[djwicker]

After 4 attempts, I am finally able to get my quad in the air, but it still seems twitchy. After the experience, I am really asking are my expectations out of line? I bought a knockoff zmr250 kit expecting to get a basic quad in the air without too much trouble and yet be able to expand it from there. My experience has been anything but easy; the first issue I ran into was the bug in 16.09 that prevents you from setting up the TX without having a flight mode switch. Of course I did not know this was the problem when I started with a 4-channel radio; so I bought a new 6 channel Flysky FS-I6X, which had the same problem! I was finally able to discover the bug, and after futzing with the radio was able to set up a switch for the flightmode (note that the TX defaults with this switches disabled, so every novice user will re-discover this problem!). In the process, I was able to burn up a flight controller when it went crazy on me, still don't know what happened there. So on my second flight controller and 3rd set of props (thankfully I had ordered spares and used them instead of my original carbon fiber ones!) the stupid quad still flips when I try to take off. Back to searching the message board and I find the suggestion to set ZeroTheIntegral. Success! The quad leaves the ground and starts to drift into the house, then down it comes minus all the props and two of the prop protectors I had just installed! So much for prop protection!

So my first question is should I have expected the default behavior to be a very conservative flight mode like the $30 quad-in-a-box that flies just fine? It seemed like a reasonable expectation, but clearly I was wrong! And what do you suggest as a next step to get it flying safely? I have used all the defaults except for the ZeroTheIntegral thing. I just want simple, stable flight, nothing fancy for now.

[hamilc]

I am a newbie but here are some suggestions. What controller are you using??? Did you go through all of the vehicle setup steps included in LP (LibrePilot)? I started with a Turnigy transmitter and a CC3D. I made sure the motor directions were correct and without props I made sure that throttle control worked as expected. By this I mean as you gradually increase throttle, do the motors all start at the (roughly) same time and with equal speed and direction? Then with no props, do the pitch and elevator controls work? You can tell by which motors speed up as you move the controls. Seems like if the controls are working properly you should be able to lift off.

I actually tied a string, about 1 foot long, to the bottom of my quad, connected it to the ground. Then with the props, I can lift the quad off the ground gradually and the string will prevent runaway and save props. Once off the ground and tethered by the short string, I can test the pitch, yaw, and elevator controls. 

Some of these suggestions might help.

I would not recommend changing the parameters in the digital loops untiul you know how to do so.

[TheOtherCliff]

It's like buying a car vs. building a car.  The car you buy is already aligned, but the one you build yourself will need to be aligned to drive straight, and the suspension will need to be tuned... etc...

If you follow the instructions, you should get something that flies well, but it will drift on first flight and even with trimming, some drift will remain.

[djwicker]

hamilc; Thanks for the suggestions. I have a Chinese cc3d. I went through all the vehicle setup without problems (many, many times!). The TX setup was the only one that gave me a problem, but eventually it worked fine. I do have a problem with the output checks in that one (or sometimes two) of the props do not respond at all. However, if I go back through the vehicle setup, I can control all four just fine. Also, in the flight data tab, several times the controls would just keep doing a slow roll even though the vehicle was resting on the ground. Once it did a 360, it seemed to stabilize again - not good if this happens in real flight! Also when I power up the drone (no props) the speed of the individual motors seems to vary each time I rev the motors. One time one motor was really screaming! So I was able to eventually get it off the ground, but I will try the string idea next time.

TheOtherCliff; Well, I did follow instructions, but it didn't work until I found the Magic Word! And I didn't build a car from scratch, I put together a kitcar that should work once it is all together. Or at least that is what I thought I was doing.

So out of curiosity, it seems like the instrumentation should allow for correction of most of the drift; if the stick is near the middle, just keep the drone at zero acceleration. I should think this would be part of the controller function.

[TheOtherCliff]

In the first flight, the FC correctly sets itself level, but the quad underneath it is off by 3 degrees and it has absolutely no way of knowing this (without a GPS). 

The reason factory built quads are better is that if the FC is mounted 3 degrees off, the factory configures it to adjust for this ... like a factory built car already gets it's front end aligned as a step in the assembly process, where your kit car needs to be aligned by you after building it. 

The controller doesn't know that it is mounted with say 3 degrees of bank and what it thinks (without being told otherwise by setting RotateVirtual, or the leveling calibration on the Attitude page, two alignment procedures that fix this) is 1.0000G downward is actually 0.0523G of lateral acceleration (after 3 seconds it is already 3 meters away and getting faster and faster) plus (in the ortho vector sense) 0.9986G of down acceleration which adds up to 1.0000G.  (really! )

[djwicker]

ok, makes sense. But doesn't the calibration in the vehicle setup compensate for this? And as you say, the leveling calibration on the Attitude page compensates for this. But I did both and still get drift. However, I did notice (and realize) that how you mount the FC seems critical; I mounted with velco (as suggested by someone online) but noticed this makes the FC wobbly, and probably not a good thing. I will correct this by mounting it with screws instead. Maybe this should be pointed out somewhere for first-timers as I did not see this mentioned anywhere. Maybe this will solve my drift problem?

One other thought I had. I was kind of annoyed that after running the vehicle wizard I still had to compensate for the mounting orientation of the FC. This also was not obvious to me as a novice. Why is this not just another step in the wizard? This would probably save a lot of people the headache of finding out this is yet another thing to go wrong that you didn't know could go wrong.

[Mateusz]

This would probably save a lot of people the headache of finding out this is yet another thing to go wrong that you didn't know could go wrong.

I agree about mounting I use nylon m3 most of time but also got these https://www.banggood.com/4-PCS-Realacc-M376-Flight-Controller-Fixed-Screws-for-FPV-Racer-p-1135527.html
Having as little vibration as possible is important and it wont fly good with flight-controller is moving.

Regarding wizard step for mounting orientation, I am not sure this is necessary, every FC has an arrow pointing in direction and usually FC and motors are numbered, its natural to see that if you flip FC motor 1 is not 1 anymore. I agree it might be confusing for someone new, so devs developing wizard could consider adding it, but it used to be like this since many years and is not end of the world if they focus on something more important

[TheOtherCliff]

The leveling calibration does the best it can, but some aircraft aren't level when they hover.  An example of this is that if the motors aren't vertical by 3 degrees, the quad will hover 3 degrees from level.

An extreme example is "race mount" motors that all lean forward.  During hover, the motors are vertical, so the body is sloped backwards.  The quad should be put on blocks so that the motors are vertical for the leveling calibration.

[djwicker]

Mateusz;

I disagree about the mounting orientation; at least on the Chinese versions, the arrow is very difficult to see; you almost need to know it is there to see it. And the natural way to mount it is so the USB connector is accessible, which is a 90 degree rotation. Also, when you run the wizard, the NW motor is still the NW motor, so there is no indication that anything is wrong. I would be willing to bet that if you took a poll, a significant number of first-time users did not catch this without something bad happening first! Whether it is worth adding to the wizard depends on the priorities of the project. Is it more important to add advanced features for a few users, or to create a quality experience for new users so they stick with the project and say good, rather than bad, things about their experience? Seems like a small effort for a potentially big payout to me.

[Mateusz]

...
Seems like a small effort for a potentially big payout to me.

Building your own aircraft has always been a project in itself, it requires some skills, knowledge and learning. Of course I agree about that it's a nice goal to have "out of box" solution that is generic enough to work for all builds, but sadly not very realistic. It's not a product that is constant, and what is more important project is not driven by number of customers, notion of number of customers is alien to GPL software. There is no "payout", even support on forum is voluntary.
I think people contributing to this project use it daily and tailor it to their own needs, improve it in their spare time, those are professionals (academia, companies, private) who spend some of their spare time on hobby like this, because they are satisfied using this software.

If you think some feature is necessary try opening ticket in JIRA, describing it, maybe make some drawings how you it should look like. Someone might be able to implement it at some point, but JIRA ticket is a good start I think.

[djwicker]

Thanks for your perspective. I did check out a few other of the open source controllers and they appear to take a similar approach, but they do seem to have builds that are "novice" targets that work out-of-the-box for most people. It seems like a good idea, in that it allows customization and new features while satisfying the requirements of a novice or someone who doesn't want to make a career out of his build, but just wants some hands-on experience.

So what is JIRA? I searched for it and found nothing referencing it except for this!

[Mateusz]

So what is JIRA? I searched for it and found nothing referencing it except for this!

Very first hit on both Google and DuckDuckGo search engines with "LibrePilot Jira" finds this https://librepilot.atlassian.net/secure/Dashboard.jspa

...they do seem to have builds that are "novice" targets that work out-of-the-box for most people. It seems like a good idea, in that it allows customization and new features while satisfying the requirements of a novice or someone who doesn't want to make a career out of his build, but just wants some hands-on experience.

Please share which software and how they solve your problem. I don't think any software around asks user to fix their board orientation, but please prove me wrong as I am curious how they deal with it, would be good perhaps for devs to know.