection results: click to see bigger

I hope I will be able to put it back… just to disassemble it was pretty hard. For now it is soaked into rice, so that some humidity will go out, especially  from the screen. I do not expect miracles, however it is worth a try, while waiting for the iPhone4 here in italy…  wish me luck!

UPDATE: I tried to revive the phone… unfortunately when I reassembled the motherboard to the display it got quite hot and no image popped up on the screen. I have to go for the iPhone 4. Or the 3GS maybe. For now I stick with my ancient nokia 6110… not quite what you call a smartphone, but it’ll do for now.

I personally use firefly to serve an enormous ‘music dump’ folder which I keep on the NAS server, so that I am not forced to have all these songs into my iTunes library on my mac HD. This folder is for casual music i listen from time to time, and if I find some gem in it, I copy it on my main library to sync with ipods and iphone.

Installation in ubuntu is very simple: just

sudo apt-get install mt-daapd

and start the daemon by running

sudo /etc/init.d/mt-daapd start

If you want the program to start every time your server boots up execute the following:

sudo update-rc.d mt-daapd defaults

Now you should see in your iTunes a new shared library, like this:

firefly server as seen by iTunes

Now we are ready to configure a folder on the server for sharing. To do this there is the configuration file /etc/mtdaapd.conf, however it is more straightforward just to use the web interface at port 3689 on your server. The username is admin and the password is mt-daapd.

From the web interface you can set up the daemon and in the configuration section you can specify your music folders to be served. Just remember to make it perform a scan after you add something new. To be sure I configured some periodic scanning to go off every hour.

Performance of the server is very good. The library behaves as if it were local, if not even better.

Avahi installation is almost hassle free; first you install the package:

sudo apt-get install avahi-daemon

Then edit your avahi configuration file located at /etc/avahi/services/afpd.service. The file should be empty, so you can

sudo gedit /etc/avahi/services/afpd.service

and then paste in the following XML





%h

_afpovertcp._tcp

After that just restart the daemon:

sudo /etc/init.d/avahi-daemon restart

Back on your mac you should now see the service advertised in the Finder sidebar, like this:

Share advertised by avahi

Unfortunately the installation of netatalk in Karmic is a bit more complicated than an apt-get call. Due to licensing issues they could not distribute a package with SSL precompiled inside, and this security layer is mandatory if you try to connect from any mac running leopard or snow leopard.

Many tutorials exist (here, here) which explain how to recompile netatalk from source code, adding SSL support. But neither of these worked for me: no matter how i did configuration or compilation, my mac always failed to connect at the username/password screen, so still an authentication problem in spite of ssl.

So I had to remove netatalk completely (apt-get autoremove netatalk) and upgrade to a newer version, not yet distributed in the official repositories: 2.0.5.

This new version works out of the box: just install it, configure the folders you want to share and you are done. It also has a special option for setting up some space to work as a drive for time machine.  Thanks to this article we can find the packages to install for the new version. Just download and double click the packages in the listed order, say yes to any message recommending an older version.

libdb4.8
libgcrypt11
netatalk2.0.5

now edit your /etc/netatalk/AppleVolumes.default file, appending a line for every share you want to export to your home macs, with this format for a generic share:

/full/path/on/linux/box   "name of the share"

or for time machine drive:

/full/path/on/linux/box   "TimeMachine"    options:tm

then restart the netatalk daemon:

/etc/init.d/netatalk restart

Finally you can check on your mac if everything went fine:
open Finder and do <command-k>, enter the IP address of your server and your (ubuntu) username and password.
You should now be connected and capable of browsing files, like this:

connected nas

Moreover, if you specified a time machine share, you will be able to set it up directly into time machine preferences pane without any terminal hacks on mac side.

The next step is to configure avahi, to advertise services on the network so that they come up in the finder sidebar (no more command-k to connect).

Setting up ssh access is very easy:

first of all install the ssh package:

sudo apt-get install ssh

then append the following line to your /etc/hosts.allow file, telling to allow incoming connection on ssh port 22. In this case I’ve configured it to accept incoming connections from all the IPs of my local network. Edit the line to suit your IP configuration.

sshd: 192.168.0.*

That’s it! Now open a terminal (in mac/unix) and do

ssh <username>@<server-IP>

If you’re on Windows, you can use PuTTY to connect to the server. Have fun!!!

MDADM (multi disk admin) is the software implementation of RAID on linux. It can be used to take a bunch of disks and make them into a single, powerful drive. This leads to several advantages, depending on the raid configuration you choose to implement.

For my setup I had chosen RAID5 with three disks. This means that if I loose one of them, I will not lose my data: pop in a fresh disk and go on, without even interrupting data availability.

MDADM does not come with ubuntu by default.

sudo apt-get mdadm


will take care of installation. Actually on my new system it took care of more: it automatically detected and assembled the array I had from a previous machine (that one had been built with MDADM as well), giving me this nice little icon on the desktop with all the content ready to go. This array is made up of three 250GB disks.

However, the work needed to setup the array if you have only new disks is a bit more complicated than just popping in the disks and let linux autodetect everything for you. Here you can find the process to set up the array:

1) Prepare the drives
Connect the disks and create an ext3 partition on each one. Partitions should be the same size and filesystem. You would want to span each drive with a single primary partition. Use fdisk and mkfs.ext3 or gparted to do that. You should now have many partitions as your disks named for example /dev/sda1, /dev/sda2 and so on (if you are using IDE disks those will be /dev/hda1, /dev/hdb1…)

2) Create the array
Merge the partitions into an array using mdadm:

mdadm --create /dev/md0 --level=5 --raid-devices=3 /dev/sda1 /dev/sdb1/dev/sdc1

This for example will create an array, accessible at /dev/md0, made up of your three partitions, using raid 5 as the level
check the status of mdadm with:

cat /proc/mdstat

3) Mount the array and use it

sudo mkdir /media/raid
sudo mount  /dev/md0 /media/raid

if the mount fails, you have to use format the new created drive md0, use again mkfs.ext3.

4) Edit/check configuration files
for automatic recognition and mount in the next boot sequences edit the following files:

sudo gedit /etc/mdadm/mdadm.conf
sudo gedit /etc/fstab

Now data written in /media/raid is safely stored, and an hard disk failure cannot compromise it. Being on a NAS server, this can be a solid foundation for all automatic backup needs of the home network;

atom NAS server assembled (well... almost)

The Hardware

The motherboard: asus AT3N7A-I: based upon the new combination of Intel Atom processor and Nvidia ION chipset. This little thing is capable of low power consumption and reproduction of 1080p video. Given this it will double as a HTPC, which is the segment it was designed for.

The major limitations of the board for becoming an expandable NAS server are the small number of SATA II ports (three) and the reduced expansion capabilities (a single PCI slot). The 3 RAID hard drives in the picture are attached through a SATA expansion card, which sports four additional sata ports, while the main drive for the OS and additional space is attached to the motherboard. With this configuration I have a spare on the expansion card and two spares on the motherboard. Will do some performance tests to find the perfect combination.

Then the board has gigabit ethernet, bluetooth and HDMI out. Perfect for HTPC. I tested it with XMBC and the overall system turns out to be ideal as a media center. With plenty of disk space attached it will provide an endless supply of video! cool!

A bad point of the motherboard is surely that nasty little fan that runs over the heatsink. I will have to find a way to reduce RPM, or to cut it off completely. The system must be very silent and up 24/7. It cannot be a noise machine as it currently is. I mean the fan noise is tolerable, but steady and in the long run can be annoying, at least for me. Right now – however – it is overtaken by the PSU fan, which is a cheapo one and for that very very noisy. I’ll find a way to silence that too!

The software

Everything will run upon the latest Ubuntu OS (regular version, not server, so to run XBMC or boxee). Right now I am using Karmic, which installed flawlessly off an usb drive. The only minor thing has been installation of the latest Nvidia driver: the one installed by default by Ubuntu at this point was too old and did not play well (at least at 1080p): compiz was not able to run, and the overall performance was slow and buggy. The latest Nvidia driver fixed this, but it had to be installed manually.

These are the components i’ve set up so far. Each has its own dedicated post with instructions, just click on the big buttons. If you’d like, you can suggest more home server sw in the comments!

The installs have been made with Karmic, so some distribution specific workarounds apply (netatalk, xbmc). If you are using a newer version these workarounds may not be applicable anymore and installing might be as simple as doing apt-get install.