CAL Multirotor Blog

If you haven't worked with carbon fiber before, you may be surprised to learn that is a surprisingly good conductor of electricity.  Not as good as aluminum, but far better than most materials. 

For example:  I have a piece of 16mm OD, 13mm ID thin wall carbon fiber tubing.  A piece 16" long has a measured resistance of less than 10 ohms.  I put 5 Amps through the tube, and it got hot quickly, so you could use a carbon fiber tube or sheet as a heater - if you keep in mind that the stuff can't tolerate high temperatures unless you buy special high-temperature material.  Using high-temp material, you could build a heated bed for a 3D printer. 

In building multirotor aircraft, I like my designs to have a clean look, which usually means running the motor wires through the arms. The ends of carbon fiber tubes are sharp - sharp enough to cut through the silicone insulated wires that I use.  So I had to print some special plastic grommets to slide into the ends of the tubes. The grommets are rounded, so it won't cut the insulation, and even if it did, the plastic grommets are non-conductive.

I was testing my foldable quad. It was powered by the battery, but also connected (via USB) to my computer.

I noticed that my computer was rebooting - and smoke was coming from my flight controller!

I shut everything off and detected that horrible "something is terribly wrong" burning smell coming from the flight controller.

I disconnected all wires and did a short check.  The battery (+) lead  (16V) was shorted to the 5V power feed to the controller!  The little "buck" regulator on the power module - which measures battery voltage and current and also provides 5V to the controller, had failed and given the system 16V instead of 5.

The controller is supposed to be protected against such things, but after inspecting the controller carefully, I found that the protection diode had not even been installed!  Note to self - do not buy cheap Chinese-made boards.  But wait.... I HAVE to buy cheap Chinese made boards, since that is the only place they are made.  But next time, I'll inspect the PCB before I install the replacement.  If there is no protection diode (Zener), I'll add one.

So, I inspected the new flight controller (and added the Zener) and installed it.  It seemed to work, but the compass and GPS did not. I checked the GPS/compass module and - sure enough, it was "blown" as well. After I took off the cover, I found a black spot on the board where the voltage regulator normally sits.

So, I replaced the GPS/compass unit and re-applied power.  But the flight controller STILL did not recognize the compass!.  After doing some testing, I found out that when the original GPS/compass unit was destroyed by the over-voltage, it shorted and fed a solid 5V to the data lines going back to the controller.  When I put in the new controller, the old GPS/compass unit had fed power to the data lines of the NEW controller, destroying it! 

Next note to self:  Check all components in the system individually after a failure. Don't just replace the part you feel is defective.

So, I replaced the flight controller again (this was getting expensive), hooked up the new GPS/compass and that all worked.  Great!

But the camera gimbal didn't After checking, I found that the 16V applied to the 5V control line of the camera gimbal controller had destroyed that!

What a mess!  Fortunately, I had a few other devices on-board that were powered separately (GoPro, Raspberry pi, 4G LTE dongle).

Final note to self:

If you run redundant power to your controller (which I always do), diode OR them with Schottky's. 

Use hefty 5.6V Zener or 5V TransZorbs  on every 5V powered device. Flight controller, GPS/compass, radios./gimbal controller.

Add 3A fuses in line with the flight controller and gimbal.

I took my foldable quad to Mexico last week to get in some non VLOS flight time.  The first two days were extremely windy, so I didn't even try to put any 'up'.

But on Wednesday, the conditions were perfect, so I got up early and went to a vacant soccer field for my launch point. I programmed in a path that took it far out over the ocean (4 miles) and back.  I wanted to take some pictures of a rock outcropping that existed out there. I sat the quad down on the sand of the soccer field and waited for the GPS to get a lock, which took only about 20 seconds. I lifted it off manually, and when the copter was flying nicely above 'ground effect' range, I switched to AUTO.  It went straight up to its normal flying altitude  100Meters = 328', and headed out over the ocean.  

It hadn't gone  more than a mile when I got the signal on the telemetry channel "radio signal lost".  That was OK, and I expected it. This had happened dozens of times before.  I almost always fly further than the range of  my control radio (2.4GHz).  The telemetry channel is at 915MHz, and has approximately 2.5 mile range.  

But none of that matters much - once the aircraft is in AUTO mode, it doesn't need any radio communication to/from the ground at all. It is guided only by GPS, compass, gyroscope and altimeter.

So even though the telemetry signals was fading in and out, I wasn't concerned. It was a normal occurrence.

I got a signal from the craft just as it was turning around (about 4 miles away), and was surprised that I got any signal at all, but a signal is always comforting - you know it is doing what it was programmed to do.

About 15 seconds after it turned around, I got another message from it.  "LANDING MODE".

But it was over the ocean!  What was going on? I took my laptop (which has a telemetry antenna attached) and held it higher in an attempt to get a better signal to the aircraft.  But it wasn't responding.  I got one more message "70 Meters altitude".  That meant it was coming down.  It was actually landing in the ocean!  And there was nothing I could do. 20 or 25 seconds later, I got the message "no telemetry packets received for 10 seconds".  It had landed.  And promptly sunk.

I examined the log files which were stored on my laptop, but not totally complete because of the bad signal quality.  It appears that for some reason, the GPS malfunctioned.  And when a device that is GPS guided has a GPS malfunction, it simply doesn't know where it is.  The flight controller gives the user two options when there is a GPS failure  1. Try to hold your current position as much as possible or   2. LAND.   Unfortunately, I had mine set to number 2.  That is a good option when flying over land, but a very bad one when flying over water.

If I had used setting number 1 instead, I could have moved my antenna around in an attempt to get a signal to the quad.  It would take only one short command to get it to climb, a good thing to do since higher almost always gives better reception.  And once I had a reliable radio connection I could have brought it home (and down to the ground) manually.

So.. I'm already planning on a rebuild.  And it will have a better and lighter airframe this time.  And it will have 2 (redundant) GPSs.

My friend, Andrew Bloedel has come to the Bay Area several times recently.  When he saw my 'fliers' he got the "bug" to build one himself.  I told him that I had lots of parts, and could give him most of the stuff needed.

I looked at all the pieces I had lying around and realized that it would take some time to build one using my "leftovers",  so instead, I bought a partial kit - an S500

https://www.ebay.com/i/122221271700?chn=ps&dispItem=1

This $139 kit has everything needed except the radio, controller and battery.  It doesn't include a gimbal or camera.

It comes with an APM Controller, GPS+compass, 30A ESCs, 2212 920KV motors and 10" GemFan props. The frame is fiberglass with carbon fiber rods for the landing gear.

We put it together.  I designed and printed a new battery holder so that it could carry a big battery - 8000mAH/3 Cell.  We did the assembly and installed the radio. 

It flew fairly well,  but on one flight, we did have a glitch with the compass, but I fixed that by adding a 220 ohm pull up resistor to the SCL line.  The craft was low on power, and while it flew, it definitely had problems when the battery voltage got low.  Below 11V, the motors had barely enough power to keep it in the air.

So we modified the arms slightly by drilling out some holes and changed the motors to SunnySky X2614 / 1100 KV models.

What a difference!  It really flies well now, and climbs like a rocket. 

I have been searching for a good controller (ground side, handheld) that will work with Mission Planner.

The idea is:  

Most everything I'm working on involves long-range communication and control. That is why I'm using 4G.  A 4G/LTE link works great for telemetry and live video, but I want control as well. I want to be able to fly in "manual" mode when I want to.  If my craft is in AUTO mode, and is flying on its own somewhere, and I see something that I want to look at more closely, I want to be able to switch to manual mode, go over and take a look, and then go back to AUTO mode. 

Everyone is using 2.4Ghz for control.  That is too limiting.  I want to use 4G for CONTROL as well.  That means that I have to send that control information over a protocol called MavLink.  My laptop computer can generate MavLink packets and send them to the craft over 4G.  But how do I get my commands into my laptop?

The program I use in my laptop (Mission Planner) has a provision for using a Logitech C310 game controller to generate commands. This thing is a joke!  It does have two short "joysticks" but both snap to the center when released.  That is OK for a game, but for a copter that is armed and sitting on the ground, slipping off the throttle will make it take off.  That is dangerous.

I tried to fly using the Logitech controller.  I don't think I could ever learn to do it.

There are some other controllers that look good - they are used for flight simulators and have two joysticks.  But they have a very narrow range of outputs (the maximum to minimum values are not adequate).  And only one or two have more than 4 channels (roll, pitch, yaw, throttle).  But I need at least 5 since I also need a MODE (auto, manual, return to launch, etc).  

I have contacted Logitech and several other companies to try to get them to build a suitable controller. No one is interested.  So I have been working on my own, and it is almost working.  Hopefully, I'll get that going in the next week or so.

After working diligently on a 'Santa Cruz flyer' for much longer than I had planned, I realized that - the only way I could get there was to drop 3 batteries along the way ($300).  Not to mention that there was a small possibility that one of the batteries would fall and hurt someone.  And still - if anything went wrong - like an unexpected side or head wind, it STILL might not make it.  That raised the possibility that I might lose the whole craft somehow.  

Too many unknowns.!

So, I started thinking about something that had a higher energy density than batteries.  Like a petroleum product.  

A gas engine would certainly provide the required power for a longer time than batteries, but multirotors will not fly if all the props are running the same speed (and therefore have the same lift).  That is why - with my partner Stan Weiss we are building a gasoline-driven 'heavy lifter' that uses 6 variable pitch props. The pitch changes in order to keep the craft balanced.

But there is another, intermediate way:  Multirotors use approx 80% of their power for lift and 20% to maintain balance.  What if you used a gasoline engine to power 2 large belt-driven props running in opposite directions (to cancel out the rotational vector), and 4 smaller, electric-driven ones to provide balance.  Since the  electric 'balance props' only have to handle 20% of the lift, the circuits that drive them don't need to deal with hundreds of amperes.  In fact, they don't have to deal with currents larger than a normal large quadcopter.  These parts are available off the shelf.   And the flight controller doesn't even need to know that 80% of the lift is being handled by the larger props.  

Since the big props are belt-driven, the mechanical losses are low. The gasoline engine does have to run at variable speeds, so that it always maintains approx 80% of the lift, regardless of the throttle position.  A servo connected to the motor throttle can handle that with a new output from the flight controller.  I can write that code.

I know you are getting ready to ask how the electric motors get their power.  Easy - from an alternator. A brushless DC motor can be used as a lightweight, high-efficiency alternator.  The only problem is that the output voltage varies with the motor speed, and i- in general, alternators aren't very efficient (approx 45%).  But with good electronics, the voltage output can be constant and the efficiency can be greater than 85%. 

I did the calculations, and I came up with the following:

Gasoline engine:   RCGF 56cc 2-stroke model airplane engine  5.3HP @ 7500 RPM

Alternator:  NTM 5060 380KV   2665 Watt brushless outrunner 

Props  Quan 2   20" X 8" pitch.    Quan 4  12" X 6.5" pitch.

Controllers - PixHawk + Rasperry pi +  Teensy3.2 + PIC18F8723

A picture of the test setup as it is being built is below: