Plea for gimbal tuning wisdom....

[Motopreserve]

Hey folks. Please excuse the long-winded post. Trying to be as clear as possible.

I've been toiling away with a used 2-axis gimbal I picked up for practicing with a Nex5n. Most of the controllers I've used until now have been proprietary (zenmuse and tarot for gopro).

This new gimbal make is unknown to me, but I know the controller is an 'authentic' Martinez V3 (Ready to Fly Quads version) based on ID markings on the board. It is loaded with the latest firmware (V50 r217).

I have spent some time balancing the camera/gimbal based on videos and how-to's, and as far as I can tell it's pretty good. I can move it to any position and have it stay. I think I have the wires routed well (zip ties to avoid them pulling/influencing motors - but leaving enough length for articulation). The motor directions have been checked and appear correct. The IMU is mounted in the correct, standard position - parallel to the camera/plate.

my biggest problem has been a good method of basic tuning. From what I've read, it's best to start with PWM at zero and PIDs all at 5 (both pitch and roll), then starting with Roll only: slowly raise PWM until "vibration." Then back off until it stops. Then begin to raise P until vibes, raise D to calm it, back and forth until P is as high as it can get. I is still a mystery to me and makes the gimbal shake horribly any time I raise it.

Couple issues/questions:

1. there is no clarification to what "vibrations" means. I get very low oscillation (bouncing really) when I raise PWM. Is this different than vibration? I was expecting high freq vibes. I have experienced high freq vibes when the D gets too high. So I know there is a difference. Is one caused by specific issues???

2. Is is there a better method to do the basic tuning? I've read one alt method to set everything to zero and pick a PWM at random. This doesn't make sense to me.

3. One issue may be that the IMU is mounted on plate under camera so that the servo connector will hit rear motor if it articulates furthest forward. I won't want it tilting that far - but not sure if this is a factor. I may have to solder new header pins to the IMU for clearance.

Any guidance will be MUCH appreciated.

Thanks

Scott

 

[kloner]

I've never seen a martinez board that i knew of that had good footage. i'd switch up to alexmos or phobotic asap....

One thing i do quite bit is start with one axis at a time, like i'll do the tilt, make roll come up leaving tilt on, then do yaw.....

 

[Motopreserve]

I've never seen a martinez board that i knew of that had good footage. i'd switch up to alexmos or phobotic asap....

One thing i do quite bit is start with one axis at a time, like i'll do the tilt, make roll come up leaving tilt on, then do yaw.....

Thanks Kloner. I was afraid of that. It seems the "good" thread on another forum doesn't seem willing to answer even the simplest questions about this board - so I was leaning toward an alexmos. Just seems there is A) more people using them and B) more success.

Appreciate the help.

 

[Motopreserve]

Quick followup:

Any opinion on the 8bit vs 32 bit board? For now I would be using just 2-axis, but might make sense to purchase the 3-axis for use in the future. But the little I've read has varying opinions on the 2 versions.

 

[PeterLester]

The 32 bit 3 axis BGC is largely a big unwritten book yet to be written! The basic controls for 3 axes is in place but challenges remain with yaw axis follow mode. However i believe the yaw functionality is mostly limited by the simple and cheap IMUs that are being used. 9 to 12 DOF imus are required for better yaw control.


Sent from my iPad using Tapatalk

 

[Motopreserve]

The 32 bit 3 axis BGC is largely a big unwritten book yet to be written! The basic controls for 3 axes is in place but challenges remain with yaw axis follow mode. However i believe the yaw functionality is mostly limited by the simple and cheap IMUs that are being used. 9 to 12 DOF imus are required for better yaw control.


Sent from my iPad using Tapatalk

Thanks peter.

I literally just just pulled the trigger on an 8 bit version. I'm not using the 3rd axis yet, so I went cheaper and will wait on the updates to take hold before moving on to 32bit and a third axis.

I figure I can wade in and learn the new style of GUI and firmware - and then place the Martinez under the tire of my truck and drive!

 

[jfro]

32 bit board with autotune. Just tried it and started a new thread. In a nutshell, I have a couple 8 bit boards and my 32 bit is better than my 8bit Alexmos boards IMO. Will be switching them all to 32 bit or maybe the new board Kloner spoke of. The smallest is $400 so since I had a good test today with the 32bit & autotune, I'm going to give it a chance and probably wait the winter to decide if the Alexmos 32bit and autotune is going to be what I need.

I don't think you will be happy with the 32bit board if you are used to the Zenmuse.

 

[Motopreserve]

32 bit board with autotune. Just tried it and started a new thread. In a nutshell, I have a couple 8 bit boards and my 32 bit is better than my 8bit Alexmos boards IMO. Will be switching them all to 32 bit or maybe the new board Kloner spoke of. The smallest is $400 so since I had a good test today with the 32bit & autotune, I'm going to give it a chance and probably wait the winter to decide if the Alexmos 32bit and autotune is going to be what I need.

I don't think you will be happy with the 32bit board if you are used to the Zenmuse.

just a few minutes too late .

I read enough to see that people are pleased with the 8 bit version. Like all of these things, they're a work in progress, and the progress has been happening so rapidly that I'm sure there are bound to be growing pains. But the 8 bit has been around long enough to provide more research material and experiences. I'll need that!

Im fine getting the 32 bit down the road when I finally get going on the third axis. For now I'll use this cheaper version to learn the ropes of this new code.

Im not married to the zenmuse. I've only used the smaller gopro version a bit. They aren't perfect either. This will be the first gimbal I'm using a slightly larger camera on (Nex5) so I'm sure it will be a challenge, regardless. I'm game though...

 

[jfro]

The 8 bit boards do work. I think I'm just pretty pumped about auto tune. Spent too much time tuning on my 8 bit boards.

 

[Motopreserve]

The 8 bit boards do work. I think I'm just pretty pumped about auto tune. Spent too much time tuning on my 8 bit boards.

That alone would have been worth it. How did it work for you?

I'm sure I'll move there at some point (newer 32bit) - but unless auto tune is flawless it will still be good for me to know the inner workings of the software.

Id have to imagine anything would be better than this Martinez nightmare I'm stuck in...

 

[PeterLester]

Conventional PID loop tuning was straight forward and well explained by Dr. Greg Shinsky. But it seemed to take a turn when applied to Multi rotor and brushless gimbals. This might be due to aerospace modifications/applications of pid loop control relative to conventional industrial process control application. In any case it is harder now to understand and apply correctly. P = proportional, I = Integral, D = Derivative. So the last two variables come from Calculus but are not applied according to conventional theory making it so much more confusing to work thru correctly.


Sent from my iPad using Tapatalk

 

[PeterLester]

Martinez worked fine for me with 2 axis and GoPro cameras. But i added the external v-divider resistors then activated v correction function to account for battery voltage drift down over time. Alexmos is simply easier all around and has more functions.


Sent from my iPad using Tapatalk

 

[Motopreserve]

Martinez worked fine for me with 2 axis and GoPro cameras. But i added the external v-divider resistors then activated v correction function to account for battery voltage drift down over time. Alexmos is simply easier all around and has more functions.


Sent from my iPad using Tapatalk

i realized late that most help online for the Martinez has been based on the much lighter gopro.

I think my power has been solid since I've been tuning only with a regulator. I set the output for 9v which was suggested. The experience has been rather bizarre. After about 20 hours, I had the roll decent - but the slightest PWM sent to the pitch motor made it bounce horribly. I have checked and rechecked the orientation of the IMU and done tests to make sure the motors were acting properly (reversed). No love whatsoever....

just looking at the alexmos GUI, it seems like it should at least be more informative and helpful for tuning.

I was was pretty comfortable with PID tuning for multiwii, but this gimbal controller has kicked my a**

 

[jfro]

Kloner has a pretty good tutorial on the threads some where and his footage is way better than mine.

I start with low numbers on the PID's and work up. I am able to get decent footage in calm conditions, but in the wind, I struggled with my light settings. I'm hoping for a little wind to try the 32bit auto tune and see if they work better than what I was getting. Maybe more holding power with higher pid's....

 

[Motopreserve]

Kloner has a pretty good tutorial on the threads some where and his footage is way better than mine.

I start with low numbers on the PID's and work up. I am able to get decent footage in calm conditions, but in the wind, I struggled with my light settings. I'm hoping for a little wind to try the 32bit auto tune and see if they work better than what I was getting. Maybe more holding power with higher pid's....

If you find that link to his tutorial, please post it here. I just searched every phrase I could think of by Kloner specifically and came up short.

thanks

 

[Motopreserve]

I still haven't found the link to Kloner's how-to, but I'd love to see if if anyone has it bookmarked.

I have switched over to AlexMos 8 bit version, and have instantly had better luck. I am really liking the GUI and features SO much more. I think my camera balance is really good (I used a tiny weight at the end because all axes were almost there). I have some micro vibrations (I know, what else is new???!!!) that I am noticing with the handheld stand I made for testing.

Couple questions:

1. Currently have 12 300g (pinkish) dampeners on the mount system. Wondering if I should go softer or harder to eliminate vibe translation to the mount (for instance: when I knock the stand or tap the gimbal mount cross beams).

2. Can anyone tell me a definitive voltage that runs these 8 bit boards the best? I am using a 3S with regulator, and have it set to 11.5v. I know it would effect my power settings, but is there a sweetspot?

Thanks!

 

[kloner]

what did i do? don't think i have a gimbal tuning vid....

looks like a cool hanger... i've found to fine tune these you want to mount it to the vibe iso it is going to use, put it on the airframe and hang that whole thing,,,, the tune is directly affected by actualy getting on the aircraft

are those 5208

 

[Motopreserve]

If you haven't made a tuning video yet - you should!

I feel like it's ready for the airframe too - but sadly I can't fit the damn thing onto the bench in the workspace I'm using in my office! So I have been using this for now - but it's about time to get it on the frame. Looking to see if I can setup an old multiwii Bluetooth module I have so I don't need to lie on the floor for tuning. I'm getting old...

The motors are iPower 4008 (roll) and 4006 (pitch). I have the power pretty low (roll is about 105 and pitch 100 or less) with the 3S at 11.5v steady.

Any thoughts on the dampener (stiffer or less stiff). I have some 200g I could slap on to test - but it's a drag to install and I'd avoid it if I'm barking up the wrong tree.

Thanks.

 

[kloner]

big things you can do there is get the pid up and going, turn up some motor power and look at how a radio reacts to it when it moves.... every motor is a little different so to just throw numbers out there isn't gonna get ya real close, but i find a pretty consistent 12-14P.... .1-.2I and 7-10D getting you going,,,,, slowly raise motor power till it shakes then back it off, watch the graph for shake.

On the vibe iso you need to fly it,,, adjust till the vibes are gone, on the hard side it induces shake when you move forward from a hover, etc , the gimbal physicaly moves around on the too soft side, and when it's just right vibes aren't an issue and the camera is steady. sometimes it's hard to get it stiff enough to not physicaly move but not have shake so people mod a softer setup to act stiffer, like nascar, add a spring shim to make the assembly stiffer, take it out to soften it up. I've seen all kinds of things work and particularly i've never had a stock iso system work perfect and it always took some sort of finaglin,,,, well except a disco.

 

[Motopreserve]

I've got it pretty close right now. Just the annoying shake or micro vibration sometimes. I'm going to try to balance out the PID with a little more power. See if that helps. I have the profile saved, so I can always go back. Seems man like everyone has this difficult period that eventually just works when you stumble on it.

I'm not against messing with thing for hours on end - it's just the lack of direction that is out there (example: start with pitch or roll? Start with P & I or P & D? Etc).

Ill wait on getting too crazy until I have it on the frame.

Quick clarification: when you talk "shake," you mean the bobbling not the fine (high freq) vibrations you can feel if touching the motor, correct?

also, what voltage are you all running?

Appreciate the help.