Unless there is something different about the Here GPS, with a regular M8N like I get from the Chinese vendors, if the arrow is pointing ahead and the unit is not upside down then I do nothing for the compass orientation.
Regarding the internal compasses, if these two params are set to zero it should only use the external:
COMPASS_USE2 , 0
COMPASS_USE3 , 0
The FBL controllers are a lot of hype and advertising, Tim, but they are not nearly as gloriously successful as you describe. We had a Radar Run at our R/C field last summer. Flying the 700 manually with the governor on the 105HZ turned up to 100% with a tuned nitro pipe on it - about 4hp @ 16,000rpm - I hit 85 mph on a 1/4 mile run. Two of the FBL helicopters that attempted the radar run lost control and crashed when they went into uncontrollable pitch up and roll, and subsequent oscillation when they went into RBS. The flybar, being a ādumbā mechanical device like a flyball governor automatically compensates for that because it doesnāt like to rotate in a different plane than the rotor. So it brings the stalled side of the disc back under critical just like nothing happened as soon as the heli noses up a bit and starts to roll in RBS. The FBL controllers wonk out and lose it. The flybar does the same thing in the wind - it automatically compensates for it and the autopilot really doesnāt have to work that hard. I know FBL is all the big fad. But Iāve never been impressed too much by fads. How much you wanna bet that the successful unmanned commercial heliās that have gotten FAA cert in the US (Yamaha Rmax and Fazer) use a flybar head for a reason? The engineers that designed those unmanned heliās were not into fads. They wanted reliability that works every time.
But thatās all just my opinion. Because Iām probably the only one that flies flybar heliās with ArduPilot.
Chris,
I do understand your position on the market, they are getting way more money for the units than they are worth and āsomeā are not good for speed. That being said, the fastest helicopters out there are in fact specially designed fbl helis. The TDR2, the Goblin Comet and thats about the listā¦ unless you consider the TDR1 still relavent. These are specially designed small production run kits with lower head geometry and actual engineering behind them.
As to FBL controllers, āmostā of the units out there are awful at high speed flight, there are a couple though, the Bavarian Demon is well known in the speed circuit as well as the Spirit Pro. Ironically both of those companies liscence the same control algorithms for the āun namedā company. They seem to guard that information well. I can say for certain if you take your heli out and speed fly it to the limit with an Ikon, Ikon2 or an MSH Brain, or a Beast X along with many others you are going to have a bad day.
I completely agree that most are selling hype, but there are a couple out there offering a rock solid unit that just plain flys well out of the box with almost any setup and handles high speed flight all the way to the record books.
Im not an fbl salesman by any account, im just giving my perspective have had many of these units, and from being at a speed flying event where the men get seperated from the boys, and the guy flying the Ikon hits RBS at sone ridiculous speed and watches his heli desigentrate midair into a shower of parts.
Its those couple units out there I speak of, they have some solid control theory behind them, awesone vibration resistance and in the case of Spirit, soon GPS and waypoint navigation.
You dont have to sell me on the redundancy of a flybar, i worry all too much what will happen with my UAS should the Pixhawk crap out midflight. Im going to give it an honest go with ājust the Pixhawkā, but if i have to i will add an fbl into the mix as a plan-b. Im also likely going to have a trex 700 flybar heli to play with very soon as well which will also get a pixhawk 2.1. My end game is a commercially viable platform so it needs to be reliable, but also i have to be able to get parts and new airframes to work with unless Iām going to go the 100% custom route, and that would cost a fortuneā¦ So thats off the table.
And thanks for the info regarding compassās Chris, that confirms what i thought and makes me rest a little easier.
Despite the fact that the arguments are endless about everybodyās favorite FBL unit, this is what Iām saying about them being over-rated. Iāve seen plenty of them flip helicopters right over on the ground too.
You see, the Pixhawk and ArduPIlot is fully capable of extreme high speed flight in full autopilot, right up to the 100mph speed limit in the US for unmanned commercial aircraft. If you know how to set the params for it. Iām not going to post what I use to do that because I donāt want anybody trying it and crashing their heli because they donāt have it set up right.
When EKF2 goes south, itās usually an issue with the IMUās not agreeing. Or multiple compasses not agreeing. Which is usually related to an issue with the mount for the flight controller - either letting it āfloat aroundā in the frame too much under high G-loads, or vibration transferred to it that causes āaliasingā of the IMUās on one or more axis. Actually getting there may sometimes take lots of tuning and testing and tweaking. But once you do, the old Pixhawk 1 kicks the butt of any FBL controller out there.
The FBL units are marketed and sold to the 3D crowd. The Pixhawk/ArduPilot system is designed for UAV type flight. They serve two fundamentally different purposes. In theory, the Pixhawk is capable of flying a full-size manned helicopter, fully autonomous, from Chicago to St Louis if you set it up with the right servos - none of the FBL units are. While Pixhawk/ArduPilot may not have the ease of setup of some of the FBL units, it is considerably more advanced and configurable. So I find it really hard to compare the two because one is a toy, the other is a real autopilot.
Back to business at hand,
So iām working on wiring harnessās and power delivery along with routing at the moment, but now that I seem to have a reliable booting board without errors constantly Iām starting to think about some good starting values for the PIDās.
I understand every helicopters different, but there must be a value that I should strive to get to? Ive read many, many posts on the issue and have seen quite varied opinions. Now im looking for another one, as it seems ive found some veteran users of the Pixhawk.
Thanks,
Tim
Chris,
I think you are misunderstanding my opinion of FBL units. I totally agree they are not even in the same realm as the Pixhawk, thats why i have purchased some and am migrating to them for my ānon hobbyā applications.
All I was getting at is a couple companies out there have developed some seemingly bullettproof tech based on accellerometers and gyros that doesent get saturated and lose orientation no matter how hard you push it. Its that one thing, the code they use, that im saying would be a good addition to Pixhawk. Somehow they have solved filtering the vibration charteristics of RC helis and that i believe would be valuable.
Definalty not trying to categorize Pixhawk in the same group as Ikon, or Spirit, or BD etc etc. They are two very different applications. Im not building a UAS carrying thousands of dollars of gimbal and cameras to have it held together with a $200 Hobby grade overpriced fbl. No redundant power systems there, no polarity protection and electronic filtering etc etc.
Tim
Oh, come on. You have way too high of an opinion of those units, Tim. The vibration affects them just as much as the Pixhawk. Every pilot I know of that flies 'em is always trying out new gel pads from HobbyKing and yadda yadda to get their heli to handle better. Donāt care what brand it is. The main advantage they have is their small size so they can be mounted on top of the tail drive case, or on a special gyro tray, which is the lowest vibration area of the helicopter. Stick one on the side of the heli on the carbon fiber frame sides that act like a big drum, and they got all the same problems as a Pixhawk.
With the Pixhawk on a helicopter, itās all about the mount. And itās considerably bigger than a FBL unit so itās harder to come up with a vibration-free mounting system.
Chris,
Not to keep this going lol, but when get a chance iāll snap a pic of where and how I have one of these Spirit Proās mountedā¦ Needless to say itās just one layer of Home Depot VHB tape held down āHardā with a Velcro strap. No gel pads, no trying to find the right place to mount it, I just slapped it somewhere, strapped it down and went on with life. Having been through many, many of these units from all sorts of different brands, itās easy to pick up when someone has some actual good code in there. Lets face it, they all use the same gyroās and accels, its not like they actually manufacture them in house.
Iām no fanboy of anything, I just unloaded a lot of Spektrum gear recently, not because I wanted the next new thing, I went FrSky because it offered more features for way less money and infinite flexibility. At my field iām the only one with Spirit, the only one with FrSky and the only one with Hobbywing etc etc. I fell for the flashy ads and the everyone else is flying it crap at first, now I just get what I āthinkā is better or suits my needs, no more no less regardless of the brand or what everyone else is doing.
But enough about that, Iām seriously sold on Pixhawk, or I wouldnāt have spoken with my wallet and more importantly, my time. My main worry right now is some missed parameter or PID gain that is way out of wack the first time I spool up. That, and how well is the GPS blending going to work with 2 Here GPSās? And is that even a good idea? Iām assuming it can add some redundancy and possibly precision in GPS guided flight modes and missions?
Tim
I would not say that Iām a veteran user of the pixhawk but I have some experience in flight controls. And I have tried to apply that experience to the pixhawk with some success. What you have to understand when tuning the PID controller of arducopter is that it is a rate controller and as such the P gain serves the same function as a flybar. It provides damping. However the benefit of the p gain over the flybar is that it knows when you make control inputs and thus doesnāt try to damp your inputs(the p gain is applied the the rate error between your requested rate and the actual rate). The I gain serves to hold attitude. In order to have a system that holds attitude well there has to be sufficient damping otherwise youāll see low frequency oscillations (< 1 hz) about an attitude.
So in my opinion, the P gain needs to be around 0.2 but that could vary depending on your setup. What I might recommend is once you think you have enough p gain, fly in acro and see if it feels damped like a flybarred heli. I think that is a good litmus test. Definitely start off with a small P gain maybe 0.04 at most and then increase it and stop once you see (if you see) oscillations. I would suggest a VFF gain of at least 0.05 or some value that gives you 6-7 deg of cyclic. You can back off on VFF as you get the P gain higher.
Donāt play with the D gain. I think a PI controller is sufficient. So the I gain is raised to a value where it holds attitude well. You have to be careful though because there is an integrator leak and you have to understand how to use the ILMI parameter with the IMAX parameter. I like Robās description of this. Think the integrator as a cup that fills with water when there is error between the desired attitude and the actual. IMAX is when the water fills to the top and spills over top thus the integrator canāt get any bigger. The integrator leak is when you put a hole in the cup. How high the hole is on the cup is an analogy of the ILMI parameter. The water will only leak out the level at which the hole is at. Or in our case the integrator only leaks to the ILMI value. This is allows the to hold some value to help hold attitude.
So my thoughts on I gain is that the value should be some where in the 0.1 to 0.2 range. In my opinion, as you get higher I gains, youāll see more lag in you response especially with higher values of IMAX and ILMI. I donāt think you need IMAX at 1.0. I think 0.4 is a good starting point but you will have to look at your data to determine if it needs to be higher for forward flight. You could make ILMI 0.1 or something small. Otherwise, it will start to lean real bad on takeoff with any type of stick input or if you are not on level ground.
Sorry for the book. Hope this helps. I know Chris will chime in with his expert opinion.
The defaults will provide a basically hoverable helicopter in no wind, although probably quite a bit on the sluggish side. The yaw defaults are really āsoftā for a big heli and it may hold a heading and it may not. I recommend cranking up the yaw and getting that locked in first because thereās nothing funner than cranking up the rate pitch and roll and having a helicopter doing a hula dance with the tail snapping back and forth 180 degrees once per second. Iāve done it and never been able to get video of it because Iām too busy trying to fly the helicopter and get it back on the ground in one piece
Bill,
I appreciate the input. The .2 P gain does ring a bell, i think ive seen that before. I think i will try to start conservative, and its very handy getting opinions as to what āconservativeā is. My experience with tubing PID specifically is through Cleanflight with multiās. Most evey FBL ive ever used has such parameters linked into āhead gainā values or something to that effect not P, I and D seperatly, so its uncharted territory for me with a helicopter.
One thing Iām wondering is this. Ive read about people having issues with inputs on the ground and having the heli tip over or dart off in a strange direction. Personally, ive gotten into the habit of not touching the sticks unless absolutely nessecary on the ground and popping it straight up into a hover rather quickly to avoid any issues. Will that method serve me well with pixhawk?
Tim
Chris,
Yeah, totally been there with a fubar tail. Love a high speed spin the second it gets light on the skids because the gyro isnt compensating correctly. Thats always fun, along with the many other issues that can arise with the tail. I think i might go a little higher on the tail gains to start so at least i dont have to chase the heli all over the place and compensate for the tail as well. Hopefully.
Tim
I donāt have an expert opinion on tuning FBL heliās. Thereās just no way to predict what it will take to tune it. I know the defaults will fly a DFC helicopter, although really sluggish. So using them as a starting point should be ok. One of the problems with low headspeed heliās is pitch oscillation that shouldnāt happen, and the only way to fix it is to speed the head up. These little RC heliās donāt run anywhere near the blade tip speed that full-sized heliās run, so theyāre not as stable. So kind of watch for that with a LHS helicopter.
It will work ok. In Acro, with the stock code, there is a problem for FBL and I would not recommend taking off in Acro until you get it tuned and get really used to it. I use the opposite method of takeoff because Iām a scale pilot. I like to lift off real gentle and graceful and smooth. With the stock code in Stabilize you can do that. The rate integrator leaks down on the ground. So if you start āflyingā the helicopter before it leaves the ground it will act kind of like a flybar. Put some roll into it, for instance, and it will try to roll, then gradually āleakā back to whatever position is vertical.
With the default settings for the rate controller I guess I would not recommend just āpoppingā it into the air like a lot of 3D pilots do because you could be in for a surprise. I would tend to be more gentle and smooth and just lift if off 6" or so and see how it handles first.
Chris,
That makes sence, smooth and gentle on takeoff it is. I got into the habit of getting off the ground quickly after a few beast x tip-overs, even though i let my helis sit on the ground spooled up now for extended peroids looking for vibrations and tracking issues, i just got into that habit once im ready to take off.
Ive used many, many different flight controllers across helis, planes and multirotors and i have found its āthe little details and quirksā that can get you and some are way better than others with regards to tuning, flight and bad behaviors.
That is one of the main reasons im trying to build a mental picture of what to expect from this unit, especially when it comes to tuning the PIDs and other parameters that greatly affect flight.
I saw Bill noted that he wasenāt a veteran user of Pixhawk, but in all honesty compared to myself he is along with yourself Chris and i genuinly value the input and experience you guys offer.
There is some basic information out there and i have read through the Wiki more times than i can count along with thousands of posts regarding traditional helicopter, but like anything, it all comes down to how well that information compares to my specific use case. And since ive chosen to use the newest available hardware, there seem to be some holes in the knowledge base. Enter this thread that has already been a boon for moving my project forward.
Later today Iām going to finish up my wiring, snap some photos and get to the business of combing the full parameter list to get things hopefully in to shape for a first flight.
Before the blades go on, i think i will clear the logs, copy my final parameters and spool it up on the bench and see what i get for vibes bench testing. At which point I will upload a .bin and parameter file and hopefully someone with experience diagnosing possible problems can take a look. Ive spent quite a few hours combing logs, but its going to be awhile before i truly know what im looking for.
Thanks!,
Tim
Tim, I found a video of what kind of to expect in Stabilize. Youāll see the heli tipped about 5 degrees left and probably 2-3 degrees nose down. When you spool up, the flight controller will try to level it. So at about 900 rpm during spinup I start putting some left roll and nose down in the cyclic, watching the disc on the heli, to make it stay āplantedā until itās ready to fly. Watch the plane of the rotor disc as itās spinning up - I manually made it stay there, otherwise it will start tipping to the right. As I start feeding it collective I gradually let the cyclic re-center with the result that the left skid comes off first and the helicopter levels out. Then it starts jumping a bit because I still havenāt put in enough collective to actually lift off. But it looks good, so I do and it lifts off smooth and clean.
There should be no real surprises in Stabilize flight mode. The above is a flybar, but itās the same with either flybar or FBL - it will try to level the helicopter. If the heli is on a level surface, then it wonāt do anything. If you donāt want it to do that then use Acro. Just be aware with acro with a FBL that itās like using rate mode in a FBL unit - it will flip the helicopter over on the ground if youāre not right on top of it and make it do what YOU want instead of letting an electronic box fly it, while you make requests to the electronic box.
Chris,
Thanks for the video, it helps to get an idea about how this all works. So if iām understanding you correctly, it is okay to āflyā the helicopter on the ground to keep it in place? I remember some older FBLās did not like any input on the ground and would build up āerrorā and do some weird things, like roll right over the second the rotor got up to speed or just plain dart off in some random direction the moment it got off the ground. With Pixhawk in āstabilizeā it is okay to give input to counter the self leveling nature of the flight mode?
I am probably headed to the hardware store tomorrow to grab a couple 6ā dowels to place in a cross pattern through the landing gear just in case the first spool up doesent go well.
Tim, yes. In Stabilize it should not build up an error and cause a flip because the rate integrator leaks down fairly fast on the ground. Iāve found I can āflyā it on the ground all I want in Stabilize and it never does anything unexpected.
In Acro, itās a different matter.
So basically, the first time you spool up and intend to fly it, place the helicopter nice and level and if it starts to tip donāt just stand there and watch it. Grab control of it right away and stop it, and shut it down. You got a serious problem that needs to be corrected before proceeding, if it does that.
Thanks Chris,
I know I have asked the same question before just maybe in a different way, but I just want to be 100% certain of my understanding of how the controller is supposed to behave and the steps I have to take for a good first flight, provided the vibes arenāt crazy or I nixed something up badly.
Check, check again and re-check for good measure. I learned early on cutting corners ends badly when it comes to a helicopter. Weird sounds, odd vibrations or unexpected behavior, I learned, are a good a reason as any to cut the flight shortā¦ or the helicopter will itself.
Tim
So,
I just spooled it up on the bench without blades of course, but I wanted to have a look at the X Y Z vibrations in MP while it was running after I was done strapping down wires and such. Iām noticing there are two red bars on the display with the values of 30 and 60 that I am assuming are āthresholdsā? While running watching the readout I barely saw any activity on the screen. Maybe values of 6-10 if I had to guess according to the scale.
Iām guessing there as a much better way to get a picture of the vibration profile? Assuming in the logs is where I can find this data, and in the logs where would I look and what is considered bad and good vibrations?
Tim
