Servers by jDrones

This should be interesting

(Bill Geyer) #641

Below is a link to the software and instructions on setting up the notch filter. This software build is based on AC 3.5.0-rc7 and should be considered beta software. I have checked it in the SITL to ensure it boots properly and flown it in the SITL just to make sure there weren’t any issues. Like I said before I have transferred this piece of code to other versions without issue but it is always good to use good buildup. You might want to fly it with notch filter at zero with your current settings just to baseline things.
So you have done a few of the steps of my instructions already because you have determined your instability frequency in roll at least. Have you determined it in pitch? The big thing to remember is anytime you change the notch filter parameter, you must cycle power to the pixhawk to reset the filter. Otherwise you could have unexpected results.
Software and instructions

I also added the parameter that allows you to adjust leak rate for the integrator. The parameter name is ATC_I_LEAK_RATE. I would suggest setting this to 0.001 and setting ILMI to zero. This will help with retaining more integrator in a hover and should help with any leaning when taking off from uneven terrain. I would say keep me posted but I know you will.

(Timothy Bacon) #642

Thanks Bill. Much appreciated. I have some more logs to review, i adjusted P on pitch until i got some worse ossicilations then backed off and copied down the values at the time and which log it was associated with.
Needless to say i have a ton of logs to review. Had alot, and i mean alot of flights these past 2 weeks trying to tune the heli. Charged 8 packs at a time sometimes twice a day and flew flight after flight.
In any case, ill heed your words of warning about being cautious and rebooting the Pixhawk between filter changes etc.
Question, when i re-flash the firmware, can I just save my parameters and then re-import them once the new firmware is flashed? I would certainly check through the ones I’ve changed pertinent to my heli, but all the parameters for swash servo setup, endpoints, govoner etc. Are those possible to import so i dont have to re-do the entire setup? Or is that not advisable?

(Bill Geyer) #643

Before I flash new firmware, I save my params. Then I flash the new firmware. Then I load the saved params and write them to the pixhawk. That should work well since there wasn’t a significant change in versions.

(Timothy Bacon) #644

Thats what i wanted to hear! Sweet. Now i just have to create some time this weekend to implement… Good old memorial weekend… Still dont understand how days off translate into being busier than a work day, guess its one of lifes great mysteries…or it could be the… :confused:

(Bill Geyer) #645

LOL! Yes I will certainly be glad to get back to work after this weekend. I figure if I can bang out a good portion of the honey do list today and tomorrow, then I might squeeze in time to work on the X3. or maybe fly something else.

(Pitt) #646

Tridge said it will change name and rescale parameters automatically.
There are some parameters name changed which caused few problems earlier but this got taken care of.

(Bill Geyer) #647

One other important thing to remember when tuning with the notch filter. Do not set the notch filter frequency much lower than 4 hz. The reason for this is that the notch filter is filtering the error between target and actual rate so you are filtering out your inputs too. I am working on an implementation that only filters the gyro signal.

(Timothy Bacon) #648

Pitt, thanks for the added info!

(Timothy Bacon) #649

I remember reading something about that, most likely in your thread regarding instabilities? That makes sence though and explains why you would want to filter the gyro signal vs error quite well.

(Timothy Bacon) #650

Did you just get a rider for an existing policy for your UAS? I know my insurance companies dont sweat the equipment replacement costs, even though its roughly $25,000 to replace it including camera gear etc. It was the bodily injury and major property damage they worried most about.
It should be noted we have had a couple claims with the insurace company covering the UAS equipment and they dident raise our rates. It was non UAS related, water damage by tenants in our apartment building, but nonetheless they dident give a rate hike.
I completely understand going with what you know though, hence the reason ive stuck with FBL. It is what I have the most experience with, and although i havent used Pixhawk up until this point, it certainly has a farmiliar feel to it. Ive flown a few flybarred models at the field and as I recall it was an altogether different feeling in flight. They are just very different characteristic wise. I think i have just learned to correct for the snappy behavior automatically over the years so it just feels right to me now.
I dont know how the pixhawk will respond, but my 3D models always get a varied flight. I usually do a couple min of low headspeed work as im digging the smooth flow these days, then before the esc heats up too much i flip into IU2 and beat it to death for a bit, then finally i flip into IU3 and pretty much always finish with a high speed run, bring it in and hit the wall per say and flip throttle hold and do a mini auto to land. So throughout the flight the helicopter see’s many different aspects of flight and when i tune i make sure it will respond well through all of it. That was much harder to accomplish with BeastX i will say, had some head wobbles and tail issues i could never tune out at low headspeed, not the case with the Spirit Pro’s im using now. Other than the FBL, there were no changes to the helis, just a new unit with the same e headspeeds I was running before. I guess the point of all this is clearly Spirit has progressed further than BeastX when it comes to their firmware, as they both use accelerometers and gyros and wuite possibly the same sensors, it all comes down to firmware and its implementation.
As we have talked about, i understand the market was designed and built on hype, but at this point there are some companies out there, small companies that dont hit you in the wallett for all your worth selling you $30.00 worth of sensors for $450.00 and giving you little in return.
I went with Spirit because i have active dialogue with the person who designed and built it. He does it because he loves helis, not to make millions and has no big ads on RCgroups and in Model Avation like Spektrum and others do. Tomas listens to the people using his product and has made many changes in direct response to what people asked for. In any case they have some kind of advanced signal processing that totally filters out vibrations incredibly well and turns even the worst setup, heavily vibrating gasser helicopter into something that flies great. I know from first hand experience using many other FBL’s that alot of the hyped up earlier fbls like Ikon1, beastx dont hold a candle to this guys pet project.
Whatever filtering Spirit is using would be helpful for Pixhawk I am sure. It works, i know from firsthand experience and testing. As i stated above, i have helicopters that are great at low speed flight as well as speed runs without making any changes. No pitch ups, no wobbles, no gain issues. Just great, stable flight. So using that logic, im positive Pixhawk can perform the same way, it just need the right filters and settings. As ive noted before, other FBL systems have some steps involved, like teaching geometry etc that I think are likely key to their success and the newer units have signal processing that is proving very reliable. Those are the things that would totally redefine the way FBL works with Pixhawk and improve performance tenfold i bet.
Then again, pretty lofty hopes for my opinion though.
I understand you position with FBL vs flybar. Its what you kbow and have had great results with. My position is that with some of the newer hardware, i have great flight and reliability along with mechanical simplicity and longevity. I have servos in some of my 3D helis that have endured hundreds of flights and are still going strong with no slop so it is possible to have a reliable FBL setup.

(Chris Olson) #651

No. I am a commerical pilot and my insurance company is Falcon. I have a general liability and comprehensive policy for my business and three airplanes, and the UAV’s are listed aircraft on the policy.

(Bill Geyer) #652

What airplanes do you own?

(Chris Olson) #653

Two AT-802’s and a Seneca III.

(Timothy Bacon) #654

Full scale AT 802 and Piper Seneca? When you say commercial pilot do you mean UAS part 107 or full scale commercial pilot, or both?

(Chris Olson) #655

Yes, full scale. And yes, commercial pilot, Part 61 for the past 30 years. I fly the UAV’s on a Section 333. Never went with Part 107. The UAV’s are used to map GPS tracks for the Raven controllers in the 802’s to make sure we don’t spray something we’re not supposed to and get a big fine, and for crop scouting/NVDI imagery.

(Timothy Bacon) #656

Oh wow, thats awesome. It would be nice to pick your brain sometime about the full scale side of things. I have been more than seriously considering getting a commercial rotary wing liscence. I love my day job for alot of reasons, but gunning sites is getting to be a tough gig, people just don’t want to work anymore and it makes keeping the bottom line all but impossible these days. Too much stress for little reward.
In any case the mrs has been pushing pretty hard for me to change fields and get out of the 60 hour a week grind. Ive heard it can be a tough gig being a commercial pilot though. Ive honestly never completely considered what job opportunities are out there, i have just wanted my pilots liscence for quite a few years.
Its also interesting you do both full scale and UAS work as I already do that side of things as side work do theres that paralell as well.
Very cool indeed Chris!

(Chris Olson) #657

Well, never got my rotary wing rating. Took 10.8 hours instruction in a Bell 47 back in the day. Got every rating known to aviation in fixed wing, and type ratings in several heavy turbine powered aircraft. But after 10 hours in that shaking vibrating 47, I’d had enough of that. I did solo that thing though but I can’t say it was the highlight of my career or anything like that.

(Timothy Bacon) #658

Since looking into all thats involved with commercial aviation, ive been all but overwhelmed by the different routes/certifications there are out there. I always thought you got a pilots liscense and that was it. Im finding out theres alot more to it than that.
We have a flight school a bit north of me that offers a BS in commercial avaition, and just south is a rotaty wing school. Thought about taking some courses at the commercial school and testing the waters. Ive spent quite a bit of time in helicopters these past few years and every flight fuels the fire more. Never had my hands on the controls of one, flown a bunch im a cessna 180 float plane , if i remember correctly, though, family friend owned one and used to take me up all the time.
But i digress, back to my helicopter…
So i flew it today, brought it with me back home, we have farnland up there and couldn’t resist the big fields. So i got her up to speed, no pitch up issues, i dident have MP hooked up to the tablet so i dident catch the speed, but it was maxed out with the available bank angle in stabilize.
It is a tad squirrely when putting on the brakes at the end of a run. It just doesent stop very cleanly if that makes sense. I also noticed that if im in a hover and i bank hard forward elevator and right aileron, then bring it back level hard it has a quick head wobble. Doesent happen with any other inputs, just those. If i pull straight back and then flattem the disk hard it has a tiny bob in the tail, but no big deal, and any other elevator or aileron inputs done on a sharp manner look good.
It seemed to track quite straight and as I said fast speed runs looked pretty good. No porpoising at speed either. I do have to flatten the collective a bit more though, and it still feels very snappy and sharp for a UAS helicopter. Still very 3D like.
I intend on trying the notch filter and seeing how that all shakes out. Btw, i reverted back to the setting i had before i messed with the ILMI and it wasent really a big difference in the handling, just seemed to lose a couple problems here, and gain a couple there. Its not flying terrible by any stretch, but still have a ways to go I think for me to feel 100% confident in it.

(Chris Olson) #659

Yeah, it makes sense, and you have to get that fixed. When you fly auto ArduCopter has a weird “jerk” at each waypoint. Or if you use spline it makes a nice smooth turn and when it hits the actual spline waypoint it “jerks” and levels the heli momentarily. That’s a flaw in the software for how it does waypoints. DJI’s stuff doesn’t do that and flies waypoints smooth as silk.

Anyway, if you’re flying waypoints at high speed with the horizontal accel’s cranked up and the helicopter is not stable, when it hits a waypoint and does its “jerk” thing it’ll set it off it’s not tuned right.

For the life of me, I can’t figure out why they can’t make ArduPilot fly smooth steady-speed waypoints like DJI’s software does.

(Timothy Bacon) #660

I wonder what makes the UAS jerk when hitting waypoints? You would think that smooth waypoint navigation would be a top priority? I am farmiliar with the way DJI navigates waypoints, and you are right, it is smooth in execution.
Hmm, are there any parameters that can adjust waypoint transitions? I havent looked into it closely yet, ive been focusing on the task at hand, namely basic tuning of my helicopter but compartmentalizing aside, it is still in the back of my mind.
Any jerky behavior would likely be a very bad thing the way it flies currently, I too could see i setting something off as quick inputs in the wrong manner induce some instabilites as it is, but when im on the sticks I can counter the ossicilation and avoid disaster.
I have quite a ways to go with tuning I feel. Hoping to get a chance to try Bills code today and furthermore i hope it helps target the problem.
I totally see the allure of the flybar you have, it puts problem resolution more on your mechanical setup skills and less on Pixhawk’s PID controller as its self stabilizing. But I have the cards I am dealt at the moment and am determined to make it work.
Rob L has a great flying FBL UAS using Pixhawks doesen’t he? The Procyon if I have it right? And if im not mistaken he is running LHS on that as well?
I guess what im getting at is it must be possible to make this happen, although Rob has exponentially more experience with Pixhawk and trad heli than someone like myself so that puts me at a substantial disadvantage, but one thing I am is determined.