I am steadily advancing with my latest in the open source control systems, The Pixhawk PX4. I have not been blessed in my earlier versions and I wasn't today. Grab a cup of coffee and I'll tell you the tale.
In this latest Pixhawk PX4, the designers have abandoned the PIDs tuning system where the user has a chart (in Mission Planner) and enters new values for P I and D. I did not know that before I bought it.
They have put in its place, an auto-tune system which analyses the flight and makes adjustments. I am guessing that this is seen as a safer option but I very much doubt the quality of that thinking. With a chart system the careful modeler will eventually solve the problems but an impatient modeler wont
I have flown auto-tuners before with great success so I was not really worried by this change. But this one is seriously different! This one does its work very slowly. On line I noted the advice that it might take 8 minutes. Now that is quite a challenge for those flying with 2.2 ah batteries.
The plan is to fly a tuning flight in very light winds in a fairly clear area. I went to my 200m square field with what seemed like nil wind. I had preset the satellites at home and walked the model to the field with a battery maintaining the preset. When I disconnected from the computer I had seen an HDOP of .88 which was based on ten satellites. It looked good. At the field I changed batteries to one which had delivered seven (tested) minutes the night before. It took another two minutes before the controller was happy again with satellite coverage (HDOP then unknown but flyable).
It was time to go and I launched. The plan suggests that you fly to a safe height (in my case 5 metres) and switch to altitude hold. I did that but noticed that at five metres there was a cross wind and the drift was significant. I flew the model directly upwind of my position but suddenly remembered that I had not plugged in my battery checker. So I switched out of alt hold and landed. plugged in the checker, reset the timer (silly mistake) and launched again.
This time everything went super smoothly. The model not only held altitude; it did not move from the switch spot where I applied the autotune command. (It could be that in auto-tune, with a really good HDOP, the contoller adds loiter to its navigation)
The tuner was clearly doing its job. It seems to test roll pitch and yaw (in that order) with each elapsing minute. My transmitter timer was faithfully telling me the minutes and I heard 5. But before 6 my battery checker was sounding the alarm. The rule is LAND. I knew the autotune had not finished but I guessed it must be close to finishing so I delayed by, maybe, twenty seconds.
Twenty seconds too long. It dropped like a stone but not quite flat. A propeller broke and that motor mount broke free (as it is supposed to do in a crash). The timer showed 6 minutes 10 seconds but it was failing to include the first flight time as I had reset the clock. The lesson learned is that you must have more than 7 minutes in the bank and must not reset the clock.
I must now wait for light winds again and repeat the exercise with 4.4ah.
If I thought that the design team was listening: I would say that an 8 minute auto-tune does not fit the market. If that total time span really is essential, then the Controller should memorise its progress in tuning, so that the complete tune can be achieved in more than one flight with multiple batteries, if required. If safety is the reason for auto-tuning that seems like a small and safer concession.
I am sure you will have heard a lot of nonsense about clones in this market. The design is Open Source
. Any manufacturer can make it! Not all of them will apply the same quality controls.
That is why I suggest that we stop calling these devices by their design names. The critical name is that of the manufacturer. In my case it is a BZUAV
and at every stage of installation it has behaved impeccably. I have no doubt at all that; when the Pids are rationalised it will work well. If it fails to meet my expectations you will hear about it.
Finally I must say that I enjoy these challenges !!!
If you are getting as much fun from some other device which was fully developed when it arrived in its big expensive box and flew perfectly on minute five, of day one; I am happy for you !!!
I love what I have and, more importantly, I treasure what I have learned in the process.