This should be interesting

Hello,
There is a ton of imformation here for sure, both Bill and Chris as well as others have offered up a veritable smorgasbord of knowladge! I would likely be staring at a pile of parts on a bench if not for this thread. That looks like a fun project you have there, im hindsight i wish i would have started with a smaller heli like a 450 or a 500 but unfortunately i just dont own any small stuff anymore.
As to the beta firmware and the new flight controller, i can say that so far its behaved quite well. Even in the face of my complete lack of knowladge regarding Pixhawk, it has managed to give me a decent flying model at this point, and im no-where near done.
I think its great someone else is going to use Pixhawk 2.1 on a heli. Everything has been “about the drones” lately and its refreshing to see more helicopter people! :slight_smile:
Tim

That’s just one of those Been There Done It things. Align helicopters do not fly very well much below about 450 feet/sec blade tip speed. At 400-425 you’re right on the limit of an unstable helicopter.

There’s a neat Android phone app called Headspeed Tachometer that works and it only costs $4.49. Just lay your phone on the ground by the helicopter and spool it up. It’s accurate to within about 25 rpm.

Correct. If you have it tuned so it feels pretty decent in Stabilize, then Try Alt Hold first. It won’t hold position in Alt Hold, but it should hold altitude ok just using the barometer. Since Loiter depends on that, test that first.

Then try it in Loiter. However I would not recommend trying it around all those powerlines and buildings and whatnot I see in your video. It’s possible you have a good GPS signal there. But that is the kind of stuff that has gotten the DJI drone people in trouble with authorities and whatnot - flying them in town where there’s loads of RFI and powerlines, etc… They get the “GPS Safe to Fly” on their phone, they don’t know how to fly a helium filled birthday balloon without wrecking it, get a GPS glitch and got a flyaway. Then just stand there in shocked disbelief that their new $1,000 toy suddenly became self-aware and went Terminator on 'em.

When you switch to Loiter just watch the heli. If it starts to rock and roll gently and seems to sort of “circle the toilet bowl” then you need to do some more tuning. Don’t let it go if it does that. Just snap it back into Stabilize and fix the issue before it darts off in some direction you don’t want. It will not be scary. Just revert back to Stabilize if it doesn’t act like it should.

Chris,
Yeah, ive heard of that headspeed tach, I will likely bite the bullet and buy it. The only telemetry sensor I wanted on this heli is the FLVSS cell voltage sensor so I can see the lowest individual cell voltage. I’ve gotten into the habit of using that to judge when to land as sometimes a pack will report a good total voltage but one of the cells is a bit weaker and gets over-discharged and viola’ puffed pack. Which is a bummer if its a $140 lipo.
I appreciate the guidance on trying loiter for the first time. I am most certainly not going to do it at the house, although my neighbors are fine with, and have gotten used to me flying big helis around, I agree totally with the “too much interference” sentiment.
It’s funny, when you mention “authorities”, I chuckled as when all the drone mania and FAA stuff hit, I was flying one time next to the road and a cop pulled up. I thought for sure I was going to get the 3rd degree, but I was wrong. It was a younger cop, he pulled over and rolled his window down and asked if he could watch. Later I talked to him and he was very interested because he saw it was a helicopter and for the next couple months he would “patrol” by the house and watch me fly at about the same time every day. I was tuning a 3D heli and flew it religiously at 6:00 every day after work if it was sunny. But for this, when it comes to GPS modes, I have a park 500ft from the house the City has given me permission to use for testing and such. Not as huge field, but big enough for medium speed circuits with a big model and most importantly, its close by.

So I am still having that drift back and to the right. I have to hold constant correction when flying. :confused: I checked the swash again and it looks fine, and level in MP corresponds with actual level on the bench. One thing I did notice is that in the parameters, RC3 has a trim of 982, which is also the MIN. ? Don’t know what that’s about?
Tim

It must’ve gotten that from your radio cal somehow? And yes, it’s normal to have the channel 3 (collective) trim at min. Might want to have a look at your rc output on the radio cal screen and see what actual min is, and just set it to that. That’s kind of important for failsafe later on if you start flying missions with it and lose radio signal. It uses the collective value for triggering the failsafe.

I’m actually very impressed with your Pixhawk2.1 project. There is some stuff in the 3.5 code that needs work yet, but there is people working on it. For now, just don’t try taking off in Acro until you get really used to the helicopter. And the rest of the stuff that needs attention in the code is not going to be a glaring issue I don’t think.

Myself, I never 100% trust the autopilot to fly the helicopter in tight quarters. I always prefer to take off and land manually, even for auto missions. And got my flight mode switches set up so one flick of my thumb and I’m in Acro where I got 100% control of the aircraft (with a flybar). For you that would be Stabilize.

One thing to consider for more of a low speed hovering platform for camera work is switching to a 3 or 5 blade scale rotor. They are much smoother and can operate at lower headspeed than a two-blade can. The lift pulses from the blades are smoother, and you could likely fly down around 900 rpm with a 800. But you’d have to change your gearing for that. You will get longer flight time with a two-blade rotor. But it’s not that significant when you consider that you can run the main rotor much slower with say a three-blade. Three blade also has better dynamic balance than two-blade. There is a reason most of the wind turbines on the planet use three-blade rotors. Even though two-blade is more efficient, they tend to self-destruct on wind turbines due to the dynamic balance issues, and they will go into a severe headshake when they yaw. A helicopter rotor is really no different, other than the fact that it operates at much higher blade tip speed, so the vibration issues are mitigated some by the speed.

Just thought I’d mention it. I know at least of the folks that flies ArduPilot helicopter uses a three-blade rotor for a camera platform and it works quite well for him. And Bill, I think, uses a four blade on his X3.

Chris,
Thanks for the explanation. As I am still new to this, seemingly simple things makes me worry.
As to a 3 blade head, I agree that would be ideal, but the blades I’ve chosen to run are hard to get matched sets for 3-blade etc. I run 800mmx73mm blades, which are quite wide compared to a lot of other blades on the market, at least the ones I’ve tried. Rail, Cyclone, the Align 708’s, Switch, Zeal, SAB, and Rotortech.and i’m probably forgetting some others ive ran? I have a few sets, and have yet to find 3 that are remotely close to eachother with spanwise and chordwise CG and weight. They are always well balanced within the pair, but not between different pairs. I suppose I could attempt to balance them with some tape, but I just threw in the towel a couple years ago and was getting good flight performance with the setup I had. I know at the same headspeeds I run now, the Align 780’s or the Zeal 800’s were much less stable than the blades I have on currently. Although, the other blades I think are designed for quick flip and roll rates and the ones I have were Aligns attempt at Aerial photography with the Trekker platform, which they have now dropped in favor of the 3D screamers…
Tim

Chris,
I played with the IM_STAB_COL parameters. 150 500 725 1000 seems to give a profile where at midstick I have a few degrees of pitch while eliminating some negative and flattening the curve. Are these likely acceptable numbers? I flipped to loiter on the desk and the swash goes to about 25% then slowl goes the rest of the way to full negative pitch. That concerns me a bit?
Tim

or is something like 125 550 600 900 a better bet? I get an “out of range” error on the 550, but it moves center stick to a few degrees of collective whilst flattening the curve a bit more. I just don’t know how my actions here are going to affect other modes?
Tim

That will be fine, Tim. Ignore the out of range message. I don’t know who came up with those ranges for helicopters, but the ranges they think are “legal” are not practical in the real world on any helicopter I’ve set up yet.

Don’t worry about the Loiter thing on the bench. Loiter don’t even work unless the FC is armed. And it’s not a good mode to take off in because it’s sluggish handling and not much collective response due to the 40-60 “deadzone” on the collective. Same thing with Alt Hold. They’re flight modes to be used in flight. Not for takeoff and landing with a heli. You CAN, but I don’t recommend it.

Chris,
Thanks, that makes me feel a bit bitter. So with loiter, are there any specific tuning parameters I should worry about, or is it a case where once stablilize is tuned loiter should just work provided the EKF is okay and you have good GPS etc?
Tim

The default settings for Loiter should “just work” to make it hold position. Loiter uses the other things in the code like the attitude controller and Alt Hold features all combined with GPS for horizontal position. There is a param for setting the failsafe for if you have an EKF fault, and that will either land it or change it to Alt Hold. I prefer the Alt Hold feature instead of just landing because if the heli is a long ways out and don’t have good visual reference to see what it’s going to hit if it lands, with Alt Hold (or Acro if I need to use it), can fly the helicopter with FPV until the EKF returns, then continue the auto flight. And I"ve had to do that before. That’s the one downside with FBL - if you lose EKF the Pixhawk can totally lose its attitude solution and have no backup. With Acro with a flybar, the helicopter is still flyable and can bring it home on FPV and once its in good visual reference range, land it safely with the EKF totally blown.

Chris,
Okay, so basically if Stabilize is good and flies nicely, and i have good GPS signal without EKF errors im good to go? Ive watched my logs thoughout the tuning process and have yet to see an EKF failsafe or issue. Although i do not want to jimx myself… This can be a superstitious hobby at times… lol
Its weird, i just had it up for a flight and the drift was all but nonexistent? I havent changed any parameters to correct it. Its almost like the drift depends on how level the helicopter was when it took off? I made a landing pad for future tests and have been using that today, and its much more level than my previous places i was taking off from. Although that could just be placebo? I do know for certain the last few flights ive been fighting almost no drift?
Tim

Nope. It’s the I-gain as you’ve turned it up closing the attitude error. Has nothing to do with taking off on a level surface. I’m sure you’ll find out, as most of us have, that you’ll end up in the mid to high 0.3’s on pitch and roll to really get it nice. And you may have your FILT param just a bit low. I use that low on the yaw, but with the P gain you’re using now you’ll probably end up turning that up to the 10’ish range.

With the VFF where you got it now, you should have pretty nice servo travel in Stabilize, unarmed on the bench, and the Loiter controller seems to like that too. If you got the VFF too low it will limit servo travel and be really “sloppy” in Loiter.

Chris,
Ive thought about adding just a tad more I gain, unfortunatly, especially with the higher headspeed, P gain is about maxed. Maybe when inadd the gimbal i can add in some more P gain? With the 6’ dowels on there my P gain way much higher than it is now. I have it hovering at roughly center stick now and response feels better around center. I charged up 6 packs, hoping the last set may be a loiter? :confused: i want to make sure its not going to jump up and down or dart off suddenly first though?
Tim

Oh and i forgot to ask, is there a parameter that equates to rates in Pixhawk, like the travel rates in your radio? Can i just turn them down in my radio instead if its too snappy? Im still suprised at how snappy the helucopter geels, and i have a pretty high 30% expo in there already to dull it down for aerial work?
Tim

It’s perfectly fine to run some expo in your radio for Stabilize. Just set it up on a switch for a Condition 1 and Normal Mode or something like you’d use for a regular helicopter to switch it off for Loiter. Maybe incorporate that into your flight mode switch so when you go to Loiter it turns off the expo.

As far as rates, that’s what you’re tuning with the rate PID’s. And the snappier it is, while still being stable, the better the Loiter controller will like it.

As far as tuning for your load, that’s totally up to you. Personally I do not do it. Nothing I hate more than have a twitchy helicopter when it’s carrying a load. The only thing I adjust is the headspeed. More load, more headspeed. I have my center stick set at 5.7 degrees of collective pitch. If it won’t hover at center stick I turn up the headspeed until it does.

I would caution you on messing with the FILT parameter if you have tuned the P gain with the FILT param at a value that attenuates the instability frequencies. Turning it up will cause the aircraft to have bad oscillations.

Bill,
Whats become of the notch filter you were working on? I saw Leonard and yourself hashing it over. Is it going to make a release? It sounds as though it would help quite a bit to be able to attenuate the specific frequency of instability without hammering the rest down with a broad filter.
Tim

Chris,
It hovers at mid stick, is that all i need to get it not to jump when switching to alt hold and loiter?
Tim

Yes. Alt Hold and Loiter have control of the altitude and they sort of “reset” whatever collective curve you are using in Stabilize and have a “deadzone” of about 40-60% stick. So if you want to climb or descend in the altititude controlled modes it takes a lot more stick movement to do it. When you switch modes, in and out of the altitude controlled modes you either have to be ready to catch it right away, or have your collective curve set so it sort of corresponds and don’t get a big collective jump.

Oic, so it’s not so much switching into Althold or loiter, its when you switch out being ready to catch the drop in collective if the curves do not correspond?