I doubt it’s static turning the rotor that slow. I would unplug the servos one at a time to eliminate the possibility of having a bad pot in a servo. Also try reverting to the old firmware and see if magically fixes it. If you don’t have any archived older firmware, PM me. I got several builds.
Well, the problem carried over when I upgraded to 3.5 (and swapped Pixhawk unit).
Will try the servo trick. Really don’t want to invest in new servos for this one, as I’m already eyeing larger helis.
Well, try the servo trick. It’s probably best to track down the source of the interference and fix it. If it’s a bad pot in a servo, you can take it apart and clean the pot’s wiper with WD-40. Done it many times on different servos. When the pot’s wiper gets dirty they usually start to groan a lot and even chatter sometimes. You’ll find the potentiometer right under the servo’s output shaft, usually engaged with a flat tang. It sends the servo’s controller the position information of the servo output shaft.
A scratchy or dirty pot can cause that issue. The input to the servo’s controller wants one position, the pot is telling the controller it doesn’t really know where the shaft is for sure, so it draws a lot of power keeping the servo’s motor “hunting” back and forth and that’s what makes it groan or chatter - and produce an excessively large magnetic field around it.
You’re right. It was the tail servo (Futaba BLS257). The reason I didn’t notice it on my first configuration, was that the GPS was a lot further down the boom. After I moved the Pixhawk, I had to move it further up and then in my final attempt, right next to the servo . Of course, this would only show up when the heli was armed and the reason it reacted when I rotated the blades, was because it was in stabilized mode and the servos moved…
Made a new shielded cable tonight appears to have fixed it?. Needs more testing, but will be out the next 2 weekends…
I don’t know if it did. Any cable with a shield has to have the shield connected to the reference ground for the system and the shield is more to protect the center conductors from common mode interference than to prevent RFI or EMI radiation from the cable. I didn’t look up the BLS257 to see if it’s a brushless servo but it probably is. Which means you have a little inverter on the circuit board in the servo that produces the square sine AC power to drive the brushless motor in it. And you have a potentiometer that sends the servo controller the position of the servo output shaft.
Excessive “noise” can come from either the pot or the inverter if it has a failed cap on the driver. There is some magnetic field around those servos, created by the mag field around stator in the brushless motor. But it is very small and weak and should not really affect the compass much even if the compass is right by it - at least nothing that a compass cal with the servos operating can’t adjust for.
It is likely that servo has a problem, especially if it seems “jumpy” or the tail is excessively “twitchy”, I’m guessing a dirty pot.
Well, I was getting zero variation in the EKF Compass Variation value, which would always move a little before (servo on or off). It was late when I finished, so only did a brief test, but will need to re-calibrate and test some more.
Meanwhile I’ll start looking for a new tail servo. Any recommendation for a 500 size servo that doesn’t break the bank?
I’ve actually had pretty good luck with the Hitec servos for less money than Fubata’s, B-K’s, etc… I’ve flown them on both helicopters and TriCopters without issues.
So, got out again, old servo, new cable/GPS position and had seemingly no compass issues. A couple of very short spikes., but on wandering compass headings like before.
However: It really struggled to maintain the set altitude. This is not an issue I’ve had before. I noticed that you cover your pixhawks in a foam material?
I raised the altitude, ran a figure-8 mission and it was glorious. Tracking perfectly, (auto)-landed as straight as you’d want.
Decided it was stable enough that I could have a camera in one hand.
It was clearly not reaching the altitude it was supposed to and then it got confused, stopped and I dropped the camera and first switched it to loiter, sorta stabilized and then switched it to stabilize.
Logs show an EKF error and a high velocity spike. Vibes seem ok.
EKF compass variations looked good, so I launched it again and ran the mission again. Apart from then again being at the wrong altitude, it flew it nicely, although wind buffed it around again during the landing.
(I’ll add what video I got later)
Yes, vibrations look fine. When the EKF blew up on you it looks like it thought the heli was doing 50 mph and it got tossed around a little bit at 30 degrees of pitch and 20 degrees of roll (or so the EKF thought). It “switched lanes” on the EKF a couple times. reset the yaw a couple times - it basically got totally confused. At least you had some air under the heli. I had it happen this morning with a 700 and crashed that one. Didn’t have enough air under it and it happened too close to the ground.
I still haven’t figured out what caused my crash. But I do know this EKF “lane switching” is bad design for helicopters.
Yes, I put some foam covering over the Pixhawk. It seems to help damp gear train noise vibration, and protects the Pixhawk from flying grass and dirt when the EKF blows up and causes it to dig the rotor tips into the earth
I think your compass issues might be related to the firmware too. I’ve already had that problem as well.
Handheld Runcam2, so not the greatest camera for filming this kind. The blow-out looked way worse in real life!
Yeah, every one I’ve had in flight has been recoverable. It’s only close to the ground where it loses attitude solution that it’s been really bad. I’ve not seen it (yet) where the EKF blows and doesn’t sort of recover after a few seconds. But it can definitely make your flight interesting. I wish ArduCopter had a different attitude controller design where it maybe uses the EKF to keep the gyro drift calibrated, but doesn’t actually depend on it for the attitude solution. And if it goes haywire revert to the gyros and fly using good old dead reckoning until it straightens itself out.
I have been flying the Trex 450 with a new tail rotor. The tail appears quite good but the servo is moving back and forward really fast. I am using the suggested setting you made at the start of this post. I have tried adjusting the P and D down but it makes no difference.
David, first try setting the FILT param down to 10 or even 6 and see if that stops the tail servo ocillation. If it doesn’t, set the P, I and D values to zero, set the VFF to .15 and see if that stops it. If it’s still doing it then there is a binding problem in the linkage, a problem with the servo itself, or interference to the servo (like the MavLink radio). If it stops oscillation with the VFF set to .15 and everything else zero, then it’s a problem with the PID loop being too aggressive.
Chris did a few tests indoor this evening and with FILT param down to 6 the oscillation almost disappears. When you rotate the heli the yaw will give 2 or 3 small oscillations but after that its rock solid you can see the servo working but my feeling it quite normal. I have tried reducing the D gain from 0.007 down but at some point the tails starts to whip. Wondering if i should change anything else.
I’m sorry David, I didn’t catch this one when it came thru.
If the tail starts to whip back and forth radically, it’s a sign of the damping being to “soft”, i.e. it needs more P or D gain. If it oscillates rapidly that’s a sign that it’s over-damped (too much P or D gain.
Some heli’s and servos seem to like a little VFF on the tail. They’re not all the same, so it’s hard to make a recommendation. But try bumping the VFF maybe .02 or .03 more from what it is and soften the P or D by a notch and see if it gets rid of those 2 or 3 small oscillations.
A lot of tail servos are .06 second full sweep and they are so fast they don’t play nice with the PID loop. That’s where using a small amount of VFF might help.
No problem wednesday is our indoor flying night and it a good opportunity to sort the tail problems as the weather is rubbish
The Tail is sorted with the settings below
ATC_RAT_YAW_D , 0.006
ATC_RAT_YAW_FILT , 6
ATC_RAT_YAW_I , 0.27
ATC_RAT_YAW_ILMI , 0.0
ATC_RAT_YAW_IMAX , 0.3
ATC_RAT_YAW_P , 0.25
ATC_RAT_YAW_VFF , 0.03
Not super sharp but what I would call nice.
What i did notice is a small vibration which i will investigate.
I will be very interesting to get it outside in some wind.
Might be time for the flybarless head.
Thanks for your help
That looks like perfectly reasonable tail settings for a smaller size heli. The tail response depends on a lot of things beside servo speed too. Headspeed (and associated tail rotor speed) can affect it. And 3D pilots are always trying different blades to find the “perfect” combination. And on some you can change the gears. On my Synergy 626, for instance, I reduced the tail rotor speed by swapping out the 4.5:1 gears for 4:1 in the tail transmission. Then replaced the stock 96 blades with 106’s. That simple change reduced the required power in hover by ~45 watts and actually improved the tail response.
So. Long story short: Looks like the real culprit of all my compass trouble was actually the pixhawk unit I was using.
It would not recognize any of my new GPS’ (A Here GPS and a mRo unit), so mounted a different pixhawk and after a lot of faffing. (The Here GPS has some quirks too), I finally got it working pretty much flawlessly today.
GPS was solid. Compass was solid. Tune is decent. Gimbal and FPV Pan & Tilt is working. (And more or less synced).
Needs some wiring clean-up, but I have yet to decide which connectors to use for payloads.
Need to get it out to larger area so I can stretch it’s legs…even 5 m/s looked too fast for that area!.
This post is very helpful. However we have a step by step mechanism to tune the roll and pitch which is well documented here. I think it would be great if we can come up with a similar mechanism for yaw as well. The guide lines mentioned here are fine but every heli is different and there a variety of factors due to which we cannot use prefixed values for yaw tune. I will do some experiment and see if we can have yaw time mechanism which would work on a variety of platforms.
I would recommend using the same procedure that is currently used to tune pitch and roll and see if it is any better or worse. You may want to keep a little P gain in there during the initial steps to set VFF. If you have some P gain then you are looking at the point when the P contribution to the PID reverses to reduce the amount of input. That then tells you that you have enough VFF. Then tune the P and D gains as is discussed for pitch and roll. I have tried it but I’m not as critical of my tail as most so I would like to see what others think.