After posting the videos of the Crazyflie and sending the link to Hackaday.com last year we got a lot of positive feedback and a lot of people asking if we would/could make a kit. So we redirected our efforts towards taking our old prototype and redesigning it to be more suitable for manufacturing. In order to fund this development we (the Crazyflie team) created a company named Bitcraze. Since we now feel that the Crazyflie project has outgrown Daedalus.nu we will start using the Bitcraze webpage to post progress and technical information.

Our goal is to be a bit more active on the new site and post more updates and information about the project since we are getting closer to completing it. So to sum it up: Go to www.bitcraze.se for the latest updates on the Crazyflie!

/The Crazyflie team

/The Crazyflie team

We have added more photo in our picasa album.

/The Crazyflie team

Ps. We recently had problems with visitors not being able to post comments (Epic anti-spam fail…). It should now be fixed, sorry about that.

We now have a couple of late nights in front of us doing burntesting and porting the firmware. We will keep posting pictures of the progress so stay tuned.

We also got the prototypes for the USB radio dongle that is used to communicate with the Crazyflie.

/The Crazyflie Team

As soon as we get the boards we will post some pictures but for now a simple 3D model of the board have to suffice.

/The crazyflie team

It is about time we post an update about the progress of the crazyflie kit. We are actively working on an improvement of our crazyflie quadcopter that we hope can be sold as a kit. The following bulletin is what we can say so far about the development.

• The schematic for the new version is done and the PCB is about to be closed. We have decided to use analog sensors to have better control over sampling points even though it requires three separate chips instead of one. It may sound like yesterdays technology but since they don’t make 6 DOF in one package with analog outputs we just have to stick with that for now.
• We now have a single-board design. It means that the radio is integrated on the main board and not as an extension board soldered in an angle. This will save some precious weight.
• The LiPo charger has been upgraded. We now use a micro-b-usb connector for charging and it can also be used for software upgrades.
• We added an extension port with power, I2C, SPI. It is a trough hole 1.27mm pitch header which means the extension board can be mounted on either side. The same header also contains the ARM 10-pins JTAG interface.
• We are preparing a Wiki on which all the copter documentation will be shared.
• We are still dedicated to release the full source code under a free software license.

/The copter team

PS. We received information about the fact that someone is selling “our copter”, using our photos and videos, on a famous Chinese shopping website. He apparently has quite a lot in stock so we are wondering why we put so much work on this copter, we could just buy a couple of those, right? No, don’t be fooled by such scam, when a kit is available, we will definitely post it here first! .DS

We have gotten a lot of questions about kits and source code and our plan is not to disappoint you here. We are working on a new version of CrazyFlie that will be more robust and cheaper (ie. with motors and propeller sourced properly). When that is ready our plan is to release the source code under some open source license including schematics, documentation and the possibility to buy kits. However we all have full time jobs and have to work on the crazyflie on our spare time so we can’t say when that could be. We plan to post more information as the project advances, so stay tuned!

We have set up a Picasa album to share some more photos, we will upload more photos of the development later:

/The crazyflie team

The process starts as for an industrial soldering: by applying solder paste on each component pad and then placing the components on the board:

Then the board is placed on the frying pan. The idea is to homogeneously warm up the board in following the right temperature profile (this profile is usually given by the components datasheet, it is the same as for an industrial oven). We followed the temperature profile using an infrared thermometer and regulating the stove power manually:

Finally the board ends up well soldered. Like with an oven the solder paste keep the component in place when it is in fusion and so all the chips, resistors and other component are well aligned:

We have done this process for our three copters and have had very few fixes to do after that.

/The crazyflie team

The main architecture of the system is as follow:

The copter itself is organised around the main CPU. The job of this CPU is to acquire the physical measurement given by the gyroscopes and accelerometers and to control the motors to keep the copter stable. This is done by a regulation loop which controls the motors speed 250 times per second. The radio communication has a pretty low bandwidth and is used to send commands to the copter and receive telemetry data from it. The CPU program can be updated by radio.

The computer runs control and telemetry programs. The control program reads the input from a game-pad and sends control commands to the copter. We also have programs that can configure the copter regulation parameters and log the measurements in order to make easier to tune the regulation.

All the development is done on Windows and Linux. Indeed we are 3 to work on this project, two of us work on Linux and one is mainly on Windows. Using FLOSS permit to handle that in a very effective way. We are mainly using the GCC compiler from CodeSourcery for the copter program compilation, GNU Make for the project build, Mercurial for the source control, and python/pyusb for the communication with the copter. All these softwares works seamlessly on both Linux and Windows and made the project pretty easy to handle.

The distance between the motors (horizontally and vertically) is around 8 cm and the total weight is around 20 g.

Details of the top side of the copter electronic board:

The top of the board contains the sensors and the control/communication electronic:

• The JTAG/Serial connector permit to program the processor and to communicate debug messages.
• The Main CPU is a STM32 Cortex-M3. It is clocked at 64MHz from the internal oscillator.
• The 2 Axis X-Y Gyro is an invensense IDG500. It has an analog output and thus is connected to the CPU analog to digital converters.
• The 1 Axis Z gyro is an invensense ISZ500. As the X-Y gyro this is an analog chip connected to the CPU analog to digital converters.
• The 2.4GHz Radio is based on a nRF24L01+ chip from Nordic Semiconductor. It is actually a breakout board sold by SparkFun. The radio is connected to the CPU via a SPI port.
• The 3 Axis Accelerometer is a SMB380 from Bosch. It is connected to the CPU via an I²C bus.

Details of the bottom of CrazyFlie:

Here nothing particular, except that the little battery is taped on the bottom of the copter. The battery comes from a Silverlit x-twin plane and actually permit 4.5min of flight time.

Details under the battery:

Under the battery we mostly have power supply and the “power” electronics:

• The charge connector accepts 5V and permits to charge the battery
• The power switch permits to power the copter either from the charge connector or from the battery.
• The 5V step-up permits to increase the power voltage from the 3~4V coming from the battery to 5V.
• The two 3.3V linear regulators, reduces the 5V to power the copter electronic. The digital and analog 3.3V are generated independently. This was made to reduce the noise on the analog components (here it is the two gyros, a quite critical part).
• Finally at the four sides we have the motors transistors. The power for the motors pass on the top and bottom of the part of PCB that hold the motors.

It can be noted that we also have the battery charging chip just at the right of the 3.3V regulators. It is a classical MAX1555 and is working pretty well.

The idea with the step-up was to get a more stable voltage when the battery is depleted and also to remove some of the noise coming from the motor electonic. Anyway, this boars is capable to work without by using 2.8V linear regulators instead. So most likely the step-up will be excluded on the next version and we also plan to integrate the radio on board.

The motors and propellers are taken from a Silverlit X-twin plane. They are powered by the battery voltage.