3D Printer!

For Christmas my dad gave me a “voucher” I could put towards the purchase of a 3D printer of my choosing.  After about a month of research and restless nights dreaming about filament deposition techniques I decided on the FlashForge Creator with dual extruders.  It is basically a clone of the MakerBot Replicator 1.

The other finalists in my search were the Solidoodle, Ultimaker, and PrintrBot Plus, as well as the myriad DIY RepRap kits and other Replicator clones.  The Solidoodle is a very nice machine but I really wanted dual extruder capability so I could print dual color and eventually dissolvable support for complex shapes. The Ultimaker and PrintrBot both have dual extruder upgrades but the former is expensive and has to take a several week journey from Holland and the latter has a tendency to knock prints over and get caught on things on your desk because of the moving build plate.  I seriously considered getting a RepRap like the Prusa but with so many variations and such unorganized documentation I didn’t want to spend a month just building it and another month tweaking and upgrading.

web_IMG_1232The Flash Forge Creator had great reviews and all the features I wanted at a price I could afford.  Plus, Amazon had it in stock and it arrived the NEXT DAY after I ordered it (thank you Amazon Prime)!  It was covered in enough shipping labels and customs stickers to make even the most seasoned delivery driver’s head spin.

web_2014-01-15 22.32.20The machine itself was packed well for it’s journey overseas.  So well in fact that unpacking it was a bit challenging.  I finally got it out of the box and set it up and was printing within the hour.  Talk about instant gratification!  It took longer by far to sand and urethane that nice table top for it to sit on.  Tucked into a corner of the box I found a little stuffed reindeer dressed in a sweater and scarf. . .  I’m guessing it was a Christmas gift, thank you FlashForge!


web_DSC_6542The setup process was super simple, just bolt on the extruder assembly and run through the plate leveling procedure.  Installing the spool holders and loading filament took another couple minutes.  I was so excited that when it started printing I just sat there and watched it for about 45 minutes.  In fact everyone in the house came down to ooh and ahh over it.

Browsing through the models on Thingiverse and YouMagine and Cubehero  and My Mini Factory was a bit overwhelming and I didn’t know what to print first!  Here are a few of the first prints I made in the following days while I got used to MakerWare and ReplicatorG software.


A GoPro head mount, still on the raft.

GoPro Head Mount


A secret heart box for valentines day.  This piece is amazing because it prints working hinges and sliders all together, no assembly required!



The yaw mechanism for a tricopter.

yaw mount 2


A tripod quick release plate for GoPro mount.

GoPro Slik mount 2


I’m still trying out different 3D solid modeling software packages for making my own parts.  I use Sketchup for architectural design but as soon as I tried to use it for parts modeling I realized it was not up to the task.  So far I’ve tried 123D Design, 3D Crafter, NetFabb, OpenSCAD. But I’ll probably end up using SolidWorks or Inventor.  Any suggestions are welcome!

Foam Cutter

I’ve wanted to make one of these for a long time now and I finally found an easy, cheap solution. That plus the motivation of making this as a Christmas gift got me going.  I wanted it to be portable and inexpensive but durable and versatile. If I can buy one at the store for $20 then the one I make myself has to be much better and cost less.

The cutting wire is René 41 27AWG. This wire is made of Chromium and Molybdenum instead of the more common Nickle Chromium resistance wire.  René wire is stiffer, springier, and doesn’t sag at high temperatures as much as NiChrome.


The power supply is an old AC/DC 12V, 1.5Amp brick I found that has a nice long cord.  The supply goes into a brushed speed controller with a servo tester to vary the output power.  I tested it on my bench supply and at 12V it draws 1.25 Amps. With the AC/DC brick it draws about 1.1Amps.


I made everything modular with 3mm barrel plugs and .1″ pin header connecters for future improvements. Unfortunately, I had to glue the PVC together in order to make it ridged enough. Oh well.



My initial design had it powered from either wall power or internally rechargeable batteries. Unfortunately I couldn’t find quite the right circuitry so I had to ditch that idea.  Here’s the finished product powered up and ready to cut some foam!


Museum Garden

The other day I was asked to take aerial photos of the Museum Garden at The National Colonial Farm.  Most of what I’ve done till now has been video and I have a very good setup for the GoPro.  For this job I wanted higher resolution stills aimed directly down which called for a real camera.  I also now have more faith in my skills and my equipment so I decided to use my Panasonic LX3 (aka Leica D-Lux 3).  Using some aluminum stock I fashioned a 90 degree mounting bracket.  The bracket is offset like that in an effort to keep the center of gravity in line.

LX3 mount_back

Now the LX3 has no IR receiver nor cable release so I had to trigger it the old fashioned way, by pushing the button.  The last time I made a servo trigger I crafted an awkward bracket out of Shapelock thermoplastic that was heavy and ugly and flexed under stress. luckily the LX3 has a hot shoe and I had a hot shoe to 1/4 20 lying around.  Attached to that, I had a piece of Lexan and a scrap of some other flimsy plastic, a 9g servo, and a short piece of silicone fuel tube. All cut, drilled, and fastened and voila!

LX3 mount_servo

What you’d see from the ground.

LX3 mount_front

And here’s the resulting photo

museum garden_small

It’s actually a composite of several shots because I didn’t have FPV live video set up for that flight (I couldn’t find the video out cable for the LX3) so I had to kinda fly back and forth trying to hold the copter steady while pulling the trainer switch to take pictures.

Next time I plan to have at least a down facing FPV and possibly a co pilot to help guide me into position and take the photos. Also, while I did use altitude hold mode for this, I had no idea what my framing looked like but alt hold combined with live video feed should make future shots much clearer.

Maine Wedding Crash Report

My previous post has the pictures so skip this if you don’t care for technical details.

I got to the venue a little early to set up.  Unpacking the equipment and assembling the quad took about 25 minutes.  I decided to start with the two 3s 4000mAh nano-tech batteries hooked up in parallel.  Telemetry and video link were solid and GPS was getting 9 satellites in view. Takeoff was smooth and I had fun tooling around in stabilize mode.  Altitude hold mode proved to be more stable for general video pans. I never got a chance to use circle mode.

flight path

My flight path imported into Google Earth. Each color represents a switch in flight mode. Yellow is altitude hold, green is loiter, purple and red is stabilize, and blue is land.

Relative Altitude

As you can see, the flight controller switched into land mode at the very end.  I had a battery failsafe setup to land when the voltage fell below 10.4V which is 3.5V per cell.

voltage graph

When I saw the voltage start to drop  I flew back to the starting point and was preparing to land when all of a sudden I lost all stick inputs and the copter accelerated into a smooth descending arc and went straight into the side of the building.

Roll graph

You can see my stick inputs in red and the flight controller’s motor outputs in green.  They track pretty well up until the end where it just ignores me.  This is actually normal for failsafe modes. What is not normal is the acceleration and the huge arc that you see in the next two images.

Ground Speed

flight path2

I have not had time to reassemble the copter since then. I had never tried land mode before and I want to see if it works correctly or has a glitch.  Perhaps the low battery voltage caused the glitch. The GPS had a solid lock and all other sensors, altimeter, magnetometer, gyro, were all functioning correctly according to the log file.  In any case I would like to recreate the events that led to the crash.

Maine Wedding Photos

I got back from Maine over a month ago, so it’s about time I posted about the trip.  Overall it was a success but I did have a pretty severe crash. Luckily I had redundancy.  I’ll get to the crash later, but first, lets take a look at the photos!  Most of these are stills from the video which is coming later.

Cutting it Close

Last week I took the new copter out to test the GPS functions. Everything was going well; the GPS locked and could hold a loiter fairly accurately, simple mode was working well meaning the compass was not being interfered with by the high current lines running right below it.  Then I switched it to RTL (Return To Launch) mode.  The copter started ascending as it should but just kept going.  I started getting concerned when it passed 60 meters and I decided to bail.  I later realized that I never changed the default RTL height of 100 meters! When I switched it back to manual mode I made the rooky mistake (it was my first time after all) of not setting mid throttle and the copter started to drop.  Being 60m high I had plenty of time to recover, I gave full throttle and it stabilized for a second but then all the motors cutout and it started to drop again! I backed off on the throttle and it love tapped the ground but seemed fine. I continued flying but about a minute later one of the motors popped off and the whole thing turned into a deranged nylon-clawed implement of tumbling destruction.2013-08-25 motor mounts  I had mounted the motors with the same zip ties I used on my other builds but the difference here is that this copter is twice as heavy and twice as powerful and the thin plastic couldn’t hold on.  I now have them hard mounted with aluminum plates.

2013-08-22 crash_smThat moment when weeks of hard work started to fall out of the sky I realized just how out of place this over weight piece of flotsam is in the sky and how badly it wants to be back on the ground that is the origin of the wood and metal it is made from.

The only damage was the landing gear and one arm were smashed and of course the props.

Technically speaking I’m pretty sure the battery couldn’t deliver enough current thereby sagging its voltage below the ESC cutoff threshold.  The battery I was using at the time was a 5000mAh 4S 30-40C so it should be able to push 150-200 Amps.  The current sensor said 147 Amps but everything happened so fast I may not have seen an additional spike or it may have happened too fast to register or the sensor I was using couldn’t handle it (it was the Chinese clone of the 3DR 90A current sensor). I have since upgraded the current sensor to a genuine 180A AttoPilot board.

Here’s the new and improved version with lighter, folding landing gear.

This new design weighs in at 2.1kg with the battery but without the camera.

Preflight checks at the field:
setup 2013-09-02

And it’s up!
flying2 2013-09-02

Hands free Return To Launch (RTL) set at 20m/2m this time.
hands off RTL 2013-09-02

On board camera.
aerial1 2013-09-02

I had quite an audience!

The carry case:
2013-09-02 carry box

The case is hideously ugly, inside and out, and smells like evil dinosaurs.  But it was free so I’ll just break out the Febreze and maybe line the thing with dryer sheets.  On the up side, everything fits, but just barely.  I timed myself setting it up and from closed box to flight ready including mounting the camera and preflight checks it clocks in at 12min.  Break down is about 6 min.

Still left to do:

  • Finalize the preflight checklist
  • Finish ground station / FPV box.
  • Tune camera gimbal
  • Tune ArduFlyer PIDs
  • Give the copter a name (it’s tentatively called Zippy after the Ziplock canopy)

I have one week, that’s cutting it close.

Heavy Lifter Quad

The first frame I built for this model was designed very carefully to fit inside a hard case I had while still maximizing wingspan. Unfortunately the frame was not rigid enough and would oscillate badly rendering it unflyable.

original frame 1

original frame 2

In order to test the electronics I cut out a  simple square and screwed the arms to it. It was supposed to be temporary but with so little time left for testing and tuning I may end up flying this setup at the wedding.  The wingspan is much shorter than I would like but I think with proper tuning it will be stable enough.  I made a test flight with the brushless gimbal and it is amazingly smooth! Even with no tuning and some heavy I-gain oscillation the video looks like cinematic crane shot. I’m using the KK2.0 board for testing so I’m not going to bother tuning until I get the ArduFlyer system in there.

heavy lifter quad 1

landing gear 4

This frame was really just thrown together with no design time put into it. It ended up coming out quite nicely if you ignore the long wires sticking out of the sides that resulted from shrinking the frame.

The landing gear you see there had a bit more thought put into it.  The whole thing is stuck on with heavy duty velcro and is surprisingly secure. The camera plate is removable by just unscrewing two screws. There’s plenty of clearance and the whole thing is very sturdy and weighs in at just 275 grams!

landing gear 1 landing gear 2landing gear 3

The all up weight with camera, gimbal, battery, and landing gear is 2.2 kg.  That may seem like a lot but the static lift capacity I measured is about 5.5 kg so it actually feels fairly nimble in the air.

Accokeek Video Teaser

Here’s a little clip I put together a while ago from some old video.  This was shot with the Hero3 on my quadcopter running KK2.0.  I made this before I learned how to use Deshaker for VirtualDub so don’t watch it if you easily get motion sickness. . .

Accokeek2012 1080p30 from robert on Vimeo.

I’m working on something much more eye/ear friendly so stay tuned.

ArduFlyer and Wedding Prep

I have been recruited to take some aerial video/photographs of my sister’s wedding in September.  I want to have more control and failsafe for flying a heavy multirotor in a crowded environment than the KK2.0 can offer so I decided to go with the ArduPilot Mega 2.5 platform (APM).  I got the RCTimer ArduFlyer clone from UAVObjects.com complete with GPS, two way telemetry, and MAVLink On Screen Display.  This flight controller gives fully autonomous flight with fully configurable failsafe with return to launch and a dozen other flight modes like loiter, altitude hold, circle, follow-me, stabilize, super simple (smart orientation like the DJI Phantom has!), and GPS waypoint flying!  And all this on an open source, arduino based platform for around $190!

I was hoping the APM would be mostly plug and play but I’ve spent most of the last couple weeks soldering wiring harnesses and configuring software.  For example, this is the harness I made to connect the on screen display (OSD) to the first person view (FPV) camera and the video/audio transmitter and supply clean power.


The whole thing is pretty much ready to go, all I need now is to finish building the frame.  If this frame doesn’t workout, with such a short time frame I may have to buy a premade frame. I want to have at least two weeks to flight test it and work out the bugs and get familiar with the new flight characteristics before showtime.

Stay tuned, more build logs and mods to come!


Flight Controller Mount

Here’s a quick and easy vibration isolating mount for a flight controller.  It uses foam poster board, rubber bands, a paper clip, and moon jell (or a sticky hand toy).  The KK2.0 Flight Control board weighs about 21 grams and therefore it takes very little energy to get it vibrating.  This means the isolation material must be extremely soft. I used one square here but I may double it up in the future.  I have seen other designs that use rubber bands attached to a frame to suspend the FC but I don’t have enough room on my flight deck to implement that.  I still use rubber bands here but this design only adds about 5mm to the length and width and about 12mm to the height.

First, slap that moon jell to the back of your board.  I realized that the gel doesn’t stick out far enough to comfortably clear the pins on the back so it will need to be spaced up from the mount.

2013-05-30 18.16.15_sm

Then measure and cut a square of foam board to about 10mm larger than the flight controller board.  Mark where the holes in the FC will go and cut slots for the rubber bands. Then cut a smaller square that the moon jell will rest on and glue it in place.

2013-05-30 18.16.00_sm

Cut up a small paper clip and bend the ends over so they can catch the rubber band.  Poke them into the sides of the board to test fit.  Remove them, add glue and reinsert permanently. I should have bent them in to a zig zag first because they have a tendency to turn in the hole.  The rubber bands are not really under tension, they are only there to keep the FC from coming loose in flight.  The moon gel is very sticky and holds the board in place.  I chose to use rubber bands to hold the board down because anything rigid would transmit vibration more readily.

2013-05-30 18.56.44_sm

All my electronics and doodads have velcro to keep them in place. The removable pieces (battery, receiver, flight controller, etc) use loop side and the base has hook side.

Here’s the finished product ready to be reinstalled into the quadcopter.  It weighs about 4 grams.

2013-05-30 18.48.33_sm

This whole project took about 20 min including “design” time.  Really there was not much design.  In fact it’s taken me longer to write up this blog post than it did to finish the mount.  I built this on a whim when the velcro on the original mount (some cut up mouse pad glued to the back) pealed up.  I may be imagining things but I think this helps the craft fly more stable.

I would like to see the next firmware add some sort of data logging feature that, among other things, would record how much vibration “noise” it was experiencing.  As far as I know, most of these vibration isolation designs are educated guesses and hokus pokus.