Jack Crossfire's blog View Details
Archive for January, 2009
Posted by Jack Crossfire | Jan 30, 2009 @ 02:53 AM | 6,381 Views
The uBlox-5 slows down with high satellite counts. In the 12-15 satellite range, 3Hz is what U get. In 9-12 U get 4-5hz. Below 9 satellites U always get 5Hz.

If U must have 5Hz, U can limit the number of satellites with the CFG-NAVX5 command. Then to see what it's discarding, use the NAV-SVINFO command.

We have no plans to reduce the satellite count.

One other thing, trirotors & quad rotors have dissymmetry of lift. The compass can be made pretty accurate, but dissymmetry of lift is always going to be spiralling U around.
Posted by Jack Crossfire | Jan 29, 2009 @ 03:32 AM | 6,072 Views
Have a video of some model airplane wiring jamming a compass. This actually happened to the autopilot in flight before we moved the wires farther away.

Model airplane wiring vs. compass (0 min 36 sec)

The reason for this was to see if the magnetic deviation was identical regardless of heading. This shows U roughly how full throttle pulls the magnetometer the same amount no matter where it's pointing, good news 4 U. U can wrap the wires as tightly as possible & move them as far as possible from the compass, but U eventually need to add fudge factors to the calibration.

That sort of worked. Got spiral free & tight control initially. 3 batteries later, heading was off & spirals were beginning. Magnetometers suck. Whether fudge factors did anything is debatable. There must be a point where it's just accurate enough, but movie magic looks impossible with this airframe.

You'll be happy to know the 24 hour GPS is paying off. Got 15 satellites for a few seconds & 13 most of the time.

Haven't crashed the tri rotor yet. Almost time to warm up the flags.
Posted by Jack Crossfire | Jan 28, 2009 @ 01:23 PM | 5,962 Views
The answer is yes. The T frame has a strong magnetic dipole because of the long wire run. U get a nifty shift in compass readings when the motors start. The symptom of a compass error is spirals & velocity commands causing movement in the wrong direction. When the compass is dead on, with the 24 hour GPS calibration, U get extremely dead accurate flying.

Your best option is to strap it down & spin up the motors to full power. Add the difference in readings to the calibration factors U measured when it was idle.

Also those 9V batteries only last a few trips to the test range. Going to need more LiPos. U taxpayers better get another emergency cash infusion ready 4 Bank of Commumerica.
Posted by Jack Crossfire | Jan 26, 2009 @ 08:20 PM | 5,922 Views
So threw down $6 for all the 12" propellers sold locally to show U how inefficient they were. The theory was to put 1 propeller on 1 motor & compare the PWM with the other motors. If the PWM was lower, the propeller was more efficient & the flight should last longer.

That didn't work. For all the 12" propellers, the PWM was much lower but the battery died instantly. What happened? Lower PWM should mean lower wattage, but lower PWM sucked more power.

The theory is at low RPM with high load, the same coils stayed on a lot longer than they did at higher RPM with low load. The less frequently the coils switch, the more time they spend in the peak of the sine wave. Normally the same coils R only on very briefly & spend most of their time in the rising & falling edges of the sine wave. The gain in power from less frequent switching was more than the loss of power due to lower PWM.
Posted by Jack Crossfire | Jan 25, 2009 @ 03:57 AM | 5,934 Views
Soldered & shrink wrapped some more power supplies. Now we're up to 4 AC bricks for VicaCopter. 1 ground station supply. 1 intermittant supply. 1 continuous supply. 1 battery charger supply. The continuous power supply feeds just the GPS. The intermittant power supply feeds the lighting & computer. Booted into 13 satellites for flight. How do U like UAV's now?

Leaving the superbright LEDs & main computer on full time is a bad idea.

Don't know if the uBlox backup battery was defective, if it even was a backup battery, or if the pros do the same power supply juggling. The hot start modes for all the GPS modules were never close to this.
Posted by Jack Crossfire | Jan 24, 2009 @ 06:44 PM | 6,592 Views
Sully Sullenberger came to Rain Ramon, so decided we might as well get some photos 4 U. The story was remarkable because plane crashes never have happy endings, yet somehow all the factors aligned just once. The only way to celebrate a story like that is to honor the pilot....Continue Reading
Posted by Jack Crossfire | Jan 23, 2009 @ 01:01 AM | 6,438 Views
So the uBlox manual says it retains Ephemeris data for 4 hours after shutdown. Theoretically if U have 11 satellites it should instantly regain all 11 satellites when power cycled. In reality, power cycles start over at some small number of satellites & either grind away for 20 minutes or have to fly to altitude before getting 11. It was the same for the EM-406. The coin cells have never given us instant 11 satellite coverage.

To get the most out of your ride, U need an uninterruptable main battery. Meet the 9V battery. Leave your airframe always plugged into AC in your dumpy apartment. Plug in the 9V & unplug the AC for transport to the airfield. Plug in the Lipo & remove the 9V for flight. Instant 11 satellite coverage!

What's it like to fly a system that instantly boots into 11 satellite coverage? It's freakish. Got 9 & 11 satellites right after the 2 battery changes. Did not do automated takeoffs since these batteries were focused on trimming. Autopilot seemed more stable than usual. V feedback required less gain. Altitude was more stable. A lot of this was also the umbillical rerouting & the presence of satellites directly overhead.

Handling was the same as the quad rotor. Yaw was just as stable.

Oh & 1 other thing. Flight times were 10% LONGER!!!!!!

Crummy 11.1V 4Ah: 8m22s
Good 11.1V 4Ah: 10m30s

Holy freaks of nature. Torque compensation on the quad rotor wasn't as efficient as we thought. Quad rotors...Continue Reading
Posted by Jack Crossfire | Jan 22, 2009 @ 02:20 AM | 6,193 Views
So the answer is yes. 2.4Ghz interferes with the GPS umbillical. The combination of 2.4Ghz & T frame was the most likely explanation for the loss of GPS in pentarotors. Moved the GPS umbillical away from the 2.4Ghz & it got 8 satellites indoors, after 20 minutes. Moved it towards the 2.4Ghz & it got 0.

Also had the PIC UART lock up again. The XBee sent it data but it didn't respond & showed a no signal status.

Unfortunately 4 U, the rain has resumed.

Oh & 1 other thing. Tried to control an HS65MG on the 200Hz PWM & it didn't work. So the Futaba S3102 & Hitec HS81MG run on 200Hz. The Hitec HS65MG only does 50Hz. U software stabilization freaks want the 200Hz servos. Unfortunately there will be no 200Hz attitude feedback.

Those HS65MG's reminded us how expensive it was to fly the T-Rex. After sucking paycheck after paycheck on broken shafts & landing gear, when the servos started burning out, it caused the B of A to go bankrupt. U taxpayers had to cough up $700 billion to cover cleaned out credit cards. The world went into a depression. Real estate investors started killing themselves. All that was because of the T-Rex.
Posted by Jack Crossfire | Jan 21, 2009 @ 01:50 PM | 6,481 Views
Tri rotor finally reached the test stand. Got a thrust gimbal which we can live with. It's based on the 1 rcgroups post which documented a thrust gimbal sufficiently to replicate it. http://www.yb2normal.com/tricopter.html

We're getting 200Hz PWM out of the thrust vector servo, one of the Futaba S3102's. In fact neither the Hitec HS81MG nor Futaba S3102 work at 50Hz when using the thrust vector header. U may not know what's happening, but it's real stable.

Now for the first time ever it's close up, focused, stable, & fully lit views of a thrust vectoring trirotor test.

First tri rotor tests (2 min 43 sec)

Posted by Jack Crossfire | Jan 20, 2009 @ 04:07 AM | 6,240 Views
2 commutes banging on thrust vectoring have really gotten nowhere. Another redesign of the thrust vectoring actuator & it's binding, flexing, wobbling. This was supposed to make it lighter & more balanced. At least we drove back down with a good servo mount. Better get ready to flip the electronics around & fabricate another crazy battery mount to balance this one.
Posted by Jack Crossfire | Jan 19, 2009 @ 04:59 AM | 6,735 Views
Since you're all now experts in 4 & 5 rotors, decided we might as well give U a full review of 3 rotors. No-one has ever documented many details about a tri rotor. They say it flies great & move on. The tri rotor weighs the same as the quad rotor. Probably won't be flying until Wednesday.
Posted by Jack Crossfire | Jan 18, 2009 @ 07:23 AM | 6,259 Views
Made up a bolt-on penta rotor. It only takes 1 commute to convert between 4 & 5. The hardest part is rotating & realigning the IMU.

U might say it's payload on demand. U can't easily increase the payload of a unicopter without expanding the tail boom, but U can design a quad rotor to take bolt-on rotors.

Unfortunately, once again the penta rotor went nowhere fast. The longest flight was 4 min & that was with the battery always below the cutoff voltage for normal operations. GPS once again never locked, regardless of whether the ESC's were on. Suspect the carbon fiber arrangement is causing some kind of ground loop.

The connection of the 5th rotor did indeed wobble. The wobbling caused resonant oscillation with the rate controller. Spreading the weight more evenly between the 5 rotors may have increased flight time a hair. Now we're pretty sure the use of angled nacelles to defeat torque is sucking the power.

We can eliminate inefficiencies in inline propellers from the list of culprits in the penta rotor....Continue Reading
Posted by Jack Crossfire | Jan 17, 2009 @ 08:44 AM | 5,887 Views
The penta rotor was disassembled in an accidental attempt to convert it into a 3 rotor. There R no plans to rebuild it. After debating 3 rotors & 6 rotors, went back to 4 rotors. 6 rotors seemed too noisy in a final revelation.

So it takes 2 commutes to go between 5 & 4. Takes a lot of soldering, a few miles of ny-ties, & a few acres of double sided tape.

There was no benefit to 5 rotors. In your wildest dreams, if U centered the weight exactly, it wouldn't be as efficient as a quad rotor.

Its only redemption is it was the only pentarotor which was ever built & flown in all human history. Someday someone famous in China will build a penta rotor & the world will beat a path to their door, but U saw it here first.

We did learn computer stabilization can make ANYTHING fly. Angled nacelles can replace counter rotating propellers. Unbalanced thrust doesn't generate yaw from roll. Angled nacelles R a stupid idea since propeller torque consumes 15% of power. U take out another 15% for sideways thrust to counteract the original 15%, sort of like commumericans....Continue Reading
Posted by Jack Crossfire | Jan 16, 2009 @ 01:30 AM | 6,646 Views
Decided we're not going to have enough time off in daylight to work out the solvable bugs, so the first flight was at night. She did indeed crash, but is still holding together. The standard crash is to flip over nose first during takeoff. She also tends to roll over sideways as expected.

Takeoffs require gunning the throttle with neutral pitch & pulling up as soon as the nose lifts off. Takeoffs always start with a nose dive. There isn't enough power to take off with the aft rotor pulling down. Once airborne & trimmed, flight is like a normal copter. Landings R easy. There's no dreaded roll/yaw coupling. That was a test stand artifact.

Yaw control without any wind is the same as the quad rotor.

4Ah 20C is not eough. You'll get 1 minute at a time. Landing gear probably has 2 more battery sets left before it falls apart. Couldn't get any GPS locks with the uBlox. It's either damaged or someone is using 1.5Ghz....Continue Reading
Posted by Jack Crossfire | Jan 15, 2009 @ 12:20 AM | 6,208 Views
So we're not going to fly the 5 rotor until the next hour off in daylight. That could be anywhere in the next 1000 years. Going to have video rolling for the takeoff & crashing.

In the mean time, most of U only fly because U want to make money off it someday, just like how we were crazy about html coding in 1997 & sick of it in 2001. So U get the business blog.

Back in our musician days, we learned the hardest pieces we could find: Rachmaninoff's sonata, the Liszt sonatas, Liszt Etudes, Schumann's Tocatta. Teachers always said we needed simpler pieces to get publicity. We never had to make any money, planned on quitting in a few years & didn't care if no-one ever heard of Transcendental Etude #12, so no-one ever heard of us & we never made any money.

VicaCopter is the most glamorous of glamorous hobbies. Everybody wants to make a powered autonomous VTOL. We bought everything except the uBlox-5 using our own money. We set our own agenda. There was no reporting to anyone. No odd projects for sponsors. No McRosoft. No babysitting airheads. No booth duty. No marketing.

That wouldn't happen if we wanted to build a VicaCopter business. It's almost impossible to build any kind of business with your own money. You have to get sponsored. You have to do the goofy side projects to raise money & get publicity. You have to use Windows. Your product has to feel accessible to the most people.

Everybody wants to make a powered autonomous...Continue Reading
Posted by Jack Crossfire | Jan 14, 2009 @ 01:35 PM | 5,695 Views
First pentarotor tests (3 min 8 sec)

With the 5 rotor on the test stand, noted the following.

Strong coupling between roll & yaw. Every attempt to roll causes a yaw. Aft rotor is tilted in rolls, causing a yaw for every roll. Requires lots of positive pitch trim.

Going to need full time attitude hold, & more Castle Creation tuning.
Posted by Jack Crossfire | Jan 12, 2009 @ 08:59 PM | 4,567 Views
Our state of the art computational fluid dynamics system showed the penta rotor would suffer from extreme weather vaning of the tail rotor. Yaw provided by the 2 front propeller torques would not be enough to overcome weather vaning. In the worst case, U would have 2 front rotors stopped & be completely unbalanced again.

Moving the tail rotor closer would make no difference. You'd watch a line of 3 propellers mysteriously turn perpendicular to the wind.

U can pile on a servo to control yaw by deflecting tail rotor thrust & add more weight.

The main problem is preventing rollover on the ground. U definitely need heavy rotor shrouds. Just because Draganfly sells underslung propeller craft for $17,000 doesn't mean you're not going to break tons of propellers. Notice how draganfly.com shows plenty of hand launches but never the landings?

Finally, we have shortened flight time from lifting a 5th motor.

There's not much U can do to increase quad rotor performance. U can reduce flight time. U can increase weight. U can reduce payload. Everything you can do to the basic 4 rotor design is just moving performance from 1 area to another.

Your best option is more software to handle unbalanced payloads gracefully. Others don't clamp PWM values. They reduce collective when a PWM is saturated. Thought that was stupid so we didn't implement it, but clearly a powered attitude controller is different than a swashplate.