I prepared a Word document to insert or paste here, but I cannot figure out how to do it without losing images and formatting. I will attempt to attach it as a Word file. Wish me luck...Arducopter%20vibes.docx
Vince the APM generally updates at 50 hertz (50 times a second) and that is both sensor input and control output and the gyros are dealt with at a very fine resolution and you need this for smooth operation of the Copter.
Emile, just save your image to your desktop as a JPEG straight from the Terminal mode of the Mission Planner by right clicking after you have windowed what you want then insert your images straight on the BLOG entry using the little image import feature. Just to the right of the little Link Icon at the top of the comment form. Illustration just done below.
Hi Vince, fair enough, the long and thin 1/16" O-rings represent a long elastic path and the sizeable plywood base plate represents a significant mass.
If the Airframe is stable in calm wind and hovering, probably everything is fine, but if there are gusts or if it is desired to make sudden maneuvering actions the airframe will move and the high mass platform will fight the elastic support bands and tend not to for a bit.
Certainly it may not exceed a half a degree, but when it comes to flight and automatic stabilization in particular a half a degree is a lot and it can definitely result in "mushy" responsiveness.
I have used O-ring suspension on both of my FlameWheels for some time now and initially I had a fair amount of corner clearance from the standoofs and was using fairly long O-rings.
I started out with some that were short enough to be fairly tight. The problem was they didn't absorb vibrations as well as I hoped so I put in some slightly longer ones, they absorbed the vibration better but definitely introduced some mushiness from the flight control board not quite following the airframe.
And as you pointed out slowly dampening oscillations can and for me have resulted as well.
Finally I moved the standoffs much closer to the APMs corners and used relatively loose tension, and made it down to less than 1/10 G vibration and since the edge was so close no detectable lag.
On the board above, the cross section of the O-rings is very small in comparison to the extremely long length of them. The board is also quite high in mass and I believe the board will definitely wallow well behind actual airframe motions.
The easy fix is to move the O-ring holes on the board further out to the corner and to use bigger cross section and shorter O-rings.
If you mounted this on non-dampening springs or elastic bands, you would probably get an increasing oscillation till the corners of the board were bottoming out.
Gary, you often raise concerns over introducing lag into the system. I know you are very knowledgeable on this whole vibration issue, but I am having difficulties picturing how the above arrangement could introduce significant lag.
Even in the most violent, control induced acceleration, I doubt if you could displace the APM more than ½ mm or ½ degree, and probably way less than that.
In my simple understanding, such lag would be extremely small compared to the normal time for the APM to command a movement, send more power to a motor, accelerate the propeller, move the frame.
My own anti-vibration system has spade loads of flex but fly’s beautifully in all modes and the nav-roll/roll graphs follow very closely. The tuning rates are now up quite high but it was still flying nicely at half the values.
Would mushiness be a symptom of a copter suffering with significant lag? My amateur understanding would suggest the controller would try harder or longer to get a response which may lead to over-control, oscillations or inability to tune high rates (ok, therefore mushiness, I get it now).
What I have had issues with, on an overly soft APM mount was (dont know the right techno word) harmonic vibrations. I guess if you ‘twanged’ the above APM mount to find its frequency and this just happened to coincide with a vibrating motor frequency it would be a problem.
Looked through you Word Document, unfortunately the low quality of the text in most of your accel graphs made interpretation difficult.
But I think you might be doing 10 times better than you think you are.
First off, every whole number increment is approximately 1/10 G.
10 = 1 G, 1 = 1/10 G, 0.1 = 1/100 G.
The clear bottom PNG file that you put in seems to indicate about 1/10 G total vibration (+ and -).
The reason the other images seem to vary so much is that the graph is normalized or you windowed to a smaller or larger vertical scale.
1/10 G is excellent.
However, your long thin O-rings and fairly heavy central movable plate might allow significant delay between when your airframe actually moves and when that motion is translated to the flight control board meaning that the sensors wont see a genuine movement of the airframe until well after it has happened.
This is undesireable and will cause mushiness or slow response.
If this is the case, you might want to try upping your O-rings to 3/32 cross section and maybe increasing tension slightly (smaller diameter).
And you could even consider putting some small low density pieces of foam between the intermediate board and the top corners of your FlameWheel frame just to snub the actual motion of the board a bit.
The use of the caulk is excellent and you have used just the right amount to deal with the small mass of the APM.
I am using 1/2" diameter 30 durometer Sorbothane hemispherical bumps to accomplish the same thing.
From what I can see you certainly don't need to improve your vibration damping anymore, but excessive long coupled intermediate board movement delay is something you may need to address.
Comments
Vince the APM generally updates at 50 hertz (50 times a second) and that is both sensor input and control output and the gyros are dealt with at a very fine resolution and you need this for smooth operation of the Copter.
Emile, just save your image to your desktop as a JPEG straight from the Terminal mode of the Mission Planner by right clicking after you have windowed what you want then insert your images straight on the BLOG entry using the little image import feature. Just to the right of the little Link Icon at the top of the comment form. Illustration just done below.
Hi Vince, fair enough, the long and thin 1/16" O-rings represent a long elastic path and the sizeable plywood base plate represents a significant mass.
If the Airframe is stable in calm wind and hovering, probably everything is fine, but if there are gusts or if it is desired to make sudden maneuvering actions the airframe will move and the high mass platform will fight the elastic support bands and tend not to for a bit.
Certainly it may not exceed a half a degree, but when it comes to flight and automatic stabilization in particular a half a degree is a lot and it can definitely result in "mushy" responsiveness.
I have used O-ring suspension on both of my FlameWheels for some time now and initially I had a fair amount of corner clearance from the standoofs and was using fairly long O-rings.
I started out with some that were short enough to be fairly tight. The problem was they didn't absorb vibrations as well as I hoped so I put in some slightly longer ones, they absorbed the vibration better but definitely introduced some mushiness from the flight control board not quite following the airframe.
And as you pointed out slowly dampening oscillations can and for me have resulted as well.
Finally I moved the standoffs much closer to the APMs corners and used relatively loose tension, and made it down to less than 1/10 G vibration and since the edge was so close no detectable lag.
On the board above, the cross section of the O-rings is very small in comparison to the extremely long length of them. The board is also quite high in mass and I believe the board will definitely wallow well behind actual airframe motions.
The easy fix is to move the O-ring holes on the board further out to the corner and to use bigger cross section and shorter O-rings.
If you mounted this on non-dampening springs or elastic bands, you would probably get an increasing oscillation till the corners of the board were bottoming out.
My thanks to all for the comments.
O-rings were on hand. Will experiment with others,
DAP caulk is water based so no corrosion expected. But I will keep an eye on it.
I did struggle with the images trying to keep them to the same vertical scale.
Does anyone know of a better way to do this?
Gary, you often raise concerns over introducing lag into the system. I know you are very knowledgeable on this whole vibration issue, but I am having difficulties picturing how the above arrangement could introduce significant lag.
Even in the most violent, control induced acceleration, I doubt if you could displace the APM more than ½ mm or ½ degree, and probably way less than that.
In my simple understanding, such lag would be extremely small compared to the normal time for the APM to command a movement, send more power to a motor, accelerate the propeller, move the frame.
My own anti-vibration system has spade loads of flex but fly’s beautifully in all modes and the nav-roll/roll graphs follow very closely. The tuning rates are now up quite high but it was still flying nicely at half the values.
Would mushiness be a symptom of a copter suffering with significant lag? My amateur understanding would suggest the controller would try harder or longer to get a response which may lead to over-control, oscillations or inability to tune high rates (ok, therefore mushiness, I get it now).
What I have had issues with, on an overly soft APM mount was (dont know the right techno word) harmonic vibrations. I guess if you ‘twanged’ the above APM mount to find its frequency and this just happened to coincide with a vibrating motor frequency it would be a problem.
Hi Emile,
Looked through you Word Document, unfortunately the low quality of the text in most of your accel graphs made interpretation difficult.
But I think you might be doing 10 times better than you think you are.
First off, every whole number increment is approximately 1/10 G.
10 = 1 G, 1 = 1/10 G, 0.1 = 1/100 G.
The clear bottom PNG file that you put in seems to indicate about 1/10 G total vibration (+ and -).
The reason the other images seem to vary so much is that the graph is normalized or you windowed to a smaller or larger vertical scale.
1/10 G is excellent.
However, your long thin O-rings and fairly heavy central movable plate might allow significant delay between when your airframe actually moves and when that motion is translated to the flight control board meaning that the sensors wont see a genuine movement of the airframe until well after it has happened.
This is undesireable and will cause mushiness or slow response.
If this is the case, you might want to try upping your O-rings to 3/32 cross section and maybe increasing tension slightly (smaller diameter).
And you could even consider putting some small low density pieces of foam between the intermediate board and the top corners of your FlameWheel frame just to snub the actual motion of the board a bit.
The use of the caulk is excellent and you have used just the right amount to deal with the small mass of the APM.
I am using 1/2" diameter 30 durometer Sorbothane hemispherical bumps to accomplish the same thing.
From what I can see you certainly don't need to improve your vibration damping anymore, but excessive long coupled intermediate board movement delay is something you may need to address.
Most definately removed the DAP asap or you will have issue with that board.