This sounds very similar to one of my experience with low-pass filter set to 4.
When I was tuning following the docs, (FILT at default 20Hz) I got P gain at ~0.080.
When Bill suggested lowering FILT to 4, I got a good oscillation starting since 0.060 P gain. I don’t know what I was thinking at that time, I decided to turn the P gain up pass 0.080 and oscillation really disappeared. And with notch filter added later, I didn’t get any further increase in P gain.
EDIT: Don’t try that! Increasing gain into oscillating 800 is not a good idea…
A person at the field also suggested increasing RPM (I was using 0-75 range at that time) and with full throttle range 0-100, my flight time got cut only slightly, something like 12mins from 14mins, but the heli now feels much better.
I think you can still use lower RPM for in-place-hovering jobs.
Hi guys, I’m not new to this scene, but have been absent for a few years, Thanks to Chris Olson for helping me get back up and running with my T500/Pixhawk on 3.4.6
Reading the thread, I see we’re all struggling to get a tune that works well in the hover and FFF, just wondering if it’s worth trying two tune profiles, when you select Auto, you also select your FFF tune profile via mixing in your RC Tx or incorporating the change into the code to blend in as the mode changes?
Then, you could have it really sorted at each end of the speed scale? a bit like we do with F3C heli flying, a “condition” (Futaba term), for every type of flying… just a thought??
Tim,
I was hoping to look at some of your latest flights. I wouldnt mind seeing the one when you were putting the aircraft through its paces trying to excite instabilities.
Hi Bob! You must’ve sat down to relax there in Australia for lunch and decided to check in here!
Yeah, the “conditions”. Normal, Idleup 1, Idleup 2, etc., that have been a staple in helicopter flying since forever. While ArduPilot is not really designed for that I’ve often thought it would be nice to have a more advanced RSC system to have a couple different governor settings available with the flip of a switch without using RSC_MODE 1 with direct radio control. Especially for combustion engine heli’s with a RevLok governor where the RSC_IDLE function for engine idle speed without the clutch engaged is kind of a pain because you have to have the FC armed to start the engine. Which I suppose is kind of a safety feature, but you have to disable the auto disarm of the FC or the engine quits before it warms up enough to start spooling it up and engage the clutch.
Hmm,
It wasent evident before, its definatly a new thing, so im left wondering how I caused it?
I guess I did have an issue before when quickly flipping into loiter while it was still moving, but that was a pronounced fast ossicilation. This is just a small .5 meter toilet bowl.
If I arrest the motion of the helicopter before flipping into loiter its fine.
Chris,
Hmm, okay. I will stop trying to tune it out then. In all honesty after getting some perspective by flying a couple known good flying 3D models ive realized that the Pixhawk is flying “awesome” at the moment.
Im definatly going to use the parameters you suggested to slow the RC movements down quite a bit.
As far as flight quality though, it up there. Very crisp but not snappy with stick movements. It doesent have that melted butter feel anymore and alot of the weird coupling issues that caused bounce and other issues with sharp elevator or elevator/aileron combinations are gone. The weird bob on pitch is just about nonexistent now as well.
One thing that helped me tune it better was eliminating most of the “expo” in my radio. I was guarding against it being sensative but im finding my new radio system has a sort of hard knee to it after 20% expo. I was running like 30% for the initial flights, which as ive said was more than ive ever used, and now that its at 12% expo i was able to feel what the helicopter was actually doing more and adjusted tuning to where it is now. Lesson learned…
In any case, control authority is far more linear across the board. It was a real pleasure to fly yesterday evening!
And as I said above, getting some perspective by flying a couple other models helped me tune immensely. It had been so long since id flown an un-stabilized helicopter, lets just say it was an interesting 30 or so seconds while my brain re-trained itself and got with the program.
I jabbed the aileron quickly and next thing you know im inverted with that “oh sh#<t” look on my face im sure realizing there is no more bank angle.
The settings you will make are the ATC_ACCEL’s for pitch and roll. The default of 110000 is usually too fast for bigger helicopters. I would suggest trying those at 72000 for a 800 and see how it acts. You were mentioning the quick yaw response you noticed in my one vid on auto-flying. You can play with the ATC_ACCEL_Y_MAX as well to tune that smoother for your camera work.
The other one is RC_FEEL.
I don’t know that I’d use expo in the radio for pitch and roll. Not unless you can flip the expo out when changing flight modes. Right now you might like it in Stabilize, but in Loiter you don’t really want much expo in the radio. Loiter is more of a sluggish, stationary, hovering mode designed to precisely position the helicopter slowly, and it’s not really that good for actually flying it. Your stick response will become really bad in Loiter if you use expo in the radio because the center stick curve will be too “flat”.
Altitude Hold is an excellent mode to have the autopilot manage collective while you fly the helicopter in the horizontal plane. And you should have the same response in Alt Hold that you have in Stabilize.
If you want flybar-type flying with only rate, then you can use Acro. Personally I do not feel that the electronics and software do a good job of replacing a flybar - it’s just not anywhere near as smooth. But if you are used to FBL units in rate mode you will probably like it. I myself hate it the way I have my DFC tuned at present, so I don’t use acro at all. I use it all the time with flybar when I want to switch the FC completely out of the picture.
Oh - I had sent you a photo of the landing gears I use on my bigger heli’s that have enough ground clearance to carry a gimbal-mounted camera, and bolt-on attachments for different payloads from aux fuel tanks to tether drops. If you want more details on how to build those just PM me and I’ll send you the details.
Chris,
I pm’d you, and as stated in the PM, i appreciate the ideas/designs for mapping. That is totally uncharted territory for me and I have been mulling over what to do there for awhile. Also on the fence with camera choices and support gear needed for mapping.
Put simply, the furthest I have gotten with mapping is “I know I need to do it”.
Tim,
I looked through your logs that you sent me. As far as the pitch bounce, its hard to tell whether that is some feedback instability or a natural mode of the aircraft/control system. It appears to die off in 2-3 cycles so it is moderately damped. You are still carrying some bias in both pitch and roll between desired and actual. It occurs more so in pitch than in roll. I really think you need to set the ATC_I_LEAK_RATE to 0.001. This will allow the integrator to build more error and help with that. I’m interested to see your param file for the stuff you did yesterday that cured the pitch bounce.
Tim, on your tail bounce could you try something for me to see if it goes away?
If you can move your batteries, move the CG slightly off center, either forward or back, it don’t matter. Rearward would be preferred. Just to see if the extra mass being slightly off-center from the center of rotation of the airframe damps it. These 3D helicopters are so well balanced to be quick and responsive that you sometimes have to shift things around a little bit to modify their inherently twitchy behavior.
Chris,
Sure thing, im willing to try anything. At the moment, where im at with this project, is data collection. Basically try something, see how it affects things and record data, then move forward.
I managed to remove alot of the bounce, and kind of have it in the back of my mind that the gimbal will likely dampen things considerably as well.
That being said, i was running the helicopter with a lower headspeed than I am using currently albeit with an FBL and I had no bounce on the pitch axis. So it been bugging me a bit.
Tim
Bill,
Ill send you three of the parameter files tonight. One being the first try with the notch, the second mid-way through after adding in I gain and such, and finally the setting I settled on at the end of the day with pitch bounce lessened.
Likely the FBL unit was mounted closer to the center of rotation of the heli in the FBL unit tray. And in that location is better able to respond to changes in angular acceleration. That was actually one of the deciding factors for me to try the Pixhawk Mini in a helicopter. It’s so small it gives me more mounting options to put it in an ideal location closer to the center of rotation of the airframe. And it has a separate “breakout board” to power the servos, etc so wiring can be more conveniently located. I haven’t installed it yet because it can only run 3.5 or newer, and I haven’t gotten back to testing 3.5 any more yet.
Chris,
Id have to measure it to be precise, but the forward location I had the FBL on is about the same distance from the mainshaft the Pixhawk is mounted on the pitch axis. Although that being said, its behind the mainshaft vs in front so that can play a role I suppose?
Ive been reluctant to try a different mounting location as the vibes seem to be good there, I am going to stiffen up the frame behind the pixhawk though, I purchased some carbon plate for that purpose, i noticed some resonant spikes that specific headspeeds seem to provoke and i had done a test where i spooled it up on the bench and the frame resonates quite a bit on spool up and spool down as well as peaking at certain headspeeds. I clamped a peice of plate to the frame with some teflon sheet in between and it reduces that substantially. But through bench testing the most vibration free place on the helicoptet without building some custom mount was where it is.
Also, every time I look at the cube mounted in the carrier board I think how nice it would be to have a custom, much smaller carrier made… I cound mount it just about anywhere then…
I don’t know. I have no clue what the FBL unit uses for angular accel or anything. From what I understand, it’s a one or two bounce thing, and then it straightens out? Not an actual pitch instability that starts and won’t go away? There’s a number of things you can try mechanically without trying to fix it in software, assuming you want to stick with the fast ATC_ACCEL’s with a big heli:
1.) Speed up the head. If it goes away, you know what the problem is
2.) Assuming you want to fly at lower headspeed, try softer dampers on the feathering shaft. If you have to, put the grey flybar ones in it from a 700
3.) Add weight to the helicopter
4.) Move weight in the helicopter
Chris,
Im running the red dampeners currently which are the softest I know of for my head. Compared to the DFC dampeners or even the black or grey ones they are very soft.
They were meant for the Trekker when running the 800 blades for aerial photography etc.
I know they solved alot of my low headspees issues when I first set the helicopter up for video work.
I was just looking at a new head for that helicopter. Its a 3 blade head and uses the newer FL style swash. I would have to change out a fair amount of parts as the servos would be stacked around the mainshaft as are most newer designs.
The only reason, other than the fact that its a 3 blade, is that there are multiple locations to mount the ball studs for both the servo attachments to the swash and then the swash to the grips. It is touted to be for adjusting servo resolution/mechanical rates. I dont know if I would gain anything though vs the push pull system as most systems, which you have stated, are for 3D and very quick rates vs what im looking for. I would have to get some specs and take some measurements with the calipers and do the math i guess. Im just wondering at this point if 3 blades might help my case out significantly?
All that aside, I think drastically changing CG is an immediate option, headspeed just saps power so quickly I want that to be a last ditch option, im very interested to see what the gimbal does to the equation and will likely do a couple tests this weekend as I am dying to find out what that is going to add to the mix.
Im also very close to pushing the buy button on the 3blade head and tail along with a couple matched sets of blades and the nessecary parts to make it happen. Although, im easily over $1000 for that and im having a hard time with it as it flew fine before, with the gimbal and without at much lower headspeed than im running now. As ive said before, I went with wide chord 800mm x 71mm blades vs a pair I had that were only 60mm as through testing I found them far more stable so I never looked back at 3 blades.
Another thing I’ve been wondering about is, what headspeed are you actually running? Have you put a tach on it? Another thing that’s pretty undesirable (from a camera standpoint) is the “blade chatter” or “lift chatter” that you get with a two-blade rotor running low headspeed. It will actually shake entire helicopter in flight making turns with it when the collective goes up. The LHS guys all brag about their “blade slap” running 900 rpm, but they’re not flying cameras on 'em either.
Very crisp but not snappy with stick movements. It doesent have that melted butter feel anymore and alot of the weird coupling issues that caused bounce and other issues with sharp elevator or elevator/aileron combinations are gone. The weird bob on pitch is just about nonexistent now as well.