[bennyh]

Hi all, I need expert electronics advice please. Posting in this section for general electronics advice, not necessarily specific to the aircraft.

Project: Cinestar hexacopter - based on Mikrocopter flight controller. Camera gimble fitted.
http://www.quadrocopter.us/
Problem: persistent glitching in the roll axis

My buddy built the ship and has been flying it for a few months. Taking HD footage is the plan but there is a consistent twitch in one axis. This happens in dead air, on the hover or on the move. It's just enough to spoil the the video recording, but not terribly noticeable in general flight. We've played extensively with the PID settings to try and iron it out. The props are balanced. The frequency of the glitch is about every 5 seconds, but not timed it for consistency.

I suspect an ESC or motor problem. Is there any way of testing the consistency of ESCs? I know how to use a multimeter but that's about it. Perhaps an ocilloscope is needed? Likewise, could it be a dicky motor?

Thanks for any comments.
Bennyh

[PDR]

To start with the obvious - if it's one one servo that's glitching have you replaced that servo?

PDR

[bennyh]

No servos!
1 flight controller board
6 ESC
6 motors

[dogshome]

No dodgy connections to the motors? Battery up to the job?

[bennyh]

I think poor connections unlikely as all the motors appear to work perfectly. Battery is a 4s 8000mAh brick. All the other batteries exhibit same problem. Dead battery would make this sucker go down rapidly!

[Yoyo]

No servos!
1 flight controller board
6 ESC
6 motors

Can you swap the motor/ESC pair from one arm to another easily? If the twitch moves to another axis then it's one of them, if it stays put then it's controller related.

[bennyh]

interesting idea there Paul, but they are all on one circular board. I could unsolder all the connections and rotate 90 degrees though - bit of a ball ache but might be worth it.

[JohnB]

We've played extensively with the PID settings to try and iron it out.

If youre meaning the settings of a full 3 term controller the Derivative part needs to be switched off, you only need PI control.

J

[Yoyo]

interesting idea there Paul, but they are all on one circular board. I could unsolder all the connections and rotate 90 degrees though - bit of a ball ache but might be worth it.

I followed the link you put in the first post - that's a serious and expensive bit of kit. I'd definitely be taking it up with them instead of trying to guess what's up.

If it isn't an ESC/motor fault then the next most likely thing is a software problem, and I'd suspect that to be the case anyway from the long period of the glitch - in ESCs everything is happening hundreds of times a second, not every few seconds, so I'd expect problems to be more frequent. Software control loops, counters etc. might glitch less often than that.

[bennyh]

Yes, we're speaking to the producers too.
Thanks for your comment about the likely source of the problem though.
For the record, how would you test an ESC?

[Yoyo]

Yes, we're speaking to the producers too.
Thanks for your comment about the likely source of the problem though.
For the record, how would you test an ESC?

It's tricky to do it in fine detail unless you have a multi-channel storage scope to catch the inputs and outputs for an extended period.

Presumably if this glitch is enough to affect filming then it will be visible in a motor running on the bench? For a confidence test I'd probably use a servo tester to feed the throttle signal into the ESC and listen to the motor at a range of speeds, any glitch should show up in the sound. That wouldn't show if the problem is down to sudden rapid changes in control input such as you're likely to find in a hexacopter, though.

[dickw]

If you could beg, borrow, or even buy an on board data logger you could monitor the controller to ESC signal and motor parameters like rpm at the same time - unfortunately only one motor/ESC combination at a time though.
The SM Unilog that I use logs at up to 20 Hz, not sure about the Eagle Tree logger - probably 10 Hz.
English translation of SM site here:-
http://translate.google.com/translate?u=http%3A%2F%2Fwww.sm-modellbau.de%2Fshop%2Findex.php%3FcPath%3D11&hl=en&ie=UTF-8&sl=de&tl=en

Dick

[bennyh]

I think there might already be an onboard data logger. I'll look into that. Cheers
Ben

[dogshome]

If youre meaning the settings of a full 3 term controller the Derivative part needs to be switched off, you only need PI control.

J

The much maligned Derivative term. But don't ever think it isn't extremely useful. Never be without it   

How's yourself J? About time we talked to each other at the next nats I think 

[JohnB]

The much maligned Derivative term. But don't ever think it isn't extremely useful. Never be without it   

I agree but it has no place in this environment and could well cause the kick that is being talked about, maybe falling on deaf ears though

How's yourself J? About time we talked to each other at the next nats I think 

Always a pleasure and youre more than welcome anytime to drop into RMFC.

J

[bennyh]

The ears are not deaf, but the bit in between will need to do more reading to understand what you mean!

[JohnB]

The ears are not deaf, but the bit in between will need to do more reading to understand what you mean!

No problem, if you have access to the PID parameters the 'D' part needs disabling or set to zero.

J

[bennyh]

Thank you, I'll definitely try this.
Have also found some slop in the geartrain of the servo on the camera gimble.