Severe vibration issues

[hogster]

Hi Max,

Thanks for your replies I'm also not quite sure where I read that - before I ordered the Cortex I read no end of threads on coax systems, some said you need bigger props on top, some said the opposite, likewise for the pitch, with user reporting advantages of every combination! ... In the end I gave up reading and just went with a platform that had been flown and seemed to work well by all accounts.

Yes you've interpreted correctly - all the top props are 15x5 and spin one way, and all the bottom props are 16x5.4 and spin the other. I believe the reason is that the Cortex manufacturer uses DJI flight controllers and that's the prop arrangement they require. Not ideal I admit, but it's what I've got to work with at the moment!

I only linked to those videos as it seems to me the current prop arrangement doesn't make the platform unflyable, in fact it seems to fly well.

What is the reason behind making changes to the mixing table when using different props on the same motors? Doesn't the control algorithm just give as much power as is required to the different motors until the vehicle responds as it's commanded to? Sorry if I'm being dense

Many thanks again for your help David

[Max]

Hi David,

Yes, lots of opinions on the topic, exactly my point. Very little scientific study. I have no idea what DJI does... I assume they have a large R&D team and know what they're doing, but they don't seem to share that info with the rest of us I have no idea how they'd make a setup like that fly efficiently, but I'm no expert. My point is that just because it works on a DJI, doesn't mean AQ will handle it the same way. Do we even know how they mix that setup (is that something that can be modified or even viewed by a user)? I think you'd need at least that basic info.

As to the why... a MR yaws by creating a difference in torque/lift provided by each set of CW and CCW props. When the power output is the same on both sets, the MR doesn't yaw. If the output is different, it will want to continuously yaw in one direction, with the speed dependent on the output power imbalance. So, say you put larger props on all CCW motors, and give them the same amount of input power as the CW ones. Now you have a power output imbalance, and the MR will want to spin CCW. Now the flight controller says, whoa, we don't want to yaw... so it reduces power on the CCW motors and increases it on the CW ones. Now you have sets of motors constantly spinning at different rates just to maintain yaw heading, and they have less overhead for further power changes. At best, some will just heat up more and you'll lose some efficiency overall. But at some point, the motors may saturate or cut out altogether when needed to perform actual maneuvers.

Check your flight log and look at MOT_PITCH, MOT_ROLL, and esp. MOT_YAW. They should all average around zero. Anything else indicates a power imbalance (mismatched props, tilted motors, ESC differences, etc). The flight controller might be able to compensate for it, but you don't want that under normal circumstances.

-Max

[hogster]

Thanks for your reply Max

Regarding the mismatch of power output, there doesn't seem to be a hard and fast rule to say which prop you need on the bottom if you have prop X on the top? Therefore I can't imagine having a coaxial setup where the top and bottom power outputs would be identical, eg. there would always be an imbalance which the flight controller would need to deal with?

I have attached the MOT_YAW plot from today's flight, and there is indeed quite an offset. Is this something I can address by reducing the mixing table values for the bottom motors? As the bottom props have a higher pitch and larger diameter, I imagine they will always require less power (or throttle %) to achieve the same torque as the top props? Could I, for example, reduce the bottom motor values to 80 across the board (instead of their current 100), do another flight and see whether there's still a yaw offset, and keep repeating this until the offset has gone?

I also repeated the vibration tests, this time with the X8 sat off the ground on a bench (with about a metre of air under each pair of props) and recorded the following (see attached). I found that at 35% the X8 would still sit quite happily on the bench without wanting to tip over. So the plot shows all props running at 35%, then the top only at 35%, then bottom only at 35%.

Another thing that still concerns me is still how noisy the motor demands are. I've attached another plot showing one of the motor outputs during the flight, and it's no wonder I can hear a scratchy noise coming from all the motors! Is there any way of smoothing these demands before they get sent to the motors? I can't help feeling they're contributing to the extra noise / vibration picked up by the sensors in flight. I guess it's a chicken and egg thing – sort out the vibration and the motor outputs will sort themselves too, but if the outputs could somehow be smoothed anyway, surely that would help matters?

Thanks so much for your help Max, I would feel horribly alone without the help from this forum

David

[Max]

Check the default mixing table in QGC for an X8. See how the rotation directions are offset (not all top/bottoms are spinning the same way)? This evens out the effect you're talking about. With a Y6 coax, yes, you can't get rid of the imbalance entirely because of only 3 props per plane, but it's still pretty minor (I've gotten it down to almost zero avg. MOT_YAW on mine). Your YAW balance is way way off, as I expected.

As for adjusting it via mixing table, yes, theoretically that's possible, which is what I've been talking about when I say you need to see the DJI mixing setup to have any hope of recreating that setup with an AQ. Back to my original advice though, get it flying with an actually proven AQ setup, same props top and bottom, and then experiment once you have a known good starting point.

With that yaw imbalance, you'll have 4 motors working very hard and 4 working a lot less. I'm not surprised if some of them sound like they're about to die and the whole rig doesn't sound right when flying. Sound is a major indicator of MR tuning, you're right to pay attention to it. When it flies well, it sounds smooth.

-Max

[LPR]

David

"Another thing that still concerns me is still how noisy the motor demands are. I've attached another plot showing one of the motor outputs during the flight, and it's no wonder I can hear a scratchy noise coming from all the motors! Is there any way of smoothing these demands before they get sent to the motors? I can't help feeling they're contributing to the extra noise / vibration picked up by the sensors in flight. I guess it's a chicken and egg thing – sort out the vibration and the motor outputs will sort themselves too, but if the outputs could somehow be smoothed anyway, surely that would help matters?"

I'm having the same type of vibration problem as you are having with my octo. My octo has 17" props and the props are overlapped 7cm.

I'm going to test for vibration by connecting all the signal wires to one PWM Rx throttle output. This will give me a constant PWM for all the motors so I'll be able to see just the vibrations caused by the motors and props.

By looking at the telemetry data I'll be able to see the PWM throttle value while testing for vibrations..

Larry

[hogster]

Max - I will ask the Cortex manufacturer if there's any mixing information he's able to share with me. However I don't imagine this settings in the DJI flight controllers go into that much detail? Will see ... I agree it's a bit of a shot in the dark without knowledge of the settings the DJI controllers use, but I'm reluctant to drop any more money on this platform than I already have. With the current T-motor props, it's not easy to get a set of 15 x 5 props of the opposite twist to the top props as they're always sold in pairs.

I will certainly get a prop balancer and put all my current props through it - I've read a few complaints that the 16 x 5.4 props can arrive out of balance - won't harm to check them. I know this doesn't get around the offset yaw problem, but it should be one improvement I can make with relatively little outlay

Larry - you don't need to fiddle with PWM signal wires to achieve the effect you're after, that's what's so neat with the AQ. In the motor mixing table on the GCS, just set all the roll, pitch and yaw values to 0 - they your throttle stick will purely drive the speed of the motors. Just don't try flying it in that manner!

Hope that helps David

[LPR]

David

That should work and it is easier.

Larry

[Astudillo]

Hey guys,

I also have a vibration problem, my acc's are doing around 1.5 in hover with spikes up to 4
So i made this setup:


I tested all 6 motors separately form each other, but with the same esc and same clockwise prop.
On this picture you can see my best and worst motor on 50%. That is quite a difference!


So my question now is what is a decent vibration level for just one motor?

Cheers,

Ray

[JussiH]

Getting under 1 can be hard. Up to +/3 on the Z-axis is in my experience ok. In general, if it flies ok and you dont have jumps in pos/alt hold, you dont need to. Its mainly the z-axis you mainly have to look at when judging vibration levels.

Both above examples should be within acceptable, but you should try and get the worst motor to perform closer to the best one.

The main problem that can come from High vibration levels, is a Z-axis saturation that leads the filter to think its freefalling and the controller responds by gaining altitude rapidly. It will only happen in pos hold mode, flicking back to manual restores normal control. Sometimes it will fly fine in low wind, but a gust can trigger the saturation when the controller revs 1 or more motors to compensate.

The Z-axis saturation usually can happen if the vibration levels on the z-axis begins to approach +/- 10. At that point, data begins to get lost and the UKF can no longer estimate vertical velocity correctly.

And its not only about amplitude, frequency comes into play at well. Low frequency vibration at high amplitude seems to be less damaging than high frequency vibrations at low amplitude.

In any case, it usually only happens with high KV motors and fast props. Gemfan 0845 EPP style propellors on a 1100KV can generate the exact right type of vibration to screw up the Z-axis acc completely. Changing to HQ 0850 makes it go away.

[Astudillo]

Hi Jussi,

Thanks for you clarification!
I was looking into my vibrations as my copter is behaving strange in alt hold as you can see in the video.

I tried tuning the PIDs, but without succes. So could this behavior be result of to much vibration?
Or do you have any other advice?

Cheers,

Ray