jittery hover but not oscillation

[sturgm]

I have just rebuilt my quad after a crash and have replaced a few parts.  I have also changed the firmware to Multiwii 2.1
It has new 10" props that have been balanced. The escs have been set along with the ACC, Mag etc and all seems fine.
However once in the air and hovering it has the jitters and takes continual stick input to keep it flying, throttle, roll and pitch.
I have tried all the PID values from 2 to 9 in increments but nothing really has any effect.
I have also tried holding the quad while applying throttle and when it starts to get light I can feel the motors twitching.  This
seems to be at random as though interference.  I have some thoughts as to what it might be but I would appreciate  any other
suggestions. 

[Gaza07]

Hi Mike have you looked at the prop adapters see if there is any wobble in them I had a big crash recently and after rebuilding had a weird little wobble not related to PID settings Im pretty sure its one or two bent prop adapters 

[sturgm]

Hi Gazza,

Thanks for the speedy suggestion ( as always ).  I will check those tomorrow and let you know how I get on.  I will also be checking many more possibilities along the vibration theme and hope to sort this quad once and for all.

Mike

[Gaza07]

I will be very interested in your findings as this is just the kind of problem that can be difficult to track down 

[sturgm]

Well I have been at this problem all day and I still have !
I found a slightly out of balance motor and replaced that.  No luck
Then balanced up a completely new set of props.  No luck
Reset all the escs individually. No luck
recalibrated ACC and Mag.  No Luck
Replaced the rx with a spectrum one and used my other Tx (a Spectrum).  No Luck
Flew the quad in acro mode. No Luck
re installed the firmware. No Luck
Every PID setting for Roll and Pitch from 2 to 9.  No Luck
It certainly is flyable but it is 'chattering' all the time on all axis and throttle.
It reminds me of my heli flying days when the gyro servo would chatter and grumble sometimes.

Anyway, I am now going to reload the Mega firmware and see what happens with that.
If that does not work then I will try another  manufacturers FC.

Mike

[Gaza07]

Hi Mike what version of megapirate are you using ? 2.7r4 by any chance,
have you tried lowering the Rate P this setting is the main one for causing wobbles

[attach=1]

[sturgm]

Hi Gaza,

I am currently using Multiwii.  I originally was using Mega but after the crash I thought I would give the Multiwii ago as the air studio boys
think that it is more easier to setup.

I am however going to load Mega back on it tomorrow and see what happens.
This is starting to be tiresome

Thanks Gaza

[Gaza07]

have you tried another flight controller on the same air frame to see if it has the same problem it could eliminate your flight controller from the equation or point to it as the faulty component 

[sturgm]

Cured
Yes I have a stable quad with no jitters.  All I have to do now is fine tune the PID's and off I go.
It was the gyro LPF setting in the multiwii code.  I had a feeling that it could be something to do with the gyro settings from my heli experience.
so I trawled the net looking for hints on gyro problems.  Then LPF popped up.  I followed this through to the Multiwii code. Found the values and changed each one until eureka the jittering vanished and was replaced with oscillations this was a defining moment for me.  I started to tune some of the PIDs quickly and the oscillation went.  All this was done while holding the quad at arms length while applying throttle till it just went light.
I then gave it a test hover and WOW this is what I had been looking for.  The code is in config.h


    /*********************    Lowpass filter for some gyros    ****************************/
      /* ITG3200 & ITG3205 Low pass filter setting. In case you cannot eliminate all vibrations to the Gyro, you can try
         to decrease the LPF frequency, only one step per try. As soon as twitching gone, stick with that setting.
         It will not help on feedback wobbles, so change only when copter is randomly twiching and all dampening and
         balancing options ran out. Uncomment only one option!
         IMPORTANT! Change low pass filter setting changes PID behaviour, so retune your PID's after changing LPF.*/
      //#define ITG3200_LPF_256HZ     // This is the default setting, no need to uncomment, just for reference
      //#define ITG3200_LPF_188HZ
      //#define ITG3200_LPF_98HZ
      //#define ITG3200_LPF_42HZ
      //#define ITG3200_LPF_20HZ
      //#define ITG3200_LPF_10HZ      // Use this only in extreme cases, rather change motors and/or props

do as instructed and enjoy the result!!

caveats
I had spent a lot of time balancing and setting up the quad mechanically so I was sure that it was not an imbalance type issue.  I feel
that there is some form of resonance in the very stiff carbon firbre frame that is inducing certain frequencies that have been upsetting the qyro
as the problem also existed in acro mode and that only uses the gyro's.
So in short this solution is the last chance salon. But what a result!!

I also made a clever little device that found some motor and prop adapter problems.  I will take a photo and post it here later.  It uses some plastic and a CD.

[Gaza07]

Im pleased you have managed to get a good result Mike, I have noticed the gyro filters before but never used them but think these plastic armed quads probably need them for the reasons you stated 

I will definately be having a play around with these settings myself now to, thank you for taking the time to report your findings im sure it will be a great help to others, It certainly will be for me 

[sturgm]

As promised I have posted a couple of photos of the CD and adapter that I used to track down any motor imbalance.

Here is the adapter turned from plastic bar that has a 6mm ID and a 22mm OD and a 15mm ledge 1.25mm depth

[attach=1]

Here it is fitted to a motor with the CD

[attach=2]

If the motor is now spun up any misalignment of the prop adapters will show up as vibration this can be felt by touching the arm or gently touching the spinning disc.  In a bad case the quad will skid along the surface (if smooth enough) slowly because of the vibration.  It will also show up any imbalance in the motor assembly.  This happened for me as I had not noticed that one of the motors had taken a slight knock in the crash and the axle had been bent out of true.

caveats again

The adapter must be turned true in one operation and with tight tolerances.  Any slop will negate any results.
Pick and check a good CD not all are the same quality.  It took me a few goes to find a good one.

One last thing I will be posting later, on how I balanced the props as there is more to it than meets the eye.

[Gaza07]

That is an excellent idea thanks for sharing Mike 

[Gaza07]

Hi Mike I have had a play with the gyro filters as you mentioned above and it really has made a huge difference the little annoying twitch / wobble I had has gone, Im at 98hz at the moment and will try more of the options but very pleased with the reults so far

[teslahed]

I gave the low pass filter a go after reading of your results sturgm;

MPU 6050 Low pass filter test

Until the wind drops enough that i can fly outside i can't be sure, but it does feel a lot smoother over my bed and seems to suffer less from prop wash induced wobbles.

[sturgm]

Hi teslahed,

I have been out in the wind quite a lot today and have already got through 2 * 5000 mAh batteries.  It flies like a dream and having also now set up the PIDs it will just stay where is is told to and will face up the wind and virtually not move.  I have yet to tune it more but it all looks great.

Thanks for the reply

[Gaza07]

Here's a video of mine after editing the gyro filter it flies much nicer and has lost its wobble' it even felt stable enough to fly nose in 

[smg id=61 type=av]

[teslahed]

Thanks for the reply

Thanks for spotting the setting

I've just tried the quadcopter above my bed again, this time with the setting;

Code Select Expand

#define MPU6050_LPF_42HZ

and it feels even more steady compared to 98HZ. I'll have to wait for the wind to drop a bit so that i can try it outside before i can be sure that there isn't any negative side effect but indoors the results are quite impressive.

It's surprising this feature is so poorly documented considering the improvement it seems to make. The comments in the code are fine but I've not read about people using it anywhere else online.

[sturgm]

It's surprising this feature is so poorly documented considering the improvement it seems to make. The comments in the code are fine but I've not read about people using it anywhere else online.

As I mentioned in one of my previous posts.  It was my experience with helicopters that pointed me towards the gyro.  I have built and flown many types of Helicopters and also fitted most autopilot systems available and twitchy Gyros have always popped up some where.  However without an FC that has accessible code that can be modified, position on the frame and mounting substrate were the only available solutions. But this can be a doubled edged sword.  With so many choices of tuning values (PIDs) and many symptoms that are similar and some times well meaning contradictory advice the answer can be lost before the question is even asked.   Anyway not wanting to get too philosophical about this I am happy that some people are starting to benefit from this enlightenment

Mike

[sturgm]

As I mentioned in a previous post I would add a guide to prop balancing.
Well, as we all know unbalanced props are the root of most stability problems as this can cause vibrations that can effect
the proper function of the quad's sensors and as an aside premature prop failure with catastrophic results!
Firstly the props have to be balanced statically using a good quality magnetic balancer.  The reason I chose 'statically' is that
they also have to be balanced dynamically.  'Why is that?' you may ask.
Well as the picture shows.  I have balanced that prop quite well, but would you to fly with ?
[attach=1]

This picture graphically shows what static balancing actually does.  It makes sure that the center of gravity (cog) is straight through
the axle.  The problem here is that once fixed to a motor the cog has nothing to do with things anymore.  It is centrifugal force that now takes
center stage. 
(Centrifugal pulls away from the axis where as Centripetal pulls towards the axis)
So if the mass distribution along the props blades is not equal on both sides vibration will occur as the rotational velocity increases.
Think of a conker on a string the longer the string the greater the pull the shorter the string the smaller the pull.  So even though the conker has the same mass in both examples
it excerpts a different quantifiable force in each case.  This is the same on a prop.  If on one side there is more of the mass closest to the axle and on the other side it is closer to the tip it will balance OK statically but will cause vibration dynamically.  Getting your car tyres balanced is a good example of dynamic balancing.

So what can be done?
You will have to make a jig similar to the one in the photos
[attach=2]

Notice the dial on the stand.  This is just touching the board and will pick up any vibration and display it as rapid swings in both directions which
can be read from the dial.
[attach=3]
The prop to be tested is put onto the motor and then brought up to speed if the dial starts to swing as a result of vibration the graduations are noted.  For example 3 grads each way.  Some electrical tape is applied and run up again.  If the swings are greater, say 4 grads then it should be changed to the other side of the prop.  Continue doing this until balanced by changing the amount of tape and position.  Once happy you can then sand down the trailing edge of the heavier side until balanced again or leave the tape on.

I hope this helps explain why vibration still occurs even though your props are 'balanced'

caveats again.

This is a very brief and simple explanation of the forces involved I unreservedly apologize to any purist out there

[Gaza07]

Have you seen this Mike what do you think to this method 

http://youtu.be/PqtUSSdf8b4

[sturgm]

If it gets you to where you want to be then it's a good idea    I personally prefer a rigid jig with one test motor so all you end up testing is the prop and not everything else that is hanging on it.  However, if the vibration is eliminated then back to my first sentence

[sturgm]

Since my last post I have been trying to reduce my vibration as I intend to use the quad for some FPV. I am also aiming for about 50% throttle hover as this seems to be a requirement if I choose to try Mega Pirates for position/altitude hold as when exiting it will use your throttle position and this can lead to some surprises, either roaring of up or down if the position does not correspond to your hover point.  I speak from experiences on this point.
It is a shame as in my heli flying days 75% hover was recommend with idle up and inverted flight so 50% for me does not really feel that comfortable at the moment.
Anyway I viewed and read once again all the topics I could find on the subject and ordered some more props.  So I ended up with six sets of props from various suppliers with a quite large difference in prices ranges from about 10$ a set to about 40$.
I also went back to basics.  What am I trying to achieve?  The answer is three 0's on my gyro with default LFP.
So I clamped my quad to a bench but with a pair of arms over the side to cut down any prop wash which effect the results.
Uploaded a version with default LFP.
Connected the quad to the MultiGui.
Selected only the gyro trace to be displayed.
Selected Acro mode on the tx
Removed all the props and ran the quad up to 50% then 75% throttle to check the balance of the motors.
The result was all 0's.  So the motors were perfect.
I stuck a piece of tape onto one motor just to see the effect and it showed up as a couple of 1's and 2's here and there
so it proved that everything was working as expected.
Now the important bit
I connected motor 1 directly to the throttle channel on the rx so there was no input from the FC into the motor as I had found that the FC will try and react to the vibration and make matters worse causing more resonance resulting in greater vibration and so the results are of no use.
I also attached a new prop that had not been statically balanced.
Another important bit
Static balancing on a prob balancer is of little or no use.  Yes you are reading right here.  I discovered over many days of balancing and testing that
this is true.  As an example  I can balance a prop to perfection on a prop balancer fit it to a motor then balance it by running it up and studying the gyro trace and carefully applying tape till I obtain my 3 0's and a silky smooth spin up then take that off of the motor put it on the prop balancer and it is clearly 'out of balance'.  So in short I miss out completely the static balancing!

It is dinner time now so I will post the rest later

Mike

[sturgm]

I thought it best that I went back to the prop balancer again.  I bought two middle of the road prop balancers to check as I had read some articles about quality issues.  Both are of the magnetic type.  This type is a must with these very light props.  Firstly I checked if the spindles and fittings were balanced.  They were both out.
So how could an out of balance prop balancer, balance props correctly ?
The shafts were straight and true but one had been machined at the ends off center.  So I machined these up which made the shaft shorter so I added another magnet to one end to compensate.
Next both of the knurled prop grippers had not been drilled true.  This could be seen by fitting both to the shaft with no prop and checking if they were out of balance.  So I gave up with those and made two new ones.
I now had a prop balancer that would balance.
I compared the results of the two balancers on a single prop.  Both gave different results.
On the stock balancer if the prop was balanced true then moved around by 180 deg it was now NOT balanced!
Also if the prop was taken off and then refitted it was also not balanced!
Try this on your balancer and see if you get the same results.

Back to the Quad
Another issue to look out for is the prop shaft adapter.  The only ones that I have found to be correct are the ones that come with
Graupner Elektro Props ( well they have to be purchased separately ) all the others are at best questionable, at worst useless.
I discovered how bad these are after I had balanced a prop on a motor took it off, then refitted it only for it to now produce vibration even though I had marked the prop and motor so it was put back on the same position.
Further investigation revealed some small slop in the fit either on the shaft or on the prop on all of the prop sets except the Graupner.

The main problem is how the adapters are made.  Being injection molded the moulds have a slight camber to enable the parts to be released in production.  This produces a slight taper in the adapter which can cause it to tilt slightly when fitted to the prop so if the prop and adapter are now pushed onto the shaft slightly out of true the adapter will allow the prop to sit slightly off center.  After being taken on a and off a few times it is unlikely that the prop will  sit in the same position twice in a row.  Also the adapter should only be used only as a guide for the prop for centering the machined bottom of the prop is used for leveling only.  This must also be checked for pimples etc from the moulding process and should be carefully filed away with no disturbance to the mating surface.

The balancing is quite straight forward.  Apply small pieces of tape on the props until it shows 0's at your hover point.  This is a time consuming and tedious job but it can be done by careful observation and and small adjustments.  You can also check on You Tube for some vids.

After each prop has been balanced all four should be replaced and the rx connected as usual.  Bring the props up to speed and look at the gyro trace.
It is unlikely that you will get 0's so recompile with a lower LPF until the 0's return.


You will now have the smoothest quad around

Lastly I now only use Graupner Elektro Props.  Three of the four I bought produced three 0's straight out of the packet.  The last one just needed a small amount of tape.  Also the adapters are machine finished and do a brilliant job.


Mike

[Gaza07]

Am I reading this right Mike your only now doing static balancing and not dynamic ?
I did think that if you balanced a prop statically and then balanced it dynamically it would alter the original static balance and then there would be no point in the original static balance, I think your right a lot of vibes are caused by cheap inaccurate prop adapters 

[sturgm]

Oh dear Gaza ::). It is the Static balancing that has little or no value.  Firstly  most popular static prop balancers are poorly made and are not even balanced themselves.  So it is impossible to balance a prop successfully.  If the prop is then dynamically balanced successfully on the motor by addition of tape etc and then taken off checked on the prop balancer is will most probably be shown to be out of balance !  Also if it is then fitted back onto the same motor using the supplied prop adapters it will most probably not be balanced again due to poorly made adapters not running true.

Caveat
Please note I am using 'probably' a lot here.  There is always a chance that none of this will be experienced by a percentage of people but the likely hood is that it is more common than not.

Mike

One last point I missed is that if three 0's cannot be achieved it is worth loosening the prop and then rotating it on the adapter through 90 degs
and testing again.  Do this three times and if an improvement is obtained at any position focus on that point and make the adjustments smaller until the best result is obtained then mark the motor and prop so if the prop has to be removed at any time it can be put back at the same position.
Why is this?
Well there are certain dynamics that exist between the prop and motor when connected that can be tuned as when joined they do become a balanceable entity in their on right.

[Gaza07]

oops did I get it the wrong way around