Just a bit of a monologue about what I did on my maiden, and I what I looked for before I started flying it properly. Some people just go full sticks right after arming, but I prefer a more cautious approach.
Step 1: All the calibrations done - radio, compass, compassmot, flight modes etc. Importantly, only 3 flight modes configured - stabilise, alt hold and Loiter. RTL was temporarily replaced with autotune (more on this later). Laptop connected up for full telemetry.
Step 2: Armed, and spooled up to near hover; to test if it will take off in a stable manner. If it tried to flip to one arm, something is wrong. It passed this test, so onto the next one.
Step 3: Lift off above the ground effect, and wait. Normally, this is when you get an idea of whether the PID's are out. Mine were - P and D were both way off, with the Y6 bouncing/warbling away. I landed immediately.
Step 4: Tweaked the PID's to get a stable flight. Not looking for optimal at this point - just getting it to within "the envelope". Do some basic maneuvers, such as figures of eight, and some throttle squirts. If it responds ok, move onto step 5.
Step 5: Checked the batteries, and fired it up in stablise, and engaged autotune. I was very nervous about this, given the problems I've had in the past. I ran it in stabilise, and it went....ok!
Step 6: Save autotune settings. Check over the motors, props, ESC connectors. Check voltage. Arm...and slowly explore the autotune settings. And for the first time ever...it's worked! It's tight when you jab the sticks, It holds a stable heading, and holds a steady attitude. Checked yaw etc. I switched to alt hold to remove one vector for analysis, allowing me to concentrate on the roll/pitch/yaw. It handles well, but it has a tendency to loose altitude.
Step 7: Start to push it around. More sticks, more speed, more power. Some straight up near-full-power ascents show a tendency for the tail to drag. More on this later. But on the whole its flying well! Refreshed PID settings screen on laptop, and was surprised to find it's really tuned down P and I, and D was quite high. Hmmm. Did a test run of loiter...I was quite worried about this, with a compassmot of 24%. I needn't have worried. It was absolutely rock solid. However, in stabilise, it moves about a surprising amount, even in a 1kph wind.
Step 8: Finish the pack(s), and head home, and start the log inspection. Noticed that roll was pretty tight, but there's a bit of overshoot. And one roll event which didn't match inputs. Pitch...has more issues. I suspect I know the reason - the same reason it was dragging the tail on burst accelerations - a draggy 4 or 6 motor. These two tail motors were consistently spinning faster than the front four as well. More testing to be done here. Maybe a CoG issue.
Step 9: Verify vibes. And it's not a good story. It hovers ok, but apply some real juice, and the vibes shoot up. Way up. Way beyond the +/- 5 I consider acceptable. Again, I suspect this is a dodgy motor...the trick will be trying to find it - and given the tail dragging, I think it's one of the rear motor.
So the next flight will be about tweaking that last few percent, and checking out that possibly dodgy motor! But all in all, the maiden went ok ;-)
Comments
Just want to say thanks for this post Euan, it was instrumental in getting me off the ground!
tuning P using the logs seems straightforward.... too low P value and the values will undershoot. too high, and they will overshoot or resonate around the desired value.
how about derivative and integral tuning using the accelerometer logs? I still haven't seen any easy to follow tutorial on how to tune these parameters. Does anyone have a solid, systematic approach for tuning these?
Before building a high dollar quad I went to Radio shack and got a "Heli-quad" 6 axis toy and crashed the heck out of it just learning to fly. When you couple the high cost and flight modes these machines can do today, the 60 bucks on the toy has paid for itself. (Get the warranty too, I had to return it once)
To follow Scott's comments above, this kind of blog post will help newbies get a better idea of what they have ahead of them, the many potential points of failure in their system, and offers the wisdom of a cautious, incremental approach before trying to do everything at once on the first flight. So I added a link to it on this page, and this page. Thanks for the post.
Euan,
Agree with all. Having just built my first quad with virtually no RC experience, this type of approach is the best way to go about it. Also, read, read, read all the post, blogs, forums, articles, watch the videos, etc. Everything about anything as far as the build and tweaking goes. Someone eventually will have somewhere near the same build and the same issue that will help out. I personally have had a few instances that were avoided (especially the potential of a catastrophic crash) by reading and knowing what the machine is doing and why. There are so many variables with respect to setup that many problems are in fact unique from a crossed wire to software tweaking. I personally go through a flight check list to ensure everything is tight, wires not pinched, connectors tight, etc before every flight.
btw, this is my setup.
DJI 450 kit with 2212 motors, 30 amp ecs's, 4s 5000 Mah 35C battery, pixhawk FC with compass and GPS module and Spektrum DXs7 using the spektrum satellite reciever to the FC.
Thanks again for the post.
Hi Euan,
I would say the batteries at the top isn't the best idea. I tried that with one of my Y6s and found that any tilt whilst trying to take off would just be exacerbated and would cause prop strike. It was just way too top heavy.
What's your CoG like? With 2 x 4500 4S, this feels pretty top heavy...is yours the same?
Nice procedural approach.
Like you experienced, I found on my quad that auto tune lowers P too much and increases D too much ( too much versus a manual tuning that feels right). Is there a pattern or is it pure coincidence?