Trex 450 Tail Rotor Tuning

bnsgeyer · February 20, 2024, 3:36am

Actually I don’t have any vibration absorption on the FPV camera. Now that you say that, it probably isn’t a bad idea but it does seem to look pretty good.

Well I’m not sure what you mean. In loiter, it behaves quite well. It drifts only as much as the GPS drifts, otherwise it holds position very well.

So is the minipix sitting on a platform or mounted directly to the side frame? Here is a few pics of my T-Rex 500.

You can’t see the double sided tape but that is what I used to mount it to the side frame.

0n-3 · February 20, 2024, 1:44pm

That’s a nice set of mods on the 500, I can hardly believe there’s no dampening on that cam, wow.
yes on my setup I used the same mount I tried earlier with the aluminum. It got my vibes down a little I figure it can’t hurt.

edit… Just for clarity I’ve installed the minipix on the left side of the heli and set ahrs_orient to roll 270
that seems to be correct but I feel like the fc Is fighting me trying to flip the machine before lift off.

I just tried hovering the heli and I’m glad I had the presence of mind to put the training gear back on because wile trying to lift I noticed it wanted to tilt weird I tried twice and just shut it down when I realized it would hit the dirt badly… would you mind looking at this log file and telling me what is up with it? I don’t know enough about log files to see what’s wrong with a setup

https://www.dropbox.com/scl/fi/sfgdw02r0rlv9vg4lvcfv/00000001New-orientation.BIN?rlkey=5o4my25u8s2hrqbvv4gu75yzn&dl=0

I can’t trust trying to put this thing in the air with it acting like this

I can see from the FFT graphs that the vibrations are way down from before though
check below (this is with zero notch filtreing)

but it’s the vibe data (likely due to the training gear) that’s totally out of wack with peaks at 60 when it was at 40 before see below

and here are the nastiest vibes I’ve gotten yet on the pinned down heli where it tried to lean left

I pinned it down and did a run up after a short while I observed hard lean right on the horizon indicator and the swash trying hard to get the heli to lean left… scary very scary I have the log for that too if you want. I’m beginning to wonder if it’s either the orientation or the newly balanced head causing more vibes than before? Which would mean that the head isn’t balanced and I made things worse? man I’m so lost

0n-3 · February 23, 2024, 1:01am

I’ve had quite an adventure over the last few days of testing. Crashed the heli today thinking I had things working well. Lucky I had the training gear on as it decided it would do it’s own thing.
I discovered that I had a bent tail rotor shaft right when I decided to balance the tail rotors due to my noticing most of the vibes were in the pitch axis. Fixed that, tested on the ground, got nice numbers, went to hover aaaaand back on the bench for repairs.

I got fed up and went back with my original mounting solution after I remembered those were the best values I’ve gotten so far. I’ve just now finished ground testing and well the graphs look good to me. The pitch axis seems to top out near 35. I think I read an acceptable range is 30?

I can’t remember right now but after a couple spots on the frame I think the original spot is my best bet because this is the only place the helicopter doesn’t want to tip it’s self over. Finally the pitch axis vibration data isn’t hitting 60 to 70. I’ll post two screen shots and on the weekend I’ll see if it hovers…

After what happened today I went back to my original settings also because I have zero idea what’s causing the heli to just freak out. having several successful hovers on my old settings gives me a bit more confidence right now so that’s what it’s going to be for the time being.

If this thing tries turn it’s self into a shovel again. I’m going to have to consider cutting my losses on ardu. Would rather not but hey my friend needs a flying heli not a hot mess that freaks out on him at the slightest and I may not be the guy to do this sort of build for him after all.

Anyway sorry for the rambling here are the shots

bnsgeyer · February 23, 2024, 3:48am

@0n-3
Sorry to read about the problems you’ve been having with your heli. In looking over you logs, I noticed that your collective is not all the way down. Do you have your collective off of the bottom stop when you spin up the heli? This is a bad practice for heli’s in ardupilot because the flight controller uses the collective to determine when you want to start flying. It changes certain things about the controls when it thinks it is flying. That combined with the vibrations causing the EKF to indicate the aircraft is moving even though it isn’t causes the flight controller to allow larger inputs while it is still on the ground.

30 is still pretty high but anything below that is indicated in mission planner to be acceptable. X and Z look acceptable and probably will get better after lift off. Just a note about the VIBE signals. X is the longitudinal acceleration (linear fore-aft movement), Y is lateral acceleration (linear side to side), and Z is vertical acceleration. The reason I mention this is that you referred to the pitch vibration being less than 60. If you are referring to the Y Vibe signal, that is really side to side movement. But catching the tail drive shaft being bent is still important to help with vibes.

One check to see if the EKF is behaving might be to look at the XKF1 message in the logs. There are the VN and VE signals which are velocity north and east. Check to make sure that these are near zero. They should be reading less than 0.5 or lower. If you look at the log you posted, you’ll that they get very large when the vibrations are bad. It causes the flight controller to think it is actually moving.

Its disheartening to read about the troubles you are going through. Part of me thinks that it is the controller. I have not been that impressed with the Radiolink controllers. If you have another controller, it might not be a bad idea to see if it gives the same readings. Personally, I stay away from radiolink products. I have heard many good things about Matek controllers. I think there are other good lower cost options.

heri · February 23, 2024, 9:03am

@0n-3

You have changed the position of the FC. The orientation parameter must be changed and the leveling must be performed again. If you have only changed the orientation, the FC will go crazy as it still has the old level values.

Could this be the cause of the FC going crazy?

BR

Heri

0n-3 · February 23, 2024, 11:13am

yes I did that also. which is why I can’t understand why it freaked out. I do that each time and calibrate the compass and well calibrate for level as well.

Ferrosan · February 23, 2024, 11:23am

@0n-3 Quote 100% what @Bill recommends. Additionally, I have flown for quite a bit of time (~1 year - about 250 flight hours) Matek boards ranging from the cheaper F405WMN to the more powerful H743 Wing on a wide range of helicopters, from the little Protos 380 FPV to big Gasser. Never had a failure or misbehaving whatsoever, doing FPV aerobatics and autonomous flights. You just need to take some extracare in mechanically damping vibrations as these boards don’t come with internal dampening of the IMU-whereever necessary.

0n-3 · February 23, 2024, 1:46pm

hm well if what you’re saying is so, then that main shaft is bent. I did runup’s on the ground after the slight crash every thing looked good although I know I need to pull it all apart ran up anyways. Blades tracked and vibes seemed regular. now that I’ve repositioned and gotten lower vibes if that Y value is not vertical I’ve got to check that main shaft and feathering spindle like I planned to anyway and I’m sure I’ll lick the problem by this evening. I did this run this morning with rubber dampeners on the landing gear
ground run dampened landing gear

Yes I bring the blades up to zero pitch before I enable motor interlock. Given I have a 3.5 negative pitch I didn’t want to put stress on the landing gear and heli with a down force but lesson learned

bnsgeyer · February 23, 2024, 6:35pm

Well then you need to be more careful with where your collective stick is while you are on the ground with the rotors turning. There were several times in your log where it declared the aircraft had taken off and there was no intent for flight. The code looks at the collective blade pitch and if it goes above half way between the hover collective and zero collective, then it triggers takeoff. So if your hover collective is 5 deg then it will indicate takeoff when the collective is above 2.5 deg. Point is that you shouldn’t just raise the collective to what you think is about right. In some cases I saw that your collective stick was 40%off of the bottom stop.
I would suggest that you mark the collective stick point that gives you zero deg pitch. That way you know it will be right. That would keep the controller from trying to make large inputs and potentially rolling your aircraft over while you have no intention of flying.

0n-3 · February 23, 2024, 11:41pm

You know what, come to think of it the fc has told me a few times that it landed. I didn’t think anything of it. I’ll definitely look into that.

Thanks for all the help.
I think I’ll finally get it going now, there’s so much more to learn about this software. Sooooo

Is 26 good?
I got home, changed the main shaft, (forgot to check the feathering spindle - will pull the head apart tomorrow), ran up a few times, while trying to sort out the blade tracking…

Got some issues…
(hoping it’s just a bad feathering spindle)
but got it tracking somewhat decent, these are the results.

Zoomed in, it’s showing the highest noise is at slightly above 26%.
I moved the throttle a bit to see if I was getting the tracking issue and it showed up as I advanced the collective.

I tried to look at the xkf1 vn ve values.
They look horrible.

So far this is a great improvement. Here’s the log if you wanna check it out.

latest log better vibes than before

Let me know what you think if you do!

Thanks again for all the help. Really appreciate all the help everyone is giving me with this. I got as far as I could alone and the rest seems harder than everything before. I really need this help here!

0n-3 · February 23, 2024, 11:53pm

I do have a matek fc that I planned to use on my 500.
(if only someone allowed me to do this ardu set up on my heli when I was good and ready rather than insisting that I use theirs as a guinea pig… insanity…)

If I can’t get the radio link going, I’ll definitely have to give that a go.
I need to 3dprint a case for it with dampening built in that’s why I’ve not looked at it yet.

Thanks for those board ideas I’ll look into getting that f405wmn, looks interesting.

I noticed something while I was going through my settings also.

Seems when I changed the orientation on the fc I may have accidentally did the same thing to the compass!

When I typed in ahrs_ and searched for orientation I edited a value not realizing I’d messed with the compass…

It was rotated to "roll 270 just like the fc… smh…

That might very well have caused the fc to go nuts and turned the heli into a shovel.

Doing all this work made me acutely aware of other places with vibrations.
I noticed yesterday that one of the tail brace supports was acting like a guitar string.

I 3D printed a stabilizer for it and got a nasty reminder that putting that thing on really just made the vibes go from low frequency to high frequency and gave me a worse reading on the vibe scale haha.

I took off and realized that the dang screw was loose, thread locked it good and the vibes went away.

Next the new landing gear with the rubber on the skids really help with the ground-run graphs finally got better graphs today.

So progress… we’re on our way finally.

Ferrosan · February 24, 2024, 12:41am

Part of preflight checks, attitude and heading on MissionPlanner consistent with aircraft movements .

You can save a lot of work time by simply removing one rotor blade, put an allen key into the spindle drag bolt and slowly turn the spindle. If it is bent you will spot straight away the opposite blade tip swinging up and down .

0n-3 · February 24, 2024, 2:17am

I always do the lean tests to see if it’s behaving correctly and it always did, however I will admit that I didn’t observe that the compass was misbehaving from my yaarpu telem display on the radio. I just now experimented with the heli held on it’s side to see what happens to the compass and it still points where it should

Thanks!! I’ve heard of this before but I’ve not flown heli in years! Thanks for the reminder!

So I just tried it and I can find no evidence of a bent spindle… so I guess I’ll have to figure out what to do about my tracking issue otherwise, but tomorrow, I’m done messing around with the heli for the night. Hoping for a great test flight tomorrow. looks like I just need to double check settings I’m ready for a hover

0n-3 · February 24, 2024, 2:56am

BTW which parameter controls the max tilt/pitch/bank angle in stab and acro modes guys?

bnsgeyer · February 24, 2024, 3:56am

The max pitch and roll angle that can be commanded is done with ANGLE_MAX. the default should be 30 deg. For the parameter, it will read 3000 centi degrees.

Acro is fully aerobatic however if you have the trainer mode on, I believe it does limit angle. But I would not recommend flying anything but stabilize or althold until you tune the controller.

0n-3 · February 24, 2024, 11:49pm

Okay gotchya. I asked because when the heli crashed last time, I was unable to correct it’s trajectory with the sticks. It veered off and ignored my inputs or I should say, my inputs slowed it down but not enough. Maybe a larger angle would help? Then again that also gives the autopilot more to play with so prob not.

I actually wondered if I had put it into acro if that would give me full control of the machine rather than the autopilot doing the flying.

I’ve hovered a little in acro before it’s a bit more sensitive but manageable.

In anycase, 23!!!

I tinkered today and the ground-run yielded a sweet looking 23.

Check it out yo! aint it puurdy?!

I wish I could say the same for the xkf1 values but this is a ground run so maybe that’s why?

and the other vibe graphs…What notch filter settings would your recommend for this one?

log file below.

https://www.dropbox.com/scl/fi/wlt975r2vmrb1zr6nli84/00000033.BIN?rlkey=3j2ppdweh0xq8e3zwvbbte4vg&dl=0

After this test, I put it on my makeshift test stand that I use to tune the tails on my birds and finally settled on the settings you posted at the beginning of my questions. I’ll stick with that!

Hover test tomorrow God willing, (first with the training gear just to be safe) hoping for a good result. Then it’s back to fighting to get stable hover.
Vibes should be much lower in the air as you’ve said, so hopefully they wont play a role in the hover and it will be easier to keep it stable and staying putt!

Previously, on my very first hovers I was playing with the
atc_hov_roll_trim
as well as the
ahrs_trim_y
as suggested by the documents but it never seemed to hold hover I don’t know what I was doing wrong. Fingers crossed this time I can get past this stage and into auto tuning.

Oh I wanted to ask, where do you go in the log file to see all the inputs the pilot makes and the parameters I’ve got set?

I need to figure out properly how to determine how much roll and pitch I’ve used to level the craft so I can put those values in to atc_hov_roll_trim and ahrs_trim_y

I’m really interested to hear your suggestions on notch filtering with these results, so let me know what you think.

Wish me luck!
Thanks again.

Edit just saw this… xkf and vibe values. The averages on the xkf1 looks actually decent no?

bnsgeyer · February 25, 2024, 3:45pm

No. I wouldn’t change the max angle. 30 deg is plenty.

You have full control in stabilize mode. It is just attitude command rather than rate command. All heli’s are rate command meaning that if you make an input with the stick and hold it, the response will be a rate (pitch rate for elevator stick input, roll rate for aileron stick input). In stabilize, the stick inputs command attitudes and when you center the stick the aircraft holds zero pitch and roll attitude (well it holds the hover roll trim for roll axis). So the autopilot isn’t doing any flying other than making the aircraft respond a certain way to your inputs.

Yes I think that is good.

Yes they look good.

I won’t have time today to look at the log file and make a recommendation on the notch filter.

To see the actual stick inputs, they are in the RCIN message and then you have the channels shown.

To see the parameters that were set for the log file, you can either click the table check box below the graph in mission planner if there isn’t a table showing. Then just filter the table for the Param message. Or you can click the check box for parameters and that will bring up the parameter list page with the parameters that were used.

0n-3 · February 25, 2024, 10:06pm

Okay thanks man, I’ll see If I can get some help via the discord if you’re around today. It’s cool.
Great Great News Though!
Did the test hover with and without the training gear.
(I needed security and a confidence builder after the last incident lol)
with the training gear the vibes were much lower then last night and without it, lower still!
It really is a thing of beauty now haha.
(aye each to their own okay)
Thanks everyone for the help and insisting that I sort out my vibes. I’ve done heli so many times without fc and never had issues that I couldn’t understand why I had to pull out my hair to get this thing working right. It makes more sense now.

check out these graphs

Vibes

FFT

RCIN

I have no clue how to translate the RC in values into degrees or rad to continue… heeeelp haha

Log
https://www.dropbox.com/scl/fi/bwkzx5ewa27sobjbixwbf/00000035.BIN?rlkey=k3fd02h4yb0c15wm5orgiupcm&dl=0

EDIT
I read farther ahead in the documentation to learn more about the logs to see if I could figure it out. Seems ATT is where I should be looking and from what I’m seeing. The last portion of my flight was when the wind was most calm and seems I need about 9 degrees of right roll and under 7 degrees of back pitch… Ummmm That’s er quite a lot!

I graphed the ahr2 values also and I think I’ll prob try
pitch value +5.45
roll value +7.74 or start with 4.46 instead
I’m basing these values on the calculated mean values displayed on the chart with the ahr2 as reference. I figure less wont hurt I can always add more and I’ll try the hover with the training gear again to be sure it’s not over kill.

Let me know what you think about that when you have a chance.

I’ve set up the notch filter as below, let me know what you think of this.

0n-3 · February 27, 2024, 10:38pm

well seems like I still have vibe issues
got some kind of ekf failure with this flight with the notch filter on.
barly made it into the air and heli the stick for stable hover after inputting a few values and the heli tried to drift fwd again

log here
https://www.dropbox.com/scl/fi/xeb40m47cm4vayk98fv10/00000037.BIN?rlkey=y7bri9log576jd9flf2ows2ol&dl=0

time to go hunting again