I've been working on testing and making a few changes to AC3.0 lately, and did a whole bunch of test flights today which went very well. AC3.0 is simply amazing, and there's more good things to come, including a real acro mode that myself, Leonard and few other people have been working on. Hopefully it should be released in 3.1, along with the changes I'm making to the TradHeli code. Most of what I've done to 3.0 are relatively minor, the official 3.0 does work fine on TradHelis, but my changes make it more stable and precise, and perfect the take-off and landing.
I shot a video to show of the performance of the system. I was hoping for a very windy day, but the weather hasn't been coorperating lately. Not only is the wind not very high, but the space weather is bad and I experienced a lot of GPS drifting today.
Anyway, here you go. It's a bit long, but I wanted to be thorough.
Comments
All my changes are available here:
https://github.com/R-Lefebvre/ardupilot/commits/Heli_Development
You could download and compile it if you wanted.
No 500, so that won't work.
I bet I could reassemble my old HK600GT. I have *tons* of parts for it. I should really look into this. The only thing I'm missing is servos I think, maybe blades too. I have a motor and ESC. I wonder what are the cheapest servos I can get from HK? Actually, I even have a couple tail servos, I really just need swash servos. I have a bunch of old S148's.... :)
Oliver, yes, I always prefer to learn from other's mistakes whenever possible, but of course the problem is that commercial gyro software is completely closed source. So we have to hack it on our own. I do think that the performance is acceptable, you'll have to wait until I shoot new video with the big heli.
Yesterday, I did figure out what was bothering me about the external gyro implementation, and fixed it. I think it has a higher chance of working now if somebody really wants to use it.
Ray: I'm not sure what makes you say that all the videos are with a flybar? This latest video is definitely flybarless. I haven't used a flybar since last September. I actually would love to have a flybarred heli setup so I can continue to keep that working. I've thought about just assembling a flybared head that I can swap on but, I don't know if that's the best approach. I probably almost have enough spare parts to assemble a flybarred heli, but mostly just don't have the time.
All the videos I've seen are Fly Bar. Is there support for FBL with APC?
Well, I guess that confirms that nobody is using external gyro anymore! I just found a bug, as soon as you enable it, the swash servos would stop moving... (not as bad as it sounds, you've never have passed pre-flight like this).
Like I say, it's not something I use, so it hasn't been tested in a long time.
I can't comment on the Helicommand, as I haven't seen how it operates. The APM Stabilize controller is sufficiently complex that problems could arise. Maybe it would be OK, I just don't know. It hasn't been tested since 2.7, that I know of. 2.8 brought in the Earth-Frame to Body-Frame rotation matrix, and I just don't know what it will do. Maybe it will actually be better, since now it's outputting true body-frame commands.
To me, an external gyro is an anachronism. I'd rather spend the time perfecting the internal system.
It looks like the SPI bus failed. I did a couple test hops to work on the COLYAW, and on one of them it just went completely out of control in oscillation and crashed. The data logs show that the sensors on the SPI bus reported crap values suddenly. (ie: MPU6000 showing 800 deg/sec rates, and the baro jumped to -16,000)
This board has been through a lot, it's been crashed heavily a number of times. And there were warning signs, over the past few days it would sometimes not boot up properly.
The 450 tail is composed of a collection of parts. The Tarot 1-piece tail box which is a nice piece, and a metal rocker-arm. Regardless, there's about 1 mm play in the pitch slider, the majority of it coming from the fit between the pitch slider ball and the rocker arm socket, and the free-rock in the bearings of the rocker arm. I don't have a lot of experience to know how much play in that system is "normal", but I know that any play is too much for a simple PID system to respond to properly, which is why I blindly state that it's crap. And I know that generally speaking, the singly-supported rocker arm is a horrible design for any mechanical linkage.
On my 600, the tail works much much better. But I can clearly see the difference between the stock singly supported rocker-arm setup, and the dual-support that I switched to. I was able to almost double the gains. In case you haven't seen this, here's a video from last year showing the 600 hovering in tail-in to 40 km/h winds. There's onboard video starting at 2:55 from a hard-mounted camera with no gimbal. It looks pretty good to me, and that is with the singly-supported rocker arm tail, and I also improved the rate controller since that was shot.
But I think most of the problem in the 450 video was due to the collective pitch compensation going in the wrong direction. So the Alt_Hold is busy on the collective, and the COLYAW was sawing away in the wrong direction on the tail.
You may still be able to use an external gyro, but I'm not sure. I've never used one myself, don't even own one. And it's been a long time since anybody tested that. I'm concerned it won't play well with the Body-Frame Stab Mode.
John, yeah, I'm not happy with the yaw either. That is with APM doing the tail control, no external gyro. Part of the problem is simply mechanics. Mechanics on a 450 scale are never great, and neither is the servo. There's a lot of slop in the system which doesn't help.
Also, after that video was shot, Joly reminded me of the potential that the Collective-Yaw compensation was working in the wrong direction. The parameter needs to be negative instead of positive. So whenever the collective increased, it was yanking thrust instead of adding it. I was just working on that yesterday morning, and saw an immediate improvement. But then I seem to have suffered an inflight failure of the APM and the heli crashed. So it will be down for a while.
I have to figure out if the negative collective-yaw compensation is the same for all helis, indicating the algorithm is backwards, or if it's just depends on servo reversing. I think this is the case. It's funny, because we had discovered this about 6 month ago, but I forgot about it while working on all the 3.0 stuff.
The tail performance was quite acceptable on my larger heli with better mechanics. Particularly with the Align double-link pitch slider which is a much better.
Hi, nice work. Yeah tail wag is also a problem i would be interested in how to solve proper…
i have been fiddling with the PID values for yaw for quite a while now and as far as i could observe they were best with 0 to low I-Value and as high as possible P (0.43 in my case) and according D parameter (0.004).
My tri runs on 4s with 12" props at a distance motor to motor of approx 29"/73cm and a total mass of 1.6kg (including gimbal).
is the i-value not of importance or why can't it be tuned via channel-6 ?
thanks Chris
Looking good. But I did see some slight tail wagging going on, are you running a dedicated HH gyro for the tail or is the APM controlling that also?
-
1
-
2
-
3
of 3 Next