This should be interesting

ArduCopter Traditional Helicopter
401

Well, I would say that’s disappointing. Mounting underneath would be the only real good option then. And that causes problems with a helicopter for attaching lifting brackets for a tethered load, or even a camera mount underneath. The side frame mount on a bigger helicopter is my favorite method.

I wonder why Pixhawk2.1 would have a problem with a sideframe mount when the regular Pixhawk don’t have any problems with it? You have to take some action for y-axis vibes because the flat side of the Pixhawk is against the frame side, which acts as a big drum and amplifies vibration. But once you overcome that it works great.

402

Chris,
Yeah i dunno. I wonder if its just erring on the side of caution because as he said “they have no data” one way or the other? The vibrations dont look bad do they from the logs? I leafed theough them and dident see anything terrible, but then again i don’t know exactly what i am looking for. :confused:
Tim

403

If for some reason i have to move it somewhere to sit level, im thinking a couple boom clamps like used with the horizontal stab fin and a carbon fiber plate platform to mount the pixhawk right up front on the tail boom? If not that, then a couple carbon struts and a plate with some hardware right behind the mainshaft. I just dont want to get into all that if its not 100% nessecary.
Tim

404

Tim,
Here is what I see in the logs and I need your help to understand what you were trying to do in each

17-59-43 P= 0.116 I= 0.11 This is just one lift off. I see some small oscillations in roll

18-09-05 P= 0.116 I= 0.11 3 spool ups and I can’t tell if you lifted off for any of these
P= 0.09 I= 0.11 Looks like you lifted off, you had some roll oscillations

18-20-21 P= 0.082 I=0.09 Lifted off with some roll oscillations
P=0.063 I=0.09 Lifted off but no significant oscillations

So far all of these have the ATC_RAT_XXX_FILT parameter set to 20. So the last log I see that you took it down to 4 hz but your P gain is 0.042 and your I gain is 0.06. It looks really good from the standpoint there aren’t any oscillations in attitude. Certainly I see the oscillatory content in the RATE message with Rdes and R, but that is not causing any large roll issues.

I’d say keep the FILT parameter at 4 hz and raise your P gain until you see noticable oscillations with quick stick inputs then back off until they subside. I’m not sure what to tell you about the mounting of the Pixhawk 2.1. I don’t know much about the system.

405

Bill,
The 3 spool ups without lifting off were the attempts to get the tail solid on spool up and initial takeoff. Which were successful.
I have one more log with the filter at 4hz bit it wouldent let me download it without error. Ill get some more flights in tomorrow with the lowered filter and ill keep trying to get the P gain up higher. I have a feeling im going to have to split the P gain on pitch and roll to get better numbers and pitch feels different than roll which makes sence i guess.
As to I gain, it there an upoer limit? I would just look for slow rythmic ossicilations with too much I gain right?
In the short term im going to leave the Pixhawk where it is and once the headspeed is exactly where I want it and if then i have no issues, whatever i guess. I cant imaging its going to harm it, it just might not dampen like they intended?

Thanks,
Tim

406

Oh and, say i cant get over .1 on P gain, but it seems to fly okay? Is that fine? Will loiter work still? If the weight helped maybe i should be tuning this with the gimbal from the getgo? What i was thinking is tuning it without for auto missions with a light mapping camera, and then tuning it with the gimbal for videography and saving the parameters for each incarnation so i can easily swap between modes?
Is that doable? Recommended?
Tim

407

Only problem with is that you are getting your accelerometer further away from the center of rotation of all the axis and it probably won’t tune or handle right. Mounted on the tail boom it’s going to measure higher acceleration in the pitch and yaw than what it really is.

408

Chris,
Hmm, that makes sence i guess, if i side mounted or underslung the batteries i could put it where the packs usually go right unter the mainshaft? Only other good place if im going to keep a canopy is right behind the mainshaft?
I mean, if im not having bad vibrations, what the point in changing where i have it? Dunno
Tim

409

I just got done looking at your logs for vibration and I don’t see a real problem there. If it was me I’d provide 'em with the data they want to say it’s ok to mount the thing on its side. It looks pretty much like a regular Pixhawk to me as far as how it’s handling it.

Your y-axis might be a bit high on vibes, but it doesn’t seem to be causing any significant aliasing of the IMU’s. And I kind of think it’s sound waves hitting the thing causing some of that on the Y-axis. If there’s some room between the unit and the frame, try slipping a really thin piece of foam in there or something - sort of like earphones have - it helps damp the sound waves. Or even put a piece of that reflective stuff like I posted in the first part of this thread on it. That helps too. The more you can dampen the vibration, the better the accel’s work.

And as far as that goes, it’s WAAAY better than my nitro heli, which flies with no issues.

410

Chris,
Well thats encouraging and thanks for the ideas to soften vibes, im sure soon enough ill be attempting to get as much out as i can. After i get a little further along tuning ill send a log Phillip’s way regarding vibes and side mounting.
Tim

411

Yep. That’s what I’d do. I just don’t see an issue with it. Assuming there is some sort of silicone rubber dampers or pegs or something on the IMU’s they should work in all axis’ and mounting orientation shouldn’t make any difference.

On the sticks, Tim, if you had that big of a difference then you were over-tuned. That type of “training gear” has been used for student pilots since like, forever. Usually with tennis balls on the ends so they don’t dig into the grass. I recently cut a 4 foot 1/2" diameter stick in two and put it on my Trex 500 with tie wraps and tennis balls on the ends so my wife could learn to hover it. I flew it with the sticks and couldn’t really tell much difference except maybe just a tad more sluggish with them on there.

She skidded it around on the ground and hovered a bit and used them for maybe the first 20 minutes and then we took them off. But they should not affect a properly tuned helicopter that much. So if anything, taking them off told you you were way over-tuned, as that long of stick were acting like dampers.

Here is the rate settings from my Trex 500DFC, which I currently have de-tuned a bit for my wife. I’m running some custom code so I took out the extra params in the custom code to eliminate confusion. And this helicopter flies beautifully on auto flights. When I detuned it I backed off the P gain a bit so it’s not quite so touchy and increased the VFF like I do for the flybars, which bypasses that rate PID loop and gives the desired control response without the oscillation problems. And basically tuned the P gain so it has enough to do it’s job, but leaned heavier on the feedforward to make it snappy but rock stable no matter how fast I want to fly it.

This has become my preferred method to tune the ArduPilot rate controller. Very happy with the way it handles and my wife as a student pilot can fly it easily. As she gets more used to it, to “sharpen” it I’ll turn up the VFF instead of the P.
ATC_RAT_PIT_D , 0.001000
ATC_RAT_PIT_FILT , 10.000000
ATC_RAT_PIT_I , 0.350000
ATC_RAT_PIT_ILMI , 0.000000
ATC_RAT_PIT_IMAX , 0.440000
ATC_RAT_PIT_P , 0.070000
ATC_RAT_PIT_VFF , 0.120000
ATC_RAT_RLL_D , 0.001000
ATC_RAT_RLL_FILT , 12.000000
ATC_RAT_RLL_I , 0.380000
ATC_RAT_RLL_ILMI , 0.000000
ATC_RAT_RLL_IMAX , 0.440000
ATC_RAT_RLL_P , 0.060000
ATC_RAT_RLL_VFF , 0.110000
ATC_RAT_YAW_D , 0.003000
ATC_RAT_YAW_FILT , 4.000000
ATC_RAT_YAW_I , 0.170000
ATC_RAT_YAW_ILMI , 0.000000
ATC_RAT_YAW_IMAX , 0.250000
ATC_RAT_YAW_P , 0.380000
ATC_RAT_YAW_VFF , 0.000000
ATC_SLEW_YAW , 6000.000000

Here’s how it flew with those settings after I tuned it for my wife to learn to fly with it. I was moving the cyclic in a circle at full stick about twice per second to see it if snapped out of a death shake and it straightens right out without even a wobble. I thought it was a bit soft with gentle cyclic just rocking and rolling it in hover but my wife loves it that way. As she gets more used to it I’ll turn up the VFF without telling her until it gets to where I like it.

412

Chris,
Yeah, it was definatly over tuned if im understanding what you mean? As Bill suggested, after the heavy gimbal is attached i can likely go higher with the PID’s, but what do i gain with gigher PIDS if it flies good where it ends up? Im guessing if i turn the VFF down i will get a better picture of what is really going on?
It is easy to get stuck in a number tuning ive learned, in this case .2 for the P gain.
What i mean by that is this, i set up my Blade 700x with a new FBL this winter, all the literature for the unit stated a cyclic gain of 50-60% was what i needed for good flight. I could not get the gain up to 50% without ossicilations in a hover and upon hard stops. I ended up at 39-43% head gain. It still flew rock solid there, no matter what i threw at it. Flips, rolls, very sloppy piro flips, tic tocs and FFF, all tight and locked in. What the manufacturer and myself deduced is that particular model has high mechanical gain built in and combined with fast servos etc it didn’t need as much gain in the FBL. As that model has a unique setup for the servos and “captured” style servo horns, without a bunch of changes i couldent reduce the mechanical gain, but in the end, i guess I dident have to if it flew well anyway.
Maybe if i swap the servo horns on the Trex im using for Pixhawk i could get some more P gain out of it? But would it be worth it in the long run?
Thinking out loud here, i have the RC_SPEED parameter set at 333HZ, could this be contributing to any negative tendencies with the cyclic tuning?
Ive read a few posts where people got poor loiter performance with low PID’s. Was that really the case? Loiter and eventually auto are two very important modes to me so…
Thanks again for all the help/videos etc.,
Tim

413

Tim,
I know that I preach 0.2 for the P gain but the amount p gain required to achieve the required damping for a good loiter mode is dependent on many things in the setup of the heli. So You may in fact achieve sufficient damping with lower P gain for your set up. That’s why I say the only way to tell if you have the P gain right is to get into a high hover (30 ft) and flip to acro mode and see what it feels like. If it feels damped like a flybar heli then your done. If you are going to do that let me know. There are some changes I would suggest to a few of the acro params.

414

Oh and Chris, i noticed you have a little bit of D gain in there?
What led you to adding some? Is there a benifet over just P & I?
Tim

415

Bill,
When it comes to acro mode, i dont think i understand it correctly? In acro mode, is it just relying on gyro and acts like a traditional FBL and doesent limit your pitch and roll like a self level?
What are the fundamental differences between acro and stabilize?
Tim

416

So stabilize is an attitude command system. What I mean by that is that your stick commands the attitude of the vehicle. For Acro, it is a rate command system which is the basic response for a helicopter, even a flybarred helicopter. What that means is that your stick commands a rate. When you center your stick, you command zero rate. So acro mode does have a special feature called trainer mode that will level the aircraft if you center your stick. I would suggest to disable that feature of acro mode if you are going to use it to see if there is enough damping. Plus it sounds like you are an accomplished pilot that can handle a lightly damped helicopter. Plus you can always flip it back to stabilize if you don’t like what you see.

417

Bill,
Okay, that makes sense. So in Acro without the training mode enabled it sounds indeed not unlike a traditional FBL only using gyros. Take my Spirit for example, if I give some left cyclic it will tilt and go left at the defined rate until I make a counter correction to stop it, that sounds like what Pixhawk is doing in Acro. If it were not a UAS helicopter, that sounds like the mode I would fly in most of the time. In this application I just need a slow stable platform that has position hold and waypoint capabilities.
When flying in stabilize, I must say the Pixhawk feels a lot like the BeastX when in the “self level” trainer mode, even down to the drift I had when using the accelerometers. :confused: I ended up just flying without the stability when working most of the time as it seemed like every other day I’d have to frig with the accelerometer trims…
Now here’s a question, say I never fix the drift in a hover, “which I’m going to, but I’m just speculating here”, when in Loiter will the Pixhawk compensate for the drift and still hold position? Or does it foul things up and lead to toilet bowling or drift?
Tim

418

In loiter mode, it is a translational velocity command, position hold system. With stick centered it should stay at the same location depending on the accuracy of the position info it is using ( GPS aided EKF I believe for the pixhawk). Your stick commands a horizontal velocity.

419

I should qualify that it will hold position as long as the PIDs are tuned well. Most of the time the toilet bowling is a poorly tuned PID

420

Bill, what exactly constitutes a poorly tuned PID vs a well tuned one. Too much gain /w oscillation? too little gain so the system cant control it by itself? Or just not a good blend of P, I and possibly D? Which leads me to the D term. I noticed Chris had a small amount on his parameters. What exactly would such a small amount do for the system? I remember you said that the heli can fly with just a P & I system, why no D?
Tim