When the Regency TR-1 transistor radio came out onto the market in the 1950s, it was hailed as a modern marvel of microelectronics. With only four transistors and a handful of other components, the TR-1 was a wonder of modern engineering. [Sprite_tm] may have those old-timers beat, though. He built an FM transmitter with the lowest parts count of any transmitter ever.

Like most of [Sprite_tm]‘s builds, it’s an unimaginably clever piece of work. [Sprite] overclocked the internal RC oscillator of an ATtiny45 to 24 MHz. After realizing the PLL running at four times the frequency of the oscillator was right in the middle of the FM band, he set about designing a tiny FM transmitter.

[Sprite_tm] remembered his work on MONOTONE and made a short song for hit ATtiny. The firmware for the build takes the notes from his song and varies the 96 MHz PLL frequency a tiny bit, thereby serving as a tiny FM transmitter.

Does it work? Well, if you want to compare it to a Mister Microphone, the range is incredibly limited. That being said it works. It’s an FM transmitter built out of a microcontroller and a battery, and that’s very impressive. Check out [Sprite_tm]‘s demo after the break.

Around this time last year, [Sprite_TM] took a 1980′s-era Macintosh SE and rebuilt it as a home file server. He used a Seagate Dockstar as the new motherboard, but over the past year he’s been annoyed with the fact that the Dockstar doesn’t have real SATA ports. Using USB to SATA converters on a server is a slow way of doing things, so [Sprite_TM] rebuilt his SE using an HP thin client. To do this, he had to break out the onboard SATA and PCIE; not an easy task, but that’s why [Sprite_TM] is around.

The first order of business was installing a pair of SATA ports. The stock thin client had two NAND-flash chips serving as the drive, both connected to a SATA controller. All [Sprite_tm] had to do was desolder the flash chips and wire up the new SATA connections. Easy enough.

Because the HP thin client only had 100Mbps Ethernet, [Sprite_tm] wasn’t looking forward to the order of magnitude difference between his expected rsync speeds and what he would get with a 1Gbps connection. The only problem is the thin client didn’t have a spare PCIE connection for an Ethernet card. That’s really no problem for [Sprite_tm], though: just desolder the GPU and run a few wires.

Just like last year’s work on his SE, [Sprite_tm] ended up with a functional and very cool home server. The old-school System 7 is still there, and of course he can still play Beyond Dark Castle. Awesome work, in our humble opinion.

[Sprite_tm] is back again, and his work never fails to impress. His latest project is a Game Boy Advance MIDI synth that takes MIDI data from a keyboard or sequencer and maps that to Game Boy sound channels.

Because he seems to never do anything the normal way, [Sprite_tm] decided to run the Game Boy without a cartridge. We’ve seen this before; the GBA boots into the synth software over the link cable with multibooting.

[Sprite_tm] designed a circuit around an ATMega168 to boot the Game Boy, receive MIDI messages, then translate and send them through the link port. The main Game Boy code also has a simple interface to display graphics on the screen and a sequencer that allows him to record MIDI messages into 8 different tracks. We’re thinking that being able to sequence Game Boy chiptunes with a keyboard is a huge improvement over Little Sound DJ.

As [Sprite_tm] freely admits he’s not the best keyboard player, he gave his GBA synth to a musician friend. The resulting song is a veritable deluge of 8 bit chiptunes. The electronics behind this build are very simple (and can fit inside a MIDI cable), so we’re ready to break out the iron for this one. All the relevant code is posted [Sprite_tm]‘s build page. There’s also a video walking through the features of his creation, viewable below.

[Sprite_TM] outgrew the features of the cheap unmanaged TL-SG1005D switch he was using on his home network. Instead of buying a new and much more costly switch he cracked the cheap one open and found that the RTL8366SB chip inside possessed the ability to work harder but was crippled for sale as a low-end model. It wasn’t as easy as that oscilloscope firmware upgrade we saw a while back. He had to add an AVR ATmega88 to send I2C commands to the switch. Turns out that the I2C protocol wasn’t standard and after much head scratching he found some Linux drivers for the chipset that gave him enough info to send the configuration commands he needed. Now he’s go the managed switch he needed for his VLAN for the cost of a microcontroller and some wire.

Want to make the above yourself? [Sprite_tm] did a thorough job documenting the build step by step (complete with pics, schematics, graphs, and links to the parts used). In summary, [Sprite_tm] busted open an Ikea CFL bulb to reuse the housing. Inside, he installed a scavenged power supply, ATtiny44, RGB LED module, and a radio receiver. A remote control allows [Sprite_tm] to change the lighting of his room to nearly any color. The cost of the project is a little under $30. The price tag isn’t so steep when one considers the insanely long lifetime of LEDs.

Over at SpriteMods, [sprite_tm] realized that a microcontroller could be used as a boost converter to power itself. A boost converter steps up voltage from a battery by switching the output of a coil. First, it is tied to ground so a magnetic field can build up in the coil. It is then released as a higher voltage than the input. Normally dedicated chips do this at an incredibly high frequency, but the PWM signal from an AVR works well enough. This can be used in low-power situations where space is an issue.

[via EMSL]

[sprite_tm], whose projects we have covered in the past, took the popular bristlebot to an extreme and created a controllable version. A bristlebot consists of a small vibrating motor mounted with a battery on the head of a toothbrush. These micro-robots buzz around randomly, and he attempted to tame them. He used a platform of twin bristlebots and added an optical sensor from a laser mouse and an ATtiny13. The optical sensor is used to determine the relative motion of the robot, so that the motors can be adjusted accordingly. He also has a video of the bot using the sensor to find a mark on the floor and stay within bounds. Although it isn’t as accurate, it acts like a traditional line-following robot.

[thanks sprite_tm]

We love old display technology, like Nixie tubes, but they’re often difficult to work with because they require higher voltages than digital logic. Vacuum florescent displays (VFD) fall into this category. While not necessarily “old”, they are becoming far less common than LCDs. The main benefit of a VFD is that it actually produces light directly; it doesn’t require a backlight. You’ll find these displays on various players and appliances: CD, DVD, VCR, microwaves, stoves, car headunits, and others.

[Sprite_tm] had written off some VFDs, but recently revisited them with renewed interest. He started by testing what sort of voltage would be required to drive the display. It took 3V for the filament plus 15V to drive the grids. There are VFD controller chips available, but he wanted to get this working with what he had on hand. He had experience with older 40xx series logic, which can be powered by much higher voltages than 5V 74xx. His final schematic has three 4094 serial to parallel chips with an ATtiny2313 controller. A 5V power supply is dropped to 3V with diodes to drive the filament while a boost converter brings it up to 15V for the 4094s that switch the segments. While the code is specific to this display, it would be a great place to start your own project.

We mentioned Pulse Width Modulation (PWM) when talking about [sprite_tm]‘s marquee control. It’s a method of power control. While [sprite_tm] did it in software, [Afroman] sent along a very straight forward introduction to PWM using just a 555. Check out his video for coverage of this fundamental electrical design technique.

In a previous job, [sprite_tm] was responsible for wrangling many different LED text ad marquees. The hardware was fairly simple and he always figured they could be pushed much further with a little work. He recently acquired ten 32×16 LED displays a decided to see what he could do with them. By the end of the project, he had full motion video running on the display. This is a great project to read up on if you’ve ever wondered about LED matrix displays. He starts by reverse engineering the electronics on the board. He then attached an ATmega88 to drive the display module. Multiple display modules were daisy chained together over serial. The article covers PWM control and refresh timing as well. Check out one of a few demo videos below.

Our flickering LED circuit combined two known circuit, and certainly wasn’t graceful because of it. [sprite_tm] saw quite a few areas where the circuit could be reduced. He ended up taking it down to just two LEDs, a battery, and an ATtiny13. The first step was getting rid of the current limiting resistors. The datasheet shows that with a 3V supply the AVR will limit the current well below the maximum current. The light sensor was removed next. [sprite_tm] referenced an earlier post on sensing with LEDs. He measures the voltage across one of the LEDs while it is off to see how much light is hitting it. The current draw while on is 10mA and 50uA while off.