Servers by jDrones

This should be interesting

(Timothy Bacon) #1

This is getting a Pixhawk 2.1…

First test bed…

Trex 700 stretched to 800mm swinging the 800mm wide chord Align blades with an 800mx 440kv motor and all gears upgraded to helical plus BK servos.

This may be the king of all bad ideas as i have learned i have to use newer untested/proven firmware and it seems big helicopters create notoriously hard to tune vibrations, but alas, i do love a challenge. As that helicopter carries a gimbal from time to time, ive spent countless hours getting it as smooth and vibration free as possible.
Time will tell, either way its going to be an interesting road for sure. Logic would denote i get an older flybar trex 700 to start with, but im hoping my tenacious attention to detail will serve me well and i can get this project off the ground.
Less fun is the 1000mm octocopter that is also getting a pixhawk 2.1 and here+ gps.
Helis are my passion, so that gets the love first! Honestly, for a video platform my helicopters vastly outperform the octa any day so that makes more sence to start with. Faster, handles wind better and after alot of tuning, no jello ever.

Trex 700/800 parameters for pixhawk
(Chris Olson) #2

I’ve flown 3.5 some and I don’t think you’ll have a problem with it, although there is a lot of new stuff in 3.5 that still needs to be tested. My settings from 3.4 carried forward into 3.5 with no problems with my 500DFC. I haven’t seen a Pixhawk2.1 on a heli yet, so that will indeed be interesting. Flybars make a better camera platform than FBL does because they are more stable and only require attitude controller compensation from the autopilot for wind. The flybar does the rest as far as rate and damping and there is no multirotor I’ve seen that can match a flybar heli for stability. A flybar can do jello-free video with no gimbal at all using a hard camera mount. So it will be interesting to see if you can get a FBL head tuned well enough for a camera platform with your new hardware and software.

Since I have not seen a Pixhawk2.1 in person yet, I hope you post some photos of the installation, as far as how it fits without (hopefully) doing a mod to the frame to mount it underneath.

(Timothy Bacon) #3

Well, regarding 3.5, that is good to hear. I totally expect problems with new firmware on anything. Heck, my phone, motherboard in my workstation or even my smart thermostat for my heating system all experience hiccups after a firmware update and usually some months later it is patched after some real world beta testing. The big difference is, it seems in this hobby/commercial world, is that people expectl perfection every time and if something goes wrong even with a beta firmware, they lash out and bash the hardware and every developer associated with it.
As i maintained, even during my issues with procuring a unit, im okay with real issues like firmware hiccups and the like. Thats all part of beta software/firmware and development. Ive had the experience repeated many times with many different things over my years of working with open source hardware/software.
But i digress, regarding flybar helis, i understand the bonuses and the detriments. I originally chose to go FBL when i designed my aerial rig as it eliminated some mechanical complexity at the sacrifice of many years of trouble free use/testing in the field already. It took alot of work getting the system to fly correctly and without ossicilation or vibration, but in the end it just plain worked well and that was using a simple AR7210BX. I never did get the stabilisation portion of the firmware to play well with the under-slung gimbal i use, but piloting it manually using just the gyros it worked awesome.
Ill see if i can dig up a photo of the rig, it is a rather large gimbal usually used on a 1000mm+ heavy lift multi adapted to hang under the helicopter. Ive flown anything from a 5DMKIII under it to a Nikon ASPC DSLR attached to an Atomos Ninja 2 RAW field recorder (My favorite combo, gets great clean video).
Ive got a few mounting ideas in store for the unit. Ive got some carbon plate im likely using to enlarge the forward gyro platform on the Trex, ive taken some caliper measurements and it looks like i can fit the Pixhawk there using foam tape for mounting. If i have to end up using the shock mount, ill fabricate a solution to go behind the mainshaft above the boom block. Either way that should get the unit centralized on the xyz axis quite well. The GPS will get an extension cable made and will go on the horizontal stabilizer mid way back on the tail boom.

Regarding using FBL, ive personally found that the key for me has been very meticulous setup, down to half turns on the links as well as just the right dampeners and a tight, slop free build up of the head in general. I balance the blades as good as my equipment ill allow and set gear lash just right as well. Ive found FBL units, for the most part will fly no matter what, but if there is an issue in setup and its not spot on, there will be vibration evident when it comes to video.
That being said, ive found that since its a large gimbal, unless its really bad, even though its still using servos, it gets pretty good footage.
Another thing ive found is to get rid of that slow ossicilation evident when hovering because of the low hanging mass of the gimbal i run lower head gain than someone would employ for 3D flight. Using the BeastX i had no ill effects of lower gain save for more sluggish behavior, but doing camera work it was never an issue. It also handled wind well even with reduced gain.
I have no illusions that this is going to fly well out of the box, im sure there is a long hard road ahead, but im up for a good challenge and am totally willing to put some genuine effort into this and maybe turn it into a good platform. Its quite clear multirotors take center stage, but hopefully a successful use case can generate some heli love here.?

(Chris Olson) #4

I think for the most part it takes a little more time to learn to fly a helicopter vs multirotors. The mechanical simplicity and “fly out of the box” with modern electronics is what has made multi’s so popular. But those who fly helicopters are pretty much already with the “heli love” thing and have gone to them because of the performance and flight time.

I’ve only had to use a shock mount for a Pixhawk on piston heli’s. On an electric or turbine you should not have a problem with the 3M 9448 double-sided tissue tape. I have had a couple helicopters that had vibration issues after mounting with the tissue tape, but felt it was better to fix the vibration issue rather than “patch” it with a shock mount.

Piston engines, on the other hand, shake the whole heli really bad. There, a shock mount works good. My heli’s have belt drive tail, which I prefer over torque tube. But one other area of vibration in torque tube drives is the driveshaft itself vibrating in the middle, and you may have to add an extra center bearing in the tail tube for that with Pixhawk.

Have you considered a frame side mount? I’ve used that with great success on bigger heli’s, just stuck to the side of the frame with tissue tape.

(Chris Olson) #5

Tim, here’s something I discovered by accident. The Pixhawk on this heli is just stuck to the side frame with 3M tissue tape. I had Z axis vibes 35-40 and x and y around 15-20.

I wanted a cover on the Pixhawk so I stole the sun shield thing out of my wife’s truck, which has a heat reflecting surface and a very thin layer of insulation behind it. I cut a piece and put it on the Pixhawk, stuck to it with mounting tape (the goo stuff and not tissue tape). Putting that stuff on the Pixhawk dampened the vibes to 20-25 on the z axis and 5-8 on x and y. And that was not its intended purpose :slight_smile:

I don’t know if this would work every time, and I can sort of understand how it dampens vibrations that get transferred to the case of the Pixhawk and keeps them from getting to the IMU. But it’s something to keep in mind to try.

New photo by Chris & Kristin Olson

(Timothy Bacon) #6

All great points! I appreciate the input!
Regarding the torque tube, already have another bearing in place, that was something i worked through on that heli eliminating vibes after i strapped the long 800 boom on it. Its amazing how much one bearing reduces vibration, dunno why align never made it part of the upgrade in the first place?
As to side mounting, that is certainly on the table. I think ultimately, in line with what you found already, where it ends up will be wherever vibes are reduced the most.
Slightly off topic, but in the realm, i have anothet heli i just put a Spirit Pro fbl on. Prior to the spirit i had a beast X on it and it flew for 2 seasons without issue. With the Spirit in the same exact place it gets vibes it does not like and i had to keep gain low to prevent unwanted ossicilations at first. I ended up puttimg different tape on, strapped it down with velcro and now the gain is way up and it flies rock solid.
Moral of the story is, even though there were no “bad or high vibrations”, the Spirir just dident like the frequency and a little mounting trickery solved the issue and made ot a happy unit.
I think alot of all of this is “whatever works best for your application” sort of a thing and even two identical models will resonate differently and require specific accommodations.
Point in hand with you using something from your wifes car, perfect example of getting the job done with what is nessecary!
Ill be putting some time into the 800 Pixhawk project soon, right now the sun is actually out so im putting some flights on my 700’s before it rains again.:confused:

(Chris Olson) #7

Cool. Be looking forward to seeing more of this project.

If you don’t mind me asking, what holds the cube to the carrier board? Does it snap on with clips, or snaps, or retained by screws, or what?

(Timothy Bacon) #8

4 small screws hold the cube in place, also there is the connector in the carrier board.
I suppose if you wanted, you could buy some of the connectors, make up a breakout cable and mount the little cube remotely from the carrier. That would be a handy thing in my opinion, could bury the cube in the frame and protect it. Then then brains of the operation would be the size of a smaller FBL unit like the Spirit for example and all of the mess could be somewhere else.

(Timothy Bacon) #9

Just thinking out loud here, but i wonder what seperating the cube from the carrier board would do to the vibration profile? Complete assumption here, but as helis tend to have lower frequency vibration i wonder if decreasing the mass of the main unit would help in that department?
You look at all the fbl’s out there and they all have a pretty small case, even if they offer things like gps and other sensors. Other things seem to be handled by another unit placed elsewhere on the frame. That may be completely by accident, or just because of manufacturing efficiencies, i wouldent know?
What ive been trying to peice together is why modern fbl controllers are nailing it with flight characteristics, and pixhawk, having access to more processing power and attached sensors, is seemingly behind the times.
In part, im sure alot of it has to do with multirotor popularity, but looking at the market it seems helicopters in the commercial world are being used more and more.
Just the other day i had my large heli based AP rig in the air and it was quite windy. Windy enough that i wouldent have bothered to even take off with the octa copter, unless I wanted to watch it balloon and fight it all over the sky. But the aerial pig of a heli i use just sat there like a rock with very minimal collective needed to keep it in a consistent vertical plane.
Let alone the huge advantage in fast forward flight or any yaw needed.
I truly believe if someone could write code that gives good default parameters for a heli that would fly, maybe not tuned really good, but flyable like every fbl on the market today, there would be many more adopters of the system. It has to be possible as many others have done just that, but boy oh boy do they hold on to their code and algorithms like its solid gold.
Either way, im going to give it a genuine effort and truly hope this new hardware was the right decesion. I intend to attempt to employ some filtering if its needed, but am relying on the smooth running platform to help. That being said, if something gets out of alignment in the air, i dont want it to suddenly ossicilate to death with no possibilty of saving it with the sticks. There has to be some way to get better gains and more stable flight with a heli out of this system with pure fbl.


(Timothy Bacon) #10

So, I have spent a little time on the project. I linked the unit to MP through my telemetry radios and all seems good. Had to do a little connector re-routing as the radios I have use different connectors. Nothing major though.
So as I sit here looking at the pile of wires coming out of the Pixhawk, something is evident. The GPS cable for the Here is quite short… I envision the GPS on the rear horizontal stabilizer of the helicopter, but with 10" maybe of length that isn’t happening without some modification. I wonder, is there an extension cable of the same type available? I’m quite meticulous with wiring and hate ending up with large bunches of random wires ive soldered up to get something working.
I’m the guy that cuts servo wires down so they fit the frame exactly without a bunch left over to zip-tie somewhere. I love SBUS and its one wire hookup to FBL’s and the like, and am anxiously awaiting ESC integration with the Spirit FBL as then i’ll have just a few wires total on my 3D heli’s whilst still having access to 16 channels and full telemetry without all the sensors.
Point is, I want to do this clean so if that same cable in length, or a pre-made extension is available for that 8 wire flat cable going to the GPS I would be a happy camper.

(Markus Müller) #11

Hi Tim,

you could use a switching board i.e. Pololu and a FBL-unit. The board switches the Pixhawk in or out on your command to either use Pixhawk+FBL-unit or pure FBL-unit.

(Chris Olson) #12

Tim, I did not have a problem getting stable flight with the Pixhawk with FBL. But I’ve only ever done one heli. The rest are flybar and those are simple.

The problem I am scared of is EKF blowup, and like Marcus suggested, you will only have manual control of the helicopter if that happens by using an external FBL unit with some way to bypass Pixhawk. The bypass is done internally with flybar because Acro is straight thru with flybar mode 1. With FBL it’s not, so you may want to consider Markus’ suggestion for ultimate failsafe reliability in the event you lose EKF. And I don’t think EKF can be turned off in Copter. There is a AHRS_EKF_TYPE param that has 0 value to disable it, but I think that reverts to regular EKF like in Copter 3.3. The other two options are EKF2 and EKF3.

On your GPS cable, does it have connectors on both ends? And if so, what type of connector?
JST? Can’t you just take the connector apart, unsolder the terminals, and make up a longer cable out of a piece of Cat5e 8-wire twisted pair and put the ends back on it? That’s what I’ve done with other uBlox M8N GPS/compass units. Except the Chinese uBlox units come with the wire going into the unit with no connector. So have to take the unit apart and de-solder the wires from the GPS board and compass module, and solder on the longer cable and put it all back together. Has always worked fine here.

(Timothy Bacon) #13

Regarding EKF, i have read a few accounts of things not going well with it. Hopefully i dont have issues, im only using sensors in the Pixhawk and the Here GNS, so im hoping with reduced load…
As for the cable, i just used cat5e and soldered. The connectors are the new ones used by the Pixhawk 2.1.
Last night i deceided to boot it up and have a look through MP. Sone issues there, horizon was pooched as i calibrated on the bench and then side mounted it. I attempted to compensate for that by adjusting the parameter for orientation, but never got it to calibrate the accels after i got the horizon displaying correctly.
I reloaded firmware and started from scratch, lemme tell ya its alot easier calibrating everything with the unit off the frame as its an 800mm helicopter…got it done though
Rc calibration went as expected, only issue i have now is a “motor interlock failure” when i attempt to arm the unit.
Also, i get no movement from the servos at all, im assuming they only operate after arming?
Ill have to look into it further tonight.

(Timothy Bacon) #14

Update… Servos are working, swash is moving, took a bit of time to get everything set up properly and moving as it should… Set to 60,60,180 with a phase offset of 60. Seemes to behave correctly.
Now, as to how the gyros behave in stabilize…? Im used to a standard fbl, and to verify correct compensation you tip the heli and the swash corrects in the opposite direction, is this true of the pixhawk as well?
Im looking for a way to verify correct operation on the ground before each flight as i do with other systems.

----Specificity regarding Pixhawk 2.1… How hot is the unit supposed to get during use? I bumped it after i was done setup this evening and was suprised how hot it was. I dont have a quantifiable number, was too lazy to go get the infra red, but ill be sure to have it nearby tomorrow for a reading.

----also pixhawk 2.1 related, i keep getting “bad gyro health” messages in MP. I did calibration and all seemed good?


(Chris Olson) #15

Yes, that is correct Tim. You’ll have to adjust the rate PID’s and feedforward to get the swash movement you want. Frankly I kinda go by “looks right”. If the swash seems to compensate too much, I turn down the feedforward. If it looks sluggish, I turn it up. When I fly it if I get rockin’ and rollin’ I turn down the P gains and turn up the FF. If it’s steady I turn up the P gains until it’s not, and then turn 'em back down to where it is.

Once you get it tuned to your liking I recommend doing quite a bit of flying at different speeds, etc in Stabilize before moving on to the altitude controlled modes. A heli that’s stable at hover can have problems in high speed flight. Better to find those problems where you got a little more manual control of the heli so you can tune them out with less drama.

(Pitt) #16

This is normal it has heated IMU.
Sensors run optimally at 60°C.

You can turn it off if you are doing long bench test by setting IMU_TARGET_TEMP to 0.

About bad gyro health, try waiting for it to warm up then reboot.

(Bill Geyer) #17

Just remember that in a flybarless rotor the P gain is much more necessary than VFF. Always add more P gain and if it gets too snappy then back off of the VFF. If you can get away with zero VFF and a high enough P gain to satisfy desired response then that is the best way to go. If you start seeing an instability then you need to back off on P gain.

(Chris Olson) #18

Tim’s 700 heli has the same servo linkages mine has, with push/pull and bellcranks. Mine is flybar, Tim’s is FBL so they use different, more powerful and very fast servos in the FBL ones. These are very snappy handling helicopters made and set up for extreme 3D flight, including the flybar ones. But they are also quite easy to tune compared to a little one because of their size. And Tim’s is a stretch mod so he’ll be running quite low headspeed. Should be a really stable machine.

Can’t tell from the photos if you put the bigger 800 blades on the tail when you stretched it, Tim? If you didn’t just be aware that the default yaw rate settings are a little “soft” on rate P so as to prevent a surprise on your first takeoff in the event you’re not right on the anti-torque to counter it. On a bigger heli I would kind of recommend turning up the yaw rate feedward to about .05 if you use the default rate P to make sure you got plenty of yaw authority with the radio to counter it if it doesn’t want to hold its heading. You can tune the rest later once you find out how it acts.

Just for reference, this is the settings I run on rate yaw on a 700 with the servo linkage in the 16mm throw position. You could likely just stick my numbers in there and not have any surprises on takeoff other than maybe a tail wiggler if the D gain is a little sharp for your setup.

Loosing yaw control on a big heli is not fun.

ATC_RAT_YAW_D , 0.007
ATC_RAT_YAW_I , 0.17
ATC_RAT_YAW_P , 0.35

(Timothy Bacon) #19

Regarding the servos and linkages, yeah the cyclics are very fast compared to most, probably overkill, but i got a great deal from Bert K at OHB this past December in FL, so i grabbed a bunch while at the funfly.
Now i decided to leave the push-pull linkages on for my intial testing to see if that allows me to get the gain a little higher, i have the newer FL style setup on another heli and was shocked how low of a head gain i can run on my regular heli fbl’s, so i assumed Pixhawk would be no different. Thise newer head styles with a direct link to the swash are very efficent and fast with todays crazy fast servos. My cyclics are faster than the tail servo that came on the heli was eapecially as its HV at 8 volts. :confused:

Now the tail, i spent many hours getting it buttery smooth. Sanding, shaping, polishing and lubricating with some Dow Corning 111 so its leagues better than it was when first built. Also, i upgraded the tail box, auto gear and jackshaft with helical gears that run at a higher ratio than stock. I also run Rail 115mm blades vs the 105mm stock aligns. Needless to say, the heli is purpose built for LHS running. Oh boy did the helicals really smooth out the drive train, and it has good holding power at some pretty low rotor speeds.
I appreciate the setup advice and the numbers, to that effect i was noticing some very sluggish tail behavior when in active mode under heli setup. Very multirotor-ish if that makes sence… I definatly have alot of setup still ahead of me.
Oddly enough, calibrating the esc went seeming well. Usually calibration with the Castle edge/ice2 esc’s can be a chore with other FBL’s ive used. Eventually i plan to swap out the Castle with a Kontronik with AFW, but for now it will do.
Hopefully being methodical and not rushing this will lead to a successful first flight. Having an 800mm heli spinning wildly in front of you, or darting off in a random direction is not a fun experience. Been there, and since its swinging the 800mm wide chord blades, they look like spinning canoe paddles, or a flying wood chipper. :confused:


(Timothy Bacon) #20

Thanks for the info. 60c is quite warm, 140f, so id say thats about where it was then. Dident realize it warmed up that much.
Will definatly try a reboot if i get “bad gyro heath” again.