Yeah, there’s a couple different ones. There’s a ATC_SLEW_YAW param that changes how fast the yaw target updates. And there’s a yaw accel param that limits the maximum acceleration in the yaw axis. I like to leave the yaw fairly aggressive on mine so it tracks the heading nice with the autopilot. And I never try to get any useful imagery or video prior to starting an auto flight anyway. Once the helicopter is in flight, as you’ll see in my auto vid it’s pretty smooth on yaw.
For manual flying I am using a bit of yaw expo in my radio to make it smoother and when I flip to Loiter or Auto the same switch removes the expo curve in the radio.
That’s the way I do it, but that’s not the only way.
I’ve been working towards trying to use an electric for some NVDI survey flights this year when my season starts in June. I don’t think I’m there yet because of the short flight time of the electric, and lack of power when the speed gets starts getting up to 40+ mph. It seems the batteries can only take that for a short time and the whole system just does not deliver the continuous horsepower of a piston engine. So I’m quite interested in seeing some timed tests in Auto with your 800 platform carrying its payload to see if you have hit on a more efficient way to do an electric helicopter.
Chris,
One thing of note is I have quite varying flight times depending upon which packs I use. If I use my beater packs, older Turnigy blues , I notice a substantial drop in the speed of which it comes up to speed on the head and also reduced flight times.
Ive settled on using Pulse 5000-5500mah 45c 6s packs mostly. I get easily 20% more flight time out of those vs a pulse 35c pack of the same capacity and also compared against other brands in similar price ranges. I have 1 set of Thunder power packs and will never ever spend that much on a set of packs again, waste of money. The middle of the road pulse seem to be the seeet spot for performance vs price.
Ive started timing flights again and ill have some data to report coming up. I tried a few loiter flights at lunch again, all were great except for one!
I flipped into loiter quite quickly and the helicopter was still descending when I did that and it had a weird fast ossicilation, it was a ways away, but from what i saw it looked to be on the z axis?
I flipped out of loiter, got it stable again and flipped back in and it was fine again. It should be noted it was windy/gusty when loitering and all but that one instance it was perfect.
Did flipping into loiter when descending cause that?
Hopefully coming up i will have some flight times for auto missions to report. Im going to add a bit of weight to simulate the camera gear, then try increasingly longer flights with different configurations to see which is most effecient.
Planned configs are: 12s 5000mah, 6s 5000mah, 6s 10,000 mah, 7s 5000 and finally 7s 10,000mah.
I will have to crunch the numbers and see if 6s is going to be enough voltage to get to my target headspeed. Im hoping that through 6s or 7s i can get back to running my ESC wide open and increase efficiency as a whole while adding capacity.
Tim
The attitude doesn’t look super tight to me. How windy was this test?
It appears you are still ‘flying’ the heck out it with the inputs as if it’s a typical FBL rate controller. I like to leave the right stick alone for a bit with several nose angles to evaluate the trim. Then go nose out, and perform a couple pitch and roll ~10 degree 'step and hold’s in both directions to see how the stabilize loops are doing.
Jeff,
Are you referring to my build? If so it was roughly 10-18mph gusts here and there, with sustained of 5-7mph. As to flying it, depending on the wind im usually counter correcting in stabilize as if i dident it would drift. I likely have many bad habits from flying 3D helicopters as well?
When ypu say the attitude controller doesent seem super tight, are you referring to desired vs actual pitch, roll and yaw?
Tim,
I agree that the attitude controller doesn’t seem to be holding pitch or roll attitudes that tightly. Is there a reason why you backed off on the pitch and roll P gain. It thought you could get those a little higher before you got oscillations? Certainly you could break out the pitch P gain and get it higher. I know you were not able to get the roll P gain much higher than 0.58 or something like that.
So now that you have some experience with the aircraft flying with the pixhawk, I would recommend trying a non zero value of ILMI (minimum leak value of the integrator). This is what can cause issues when taking off from uneven terrain but it is manageable if you use low values like 0.05 or 0.1. It just allows the integrator to help a little more in a hover to drive the attitude error to zero which I think is some of the problem with why the actual and desired values are not tracking together. I would suggest that if you do use some non zero ILMI that you could back off on your I gain and maybe start at 0.1 then 0.2 and finally 0.3. just to see what it does for you. Keeping the ILMI value low will help keep the aircraft from wanting to lean too far while on the ground.
Just some thoughts. Chris speaks from a lot more experience and what he has found has worked. I’m just commenting on what I know about control systems and the little experience I have with the X3, which non of it is using auto modes. So I guess I’m saying that if you feel this is working and it seems to fly well autonomously then that’s fine.
Bill,
So I dropped I gain down towards .3, added .055 on the ILMI on both pitch and roll and tried. It feels completely different, no weird behaviors, just felt more sluggish and planted in the sky. I landed, then added in some more P gain to pitch and roll and flew. I added enough P gain to where I would previously get oscillations and it did not oscillate?
I’ve landed to charge packs and while charging I looked through the logs. Both pitch and roll seem to follow each other as close as yaw did now when I look at ATT DesROLL and ROLL, and DesPIT and PIT. Was this to be expected? Btw, I took of from uneven ground and it did not attempt to tip over etc. Did not act any different than it did before without the ILMI. I bumped the ILMI up to .07 and bumped th eP gain up a bit more closer to .1, I will try that out and see how it feels/looks.
It should be noted the oscillation happens when I jab the stick hard to the right or left on the roll axis. I can watch it wobble about an inch or so on the mainshaft. I can get the P gains substantially higher before “Really Bad” oscillations start to happen on their own, but the closer I get to that point the worse the small events happen on quick stick movements.
There is a pretty wide gap where they first appear on sharp stick movements and when it just oscillates badly on it own without input.
Tim
Yes now that the integrator is allowed to build error, it can help but that does come at the expense of some phase lag in the response. I think this probably acceptable for stabilize mode but would be offensive in acro.
So with the new settings it seems to fly well. There is an itermittent gusting wind so if there were going to be any issue bacause of that i think id see them?
I just put it up and flipped into loiter for the entire duration of the packs and it just sat there like a good heli waiting for me to tell it to do something. Very little drift if any in all axis. Yaw is on axis good, it seems to respond good to stick commands in loiter as well. I watched it for issues with wind gusts while in loiter and it seemed to take them in stride.
Tim
BTW I’m not sure if you got your question answered earlier about how to set max angle for stabilize. That is set through the parameter ANGLE_MAX. You can test to see how well the aircraft maintains an angle by holding a constant stick input and see that it holds a constant attitude. You can go into the logs and see how much stick you held and what that equated to in attitude. Based in the parameter angle_max if it is set to 4500 then you should get 45 deg attitude at full stick deflection.
Bill,
Okay, that explains what im feeling then. So i just did 2 more flights with ILMI a tad higher, .07 i believe. P gains are a bit higher as well, I gain is down a bit as is VFF.
One flight was stabilize, in a hover the entire flight seeing how it held. Result, holds position and resists wind gusts much better, a tad less snappy on pitch and roll but not bad. If anything it feels smoother like a UAS should be, not a 3D machine.
One flight in loiter, entire pack spent watching it ls behavior with wind gusts. Result, held quite well, only the biggest 25+ mph estimated gusts got it to move a few inches, and by few i mean 2 or 3 maybe. Im astonished how well it holds position in the wind. Way better than my tuned octacopter with absolutely no balooning whatsoever. Impressive.
Tim
Bill,
The question very well could have been answered, but there was also a lot of other information thrown around and I likely didn’t catch it, need to back track and re-read earlier posts now that I have some time. It doesn’t feel like there is a lot of bank angle available, I could be wrong, but it feels like when I’m fighting wind I’m holding quite a bit of stick.
I’m sure a lot of this comes down to my experience to date, which is all with traditional FBL units even when I was using this for aerial work before. I just got used to very fine stick movements because of the fast rates, I would add in more expo than I do on my 3D helis as well as reduce the rates, but even with high expo and 70% rates with an FBL it felt way more snappy than this does with the Pixhawk. I’m sure there is going to be an adjustment period… I haven’t picked up a 3D heli in awhile now, I’m sure that will be a fun first flight after logging so much time with the Pixhawk,.
Tim
I looked at the logs a bit and Roll looks much better than Pitch I think? Which makes sense I guess as pitch still feels a bit wonky and bounces a bit, not bad, but could use some more tuning I believe.
Tim
Have you tried raising the P gain on the pitch axis by itself. That might help with that axis. I won’t have a chance to look at your logs until a little later tonight. I don’t know if you have looked at the PID messages in the log file. Now that you have made ILMI non zero, you should see the I term have larger inputs and may even hold a bias. Before when I’ve looked at the I term it was much much smaller than the input by the P term. The data for roll is PIDR and for pitch is PIDP.
