I recently built a foldable Quadrocopter for FPV, the KlappQuad.

It’s basic form is inspired by the FQuad from Warthox. I own one of these and like it a lot, but I always wanted a foldable version. Also, in order to use a brushless gimbal for the camera, the frame needed some other modifications.

The KlappQuad has a 3 piece body. The lower two plates form the rigid body, holding the foldable arms and the landing skid in place. Also the RC receiver, the flight controller and the AV sender are mounted on this rigid section of the copter.

On top of this section, hold by four soft rubber spacers, a third plate is mounted, holding the gimbal assembly (including it’s electronics PCB) and the LiPo battery (to provide inertia). Since they need to have clear view to the sky, the GPS antenna from the flight controller and the OSD (also containing a GPS antenna) are mounted on the top plate. The rubber spacers (in conjunction with the high mass of the LiPo battery) provide decent vibration decoupling of the camera from the rest of the copter.


I use a Naza Light as flightcontroller and it’s GPS antenna is supposed to be mounted on a small spike in order to get best reception. To minimize the folded size of the KlappQuad, I mounted the GPS antenna on a foldable retainer:


I machined this retainer out of brass stock on my lathe. The three pieced part took about 2hrs to produce (I’m somewhat out of practice and was happy to have something to machine after a while).

Folding the KlappQuad is a little bit more sophisticated than simply folding all four arms backwards:

In order to minimize the folded length, the two arms on the back are not only rotated to the back but also retractable. The arms are not hold in place by a single bold (like the front arms) but by two bolt heads, running in grooves on the under side of the arms. Milling these grooves was a nice and easy job for my Pleasant Mill…

One bolt head is guiding the shift and rotation of the arm with the long groove visible in the image above. The second bolt holds the arm in is “flight” end position. The groove for the second bolt is open at the end of the arm and visible in the following picture of the folded copter.


The above picture also shows the position of the front ESCs. The both ESCs for the back motors are concealed between the two lower plates in the back of the copter.

Since all arms slide partly between the two lower two plates when folded, using cable binders to mount the cables from the ESCs to the motors on the arms wasn’t an option. Instead, I drilled small holes into the arms (1mm diameter) and used 0.8mm silver wire to fix the cables to the side of the arms.

Here are some pictures of different stages of folding/unfolding the copter:


The folded copter measures LxBxH 47 x 14 x 14 cm with props and 36 x 14 x 14 cm without props. So it fits easily into a small suitcase (including RC transmitter, and some clothes).

The total weight (ready to fly, including Gimbal, GoPro Hero 1 and a 2400mAh LiPo battery) is 1150 gramms.

Last but not least:

It flights great so far. Here’s a video I shot entirely with the KlappQuad. I wanted to know if it would be possible to use this copter to shoot an “action scene”. A friend of mine an his son were so kind to help me out as actors :)

The mill’s soul

IMG_2456.JPGI just published v0.4 of Pleasant Mill‘s firmware on Github yesterday:


As you might guess from the version number, this firmware isn’t finished yet and probably still has some bugs to fix (help’s highly appreciated, BTW :).

As mentioned before, I use a Seeduino Mega and some Pololu A4983 stepper motor drivers.

One recent addition on the hardware side was a 24V/6.3A power supply. I also glued some small heat sinks on the A4983 chips.

Then I increased the current to 1.68A, the max current, the stepper motors (SY42STH47-1684B) are rated for. Driving the steppers at 24V/1.68A enabled me to dramatically increase the top feedrate for the machine (about 1800mm/min instead 600mm/min). Unfortunately, the little heat sinks weren’t enough and the thermal shutdown of the A4983 kicked in after a few minutes. So I had to lower the current to about 1.2A again (and the max feedrate to 1100mm/min).

As you can see in the image at the top of this page, I’m still using a messy bread board for the electronics. As soon as I have the feeling to know the main issues with the whole shebang, I’ll build some kind of Seeeduino Mega shield, of course.

Right now, I plan to put the stepper motor drivers on a separate daughter PCB, somehow integrated in some kind of big ass heat sink.

If anyone knows a good way to cool the tiny A4983 chips on the Pololu breakout boards, please let me know!

The just published firmware release contains code for the G2/G3 commands (arcs) and for several drilling cycles (G81, G82, G83, G85, G89 and G73, see video below).

The drilling cycle commands also recognize the L (loop) parmeter, which is nice to drill multiple holes in a row. Just switch to incremental positioning (G91) and use something like G81 X10 Y0 Z-12 R1 L7 to drill 7 holes along Y=0mm, 10mm apart and 12mm deep. Don’t forget to switch back to absolute positioning (G90) after :)

I also added the M6 (tool change) command. Of course, I have no automatic tool changer (and I don’t really plan to build one). On the Pleasant Mill, the M6 command pauses the G-code processing and displays a message on the LCD, containing the tool number, requested by the G-code program and waiting for feedback from the operator.

For example, the line

M6 T3

results in


I even wrote some code to save a “tool database” in the Seeeduino’s EEPROM, so the LCD would say something like “Insert tool: 3mm drill bit”, but this code doesn’t work properly yet.

Both, the drilling cycles and the tool change command are showed off in the following video:

And here’s the G-code, I sent to the machine during the video:

G0 Z3
G0 X2 Y2 ; Simple drilling cycle
G81 X6 Y0 Z-8 R2 F300 L4
G0 X2 Y7 ; Drilling cycle with dwell
G82 X6 Y0 P500 L4
G0 X2 Y12
G4 P1000 ; Peck Drilling cylce
G83 X6 Y0 Q2 L4
G0 X2 Y17; Peck drilling cycle, high speed
G73 X6 Y0 Q3 L4
G0 X2 Y22 ; Drilling cycle, slow retract
G85 X6 Y0 L4
G0 X2 Y27; Drilling cycle with dwell, slow retract
G89 X6 Y0 P500 L4
G0 X0 Y-25 Z19.5 ; Tool change position
M6 T2 ;  Change Tool
G0 X2 Y32
G81 X6 Y0 Z-5 R2 L4
G0 X2 Y34
G81 X6 Y0 Z-5 R2 L4
G0 X0 Y25 Z13.5

I also implemented the commands G54 to G59. Each of these commands switches the machine to one of 6 “work coordinate systems” (WCSs). These are 6 user defined “zero positions”, saved in the machines EEPROM.

To define a WCS, use the “Jog XZ”, “Jog Z” and “Jog AB” functions in the mill’s UI (“Cartesian”) to move to the desired position. Then choose the menu command “Set WCS” to save this position in the EEPROM. You also can view already saved positions with the “Show WCS” command.

When writing G-code, you can then use one of the commands G54 to G59 to load the previously saved positions as zero positions.

In order to get the whole WCS stuff working, you need to “home” the machine once (either by sending the G28 command or by choosing “Find home” from the Cartesian menu in the UI). This is necessary to give the firmware an idea of the machine’s absolute zero position. If you try to save a WCS position in the UI or to use G54 to G59 without homing, an error is displayed.



Two tricks

IMG_0399.jpegIn this post I’d like to show you two little tricks concerning 3d printing. I didn’t invent the first one (it’s rather a well known procedure to stick foil to smooth surfaces without bubbles), but as far as I can tell, the second trick is something new.


1. No bubbles, no troubles

When printing with a heated build platform, it turns out, that Kapton tape is a great surface for ABS printing. The only problem is, that normal Kapton tape is rather fragile and easily get ripped off the build platform when removing printed parts. Therefor I use 10x10cm sheets of glass with a layer of Kapton tape on top as exchangeable build surfaces on my heated platform.


One problem is to get the Kapton tape on the glass, well aligned and without bubbles.


The trick is to use soap, water and a scraper. Here’s a short how-to video, I made:

2. Snap-in, not snap-off

I recently designed a printable Tricopter:IMG_2055.JPG

One special thing of this design is, that the Tricopter is foldable for easy transport:


For this, the printed center piece (below the plywood platform) has two snap-in hinges for the front arms:

I’m sure, there are several other applications for this technique.

Speaking of technique, slightly off topic, but maybe also interesting:
A “mechanical disadvantage” of Tricopters versus Quadrocopters is, that in order to countervail the unbalanced torque of the three propellers, one of the motors needs a tilt mechanism (Quadrocopters use two CW propellers and two CCW propellers to self balance the propeller’s torque).
This tilt mechanism is usually one of the more complicated parts to build on a Tricopter. Here’s the tilt mechanism I designed for my printed Tricopter: