Lessons learned by not thinking!

[KK1W]

Thanks to some pointers from Craig and Steve in chat yesterday, as well as many hours looking at log files (after learning how to halfway understand them) I solved the ';twitchies'; my quad exhibited. Unfortunately it always comes down to the obvious. Hopefully this boring rant will help someone in the future 

Quickly, I built an RCTimer DJI clone 450 (mkII) with an APM2.5 running 2.8.1. I';m new to quads but managed a few decent hovers in the garage before winter weather turned to crap. It was nice the other day so I took it outside but it was almost unflyable. Hover and flight were OK but... it would randomly ';twitch'; in almost any direction and quickly recover. Not a feeling of confidence for sure.

I went through everything, reset the PIDS to default ';just in case';, nothing made any difference.  I looked at the vid I shot of the hover and the problem sure wasn';t there then. So what changed?

The only change was experimenting with GPS and telemetry. I hooked those up and made sure they were working but removed them for the test flights thinking why break even more stuff when I crash  But, being the lazy type I only removed the units and left the cables attached to the APM. I did rubberband them to the arms so they didn';t fly into the props tho   I really should know better, I actually am pretty good about radio and electronic stuff but...

Anyway, after setting up the video camera to shoot the twitches, then breaking a prop on landing (a twitch got me), and putting the two new props on the wrong motors (caught that in time on lift off) and still nothing worked  I figured I';d take one last shot and remove the two cables.

Amazing, it actually flies nicely now  Apparently, and especially since the two cables were unterminated, they made great antennas to pick up all the ESC hash from the arms and transport it directly to the APM - just to make sure it';s brains got scrambled :embarrassed, I really should have known better.

Lesson learned, electrical interference can indeed cause un-commanded motor changes. All it cost me was a broken prop, a lot of hours of research that will be a benefit in the long run and a reminder to always think about ';what';s changed'; before looking for other reasons. Yep, I';m a m0r0n 

Now that I had a chance to fly it around the back yard today I fully understand why people put LED strips on the arms. That';s the next mod for sure, after I move the APM to the top plate and use Stef';s Trex landing gear idea to put the Lipo on the bottom. The further from the ESC';s the better I';m sure.

Thanks again guys for letting a Yank into the chatroom yesterday too 

[Gaza07]

Thanks for sharing your findings Jim I';m sure it will help some else along the line, problems like these can be a nightmare to track down   

[KK1W]

It was hard, at least for me, to tell from the logs but it appears the motor spike was what came first and not interference with the gyro or accel. The gyro curve looked pretty stable, there are spikes in the accs but you would expect that (I guess) if the motor(s) spiked and the frame twisted. If anyone is interested I can post or send the logs. I enabled RAW on Craig';s suggestion yesterday.

If it wasn';t so crappy out I would plug the cables back in but attach the gyro and radio to see if terminating them makes a difference (I bet it will help). Maybe another day, one step at a time with this thing. Here';s a short video of the problem. You have to look closely but they are there.

Quadcopter with EMI induced ';twitches';

Thanks again, this group is by far the best source I';ve found for good information and helpful folks! Good job to all involved. 

[KK1W]

Ripped the APM off the bottom plate and moved it to the top. Relocated some of the wiring. Added a ferrite ring on the ESC data lines. Found a nice Thunder Tiger heli undercarriage at the LHS for under $20 (thanks for the idea Stef!) and it looks great. Maybe a little large for the 450 but it will grow into it 

Took it for a few test spins around the back yard. Probably can use some tuning but overall pretty easy to fly and the ';twitches'; are gone!  ::) ::)

Next step add in the GPS and telemetry and see if things are still good 

RCTimer DJI V2 clone, stock motors, 30A SimonK ESC, RCTimer APM 2.5, v2.8.1, 3DR PM.

Here';s what it looks like sporting it';s new gear. The skid set is for a TT X50 Heli, part# PV0035-2.

[Gaza07]

That is a very neat install Jim it looks great you should try arducopter 2.9.1 I found it very good and very stable 

[KK1W]

That is a very neat install Jim it looks great you should try arducopter 2.9.1 I found it very good and very stable 

I';ve been thinking about that myself. I';m new to multi-rotor but I have a decent handle on 2.8.1 - at least as far as making it work with what I';ve done so far. Reading the billions of posts on the 3DR forum about 2.8.1 makes me wonder if I should add more ';unknowns'; into the mix, especially since I don';t know much already

It';s good to hear it should be a good move, maybe I';ll give it a try later this week or next. First I';ll get GPS and telemetry working with 2.8.1. Then at least I';ll know they work correctly before I move up to 2.9.1. I want to review my Z axis vibration too before making the move. I ';think'; it was OK when I was looking at the RAW logs a few days ago but didn';t spend much time with it.

Thanks Gaz, appreciate the recommendation.

[Biffa]

That looks great with the TT landing gear, do you know what part number they are?

[KK1W]

That looks great with the TT landing gear, do you know what part number they are?

Steve,

It is TT part# PV0035-2, Landing Skid Set X50. Thanks for asking about it, I should have put it in the original post (edited and added now). Cheaper everywhere on line but a good example of why having a nearby LHS is nice. It is so much easier to find things that ';could work'; when you can hold them in your hand.

[Biffa]

Many thanks

...a good example of why having a nearby LHS is nice. It is so much easier to find things that ';could work'; when you can hold them in your hand.

Sadly my nearest LHS isn';t very local so I never get the chance to hold parts and rummage, all my buying is done online.

[rickp]

How are the vibrations with the APM mounted like that?

[KK1W]

How are the vibrations with the APM mounted like that?

On the bottom plate or the top plate?

I';m running 2.8.1 so I wasn';t as concerned as I will be when I move up to 2.9.1. The Z axis looked OK when it was on the bottom plate. I didn';t pull down the latest logs after I moved the APM to the top plate. It was mounted with two large pieces of Futaba 2mm gyro tape on the bottom plate. I';m only using the tape in the corners and in part of the middle on the top plate. I';ll pull down the logs tomorrow and see what they look like, hopefully better, although I guess these are somewhat OK. The spike at the end is when I landed on the deck.

[teslahed]

Ripped the APM off the bottom plate and moved it to the top. Relocated some of the wiring. Added a ferrite ring on the ESC data lines. Found a nice Thunder Tiger heli undercarriage at the LHS for under $20 (thanks for the idea Stef!) and it looks great. Maybe a little large for the 450 but it will grow into it 

Is it ok to use one ferrite ring on all 4 ESC lines? I';ve got some ferrite rings somewhere - i wonder if i should bung one on my ESC data lines as well?

[KK1W]

Is it ok to use one ferrite ring on all 4 ESC lines? I';ve got some ferrite rings somewhere - i wonder if i should bung one on my ESC data lines as well?

I don';t any reason why it wouldn';t, unless you couldn';t get enough turns from each one through the center of the toroid. I have looped each data line through two turns, it was snug but as long as they fit it is OK. My understanding is the toroids will help eliminate ';common mode'; type interference, basically high frequency current (ESC spikes) on the outside of the conductor.

The problems I was having with my APM and twitching were not related to the ESC data lines but instead the unterminated GPS and telem cables. I had the toroids so figured it wouldn';t hurt

Not much help I';m afraid - it all comes down to magic unless you have some good test equipment to actually look at the signals on each side of the toroid.

[rickp]

On the bottom plate or the top plate?

I';m running 2.8.1 so I wasn';t as concerned as I will be when I move up to 2.9.1. The Z axis looked OK when it was on the bottom plate. I didn';t pull down the latest logs after I moved the APM to the top plate. It was mounted with two large pieces of Futaba 2mm gyro tape on the bottom plate. I';m only using the tape in the corners and in part of the middle on the top plate. I';ll pull down the logs tomorrow and see what they look like, hopefully better, although I guess these are somewhat OK. The spike at the end is when I landed on the deck.

I';d say that those were right on the boarderline of acceptable. The suggested acceptable range is -5 to -15 on z, and if you can do better, you';ll probably see better flight performance. Its also worth setting the filter frequency to 20HZ, as that seems likel to be the default in the next release - so that may improve your performance anyway.