I just wanted to make a post to specifically address the issue of vibrations on helicopters, as I am seeing quite a few logs which exhibit excessive vibrations. This not only affects Loiter, but can also affect Stabilize as the AHRS system loses track of which way is "down". This can result in you actually having to fight the machine to keep it level.
Now, I comment so often on this, I begin to wonder if people are starting to think "he just says that about everything". So I want to provide an example of what I consider low vibration. The attached image is from my Trex 500. There's really not a lot special about the machine. It's just a good quality, solid build, with attention to detail and balancing. It started out as an HK500FBL TT, but now is almost entirely Align, as well as some other upgrades, but nothing fancy. The vibration damping system is... simply a Pixhawk, mounted sideways on the frame using the 5mm thick black foamy double-sided pads that 3DR supply with the Pixhawk. That's it. I find this 3M mounting tape to be extremely effective, and dead-easy to use. I've started using it in place of my other elaborate systems, since it pretty much just works.
I'm actually amazed when I look at this graph. It took me a long time to achieve something like this on my other helis. This one I got it right pretty early, using the same principles I learned on the others.
Compare that to an example from a machine with high vibration:
These are the signals coming from the accelerometers on the APM or Pixhawk. As you can see, my log provides a clean signal which the AHRS can use to keep track of which way is "down". There is a bit of vibration (the small lumpiness) but you can clearly see the motions of the airframe (the larger waves). I was flying aggressively in this log which is why you see big motions.
When comparing to the example with high vibration, it's pretty hard to see where the airframe motions are because they are lost in the "noise" from the vibration. The system tries it's best to extract a signal from this, but it's not great.
Now, this is not to say that everybody needs to achieve what I have. But the X and Y axis need to be lower than +/-2, and the Z-axis should be lower than 5-15. In the example above, the Z axis is actually not bad at all.
So how do you reduce vibrations?
- Make sure all shafts are straight. Check them any time blades touch the ground. All "clone" shafts come pre-bent.
- Make sure all bearings run smooth. Feeling them by hand, they should be... smooth. If they are notchy, they cause vibration. Most clone bearings, and even some genuine Align bearings are notchy right from the package. Again, these need to be checked anytime blades touch the ground.
- I find 3 main-shaft bearing supports help a lot. Most machines don't come like that anymore, I retrofit them into everything. All deep-groove radial bearings have radial free-play. Even the best quality ABEC-7 bearings do, they have to for assembly purposes. Therefore it's impossible to have a stable main shaft that is only supported by 2 bearings. 3-bearing systems are usually "over-constrained", meaning the shaft will be snug going in. This eliminates the radial free-play.
- Blades must be statically balanced in a balancer. Check them! Many blades do not come balanced. I also do a dynamic balance. What this means is that I use tape to make sure that the blades not only balance on a balancer, but are also *exactly* the same weight measured on a scale with 0.01 Gram accuracy. It happens often that the heavy blade goes up on a balancer, so you add weight to it. Now it's even heavier. So it balances on a balancer, but soon as you spin it up, there's an imbalance because it's CG must be further inboard than the light blade.
- Blades must be tracked. I exclusively use turnbuckles for the pitch links, with Left-hand threads on one end, and Right-Hand on the other. Heli Option is a nice source for these. They allow infinitely precise adjustments.
- Main gears, make sure they run straight and true and don't wobble excessively. Most clone gears come pre-bent.
- Make sure the frame is stiff. Obviously, make sure all your screws are good and tight. Use Loctite. But more than that, I find most frame designs lacking in stiffness. I try to use after-market frame sides that are thicker than normal. I also try to replace all the plastic pieces with CNC cut aluminum. So, the frame bottom, battery tray, bearing blocks, motor mount.. everything, even the tail boom mount if I can find one. All of this increases the weight of the heli. But they vibrate so much less, it's well worth the effort in a UAV application, particularly if trying to do any imaging.
- Head setup. The head should be well set up, the pitch links have to move freely. But with minimal slop. Any stickiness in the head will result in vibrations.
- Tail system. Again, this all needs to be well set up, good bearings, shafts straight, blades balanced. Tail drivetrains are particularly problematic. I always use at least two Torque-tube bearing supports. If the kit only uses 1, add another. Make sure the tail rotor shaft is straight!
I used to use a lot of clone kits, but it's just gotten to the point where I spent more time trying to solve problems with them that I don't bother anymore. You constantly have to check parts for straightness, binding, manufacturing mistakes, etc. I do use a lot of upgrade parts, such as from KDE, Heli Option, and Tarot. Some clones such as Tarot and ALZRC might be OK. But any no-name kits, and most of the stuff from Hobby King are just not acceptable quality.
Would it be possible for someone to check attached flight log and let me know if the vibration levels are acceptable? I am slowly getting my JR E6 heli setup with a Pixhawk and would like to make sure that I am on the right track as far as vibration levels go.
When Rob plots his waveforms they seem to have a much higher sampling rate than mine do. Is there a way of setting the Pixhawk data sampling rate? It would be interesting to see the vibration in the same details so it would be possible to see where they are coming from.
The heli needs quite a bit of trim to hold position in stabilize mode. Alt hold works well as far as I can tell. When I switch to loiter the heli is not well behaved at all. The best that I have had it fly in loiter was a fairly small toilet bowl effect without it going out of control. Perhaps this is related to the fact that I need trim and have not setup the swash quite right, and I will check it out later tonight.
Thanks in advance for the help.
Richard, Your vibrations are not bad. The y axis is the worst one but I think it's acceptable. The flash log you attached was for a flight in stabilize and alt hold only. I did not see the mode switched to loiter on that flight. I did notice that your roll / desired roll and pitch / desired pitch graphing was not matching up well. I don't know if this is related to the swash not being leveled properly or the PID's not being tuned well. I looked briefly at your parameters and they seemed to be in the same range as I use on my 600. Use a swash leveler and get the swash plate set up properly and try again.
David R. Boulanger
Richard, You also may want to fly a little higher to get away from the ground effect air during your testing. I know that when I loiter 2 meters high the aircraft moves around a little compared to flying at 10 meters.
David R. Boulanger
Thanks for the input. Tonight I spent quite a bit of time getting the swash plate level. The heli now flies without any trim in stabilize mode. The heli flies better now. It will fly hands off for a good amount of time before I need to apply correction. I flew it higher and in more modes to get as much data log as I could for analysis. Loiter still ends up in an ever increasing toilet bowl.
Checking my log, the roll/desired roll etc are not matching up yet. Maybe now I should go back and adjust PIDs again and see if I can get it to match up better. I suppose if these are tighter then I will get it to actually loiter. Correct?
Richard, Can you post your parameter file? I was looking at your latest flash log and it is interesting when you put the heli in loiter at about 13 minutes into the flight. The des roll and pitch vs. the actual roll and pitch are right on top of each other. Is your LOITER LAT and LON IMAX values 400 or 4000? If they are only 400 this could be the problem.
David R. Boulanger
Richard, I am attaching my parameters and a flash log of my last short mapping mission. I bet if you turn up your LOITER LAT and LON IMAX to 4500 you will solve your problem. I also noticed a few of your parameters that were still 0. YAW FF, and a couple of ATC parameters I think.
David R. Boulanger
3.2 march 20.param
You were correct with the recommendation to increase the values as indicated. I was able to successfully loiter the heli for a good amount of time! The loiter position was nice within a fairly tight space. I was very impressed to see how well it actually worked. One really strange thing is that right at the end of the flight when I switched into stabilize the heli will not even hover nicely as it has quite the Y axis (roll) oscillation now.
I tried with a fully charged pack on subsequent flights with the roll being problematic. It is almost as if something drastic changed to cause it not to behave as well as it did earlier in the week. I checked the heli and can't find any issues.
Thanks for uploading your flight information. You have very low vibration levels in your helicopter. Do you have a special mount for the Pixhawk, or are you just using the Pixhawk supplied foam tape? I am going to work on reducing my levels to see if I can get it lower. I have removed rotor heads etc and started logging vibration levels of the bare machine and working up to having the blades back on. I would like to find out where vibration is coming from and trying to minimize it.
I tried using the dubro foam, but it causes the heli to oscillate in the Y axis. It is very soft tape. I have some nice JR foam tape that they use on the Ninja quad controller that I am using on the Pixhawk. It is about 3mm thick and is a bit different than Pixhawk tape.
Attached is the parameters file that was used to make the loiter flight in the attached log file.
I really appreciate your help so far. I was making good progress with your input :-)
Richard, Glad I could help. I just use the 3DR foam that comes with the pixhawk. My blades are balanced well and the tracking is visually good. check your logs and see if your vibrations went up when you developed this roll problem. make sure the pixhawk has not loosened up on your frame. 3DR sells the 3M foam separately now. Its amazing stuff!!
David R. Boulanger
Richard, I'm not sure if the toilet bowl is from PID's or the compass having interference. You could try asking this on the Arducopter support forum along with the log. Rob is the expert and at some point would probably have some ideas for you to try.
David R. Boulanger