cory@cytranic.c Posted at 2015-1-16 06:50
After 20 or so flights myself, I'm very happy.

Well sh*t in a bag and punch it! I just spent the last 5 minutes running around the garden because...it worked!!!

This thing has been driving me nuts for weeks, I was incredibly glad when I found this thread because it meant that there was a solution. So thanks Cory for your post, I know I've helped further the solution but that wouldn't have happened if you hadn't had posted in the first place and I like many others would still be completly stumped.

Must admit its more like 98% perfect but I think if I get myself a proper spirit level then I'll get closer to 100%. The trial and error method will give you a perfect result but this will give you a quick fix thats pretty close.

I guess now I'll have to see if there is an easy way to level the IMU in the hull for a permanent fix, but first I'm going to get a surface spirit level - thats the one that measures the level in every direction.

Wow!....I can finally sleep tonight ;)

cory flat out rawks.

roger.allen@red Posted at 2015-1-19 01:26
Well sh*t in a bag and punch it! I just spent the last 5 minutes running around the garden because ...

ah yes. DJI if you are reading this then you owe me a propeller and a gimbal which broke whilst trying the trial and error method - wouldn't have happened if this IMU issue wasn't there

I have been fighting this through three different series of phantom quads:, vision, vision plus, and the phantom 2. My latest phantom 2 is the newest version with the new controller and new compass and h3-3d, mini iosd, avl 58, and fpv hub.  I thought I tried everything, I have 22 years of RC experience including helicopters, jets, and giant scale planes ( just pointing out this isn't my first rodeo) and I could not get this fixed.

I tried your solution and boom! Fixed! Great job I am not much of a forum poster but I wanted to take a moment and say thank you! This has been bothering me for a few years!  

Again great job!

Saw this in another forum: wouldn't you want to calibrate it so that the motors level?

Good job on solving this. I'll have to pop the top and put a level on it.

P2V+ V3 bought in dec 14 direct and was fine.  I noticed the drift when rotating a couple weeks ago and didnt think much about it figured it was normal or the wind. It drifts a few feet ir more. Does not spin on a dime. Did the imu calibration thinking that would help, the basic version. Then it seemed like it was flying all funky. Had a hard time landing as it was drifting so much and bouncing in no wind. Finally had to grab it in the air with my free hand.

The second battery i redid the compas calibration and it helped with the landing but still drifts. It just seems so off and i dont understand why this is the case. It seems that these copters are buggy. For this kind of money there is zero excuse for any of this, it should be dialed in before these even went into production.

Very bummed right now that my phantom wont fly right after 20 or so flights in the last three or four weeks. Ugh.

ZippyZapp Posted at 2015-1-19 11:29
P2V+ V3 bought in dec 14 direct and was fine.  I noticed the drift when rotating a couple weeks ago  ...

Zippy, open up the Phantom. put a surface spirit level on the IMU, place paper underneath the phantom until the level is lined up, then do an advanced IMU calibration. That will fix your problem.

I've been equally frustrated too but this finally fixes it

BTW. I'm going to make a video to explain this, just waiting for the surface spirit level to arrive....

roger.allen@red Posted at 2015-1-19 17:10
Zippy, open up the Phantom. put a surface spirit level on the IMU, place paper underneath the phan ...

I've never opened it up and have read conflicting stories on if this voids warranty.
Can someone from DJI please comment on this and the IMU?

Thanks for the instructions, that's great!

cory@cytranic.c Posted at 2015-1-16 06:50
After 20 or so flights myself, I'm very happy.

Wow.... i just feel like jumping of happiness with this thread ...

I have some issues wih my new V3, drifting is one of them... but i have more complicated ones in queue...

I am so "sad/mad" with my situation that i didn´t touched my Phantom since my help thread posted in this forum and i did not try this fix, but it makes a LOT of sense and i will do that in the future

EVENTUALLY (sooner than later) we ALL will calibrate the IMU... so... we ALL WILL create an Issue that NO ONE (phantom users) will know how to help....

People will tell you (us/me) that you need to read the instructions or that doesn´t hapen to them or something else that makes the person with the problem looking stupid
So.... normally... the person with the problem will shut up and live with the problem after WASTING a LOT of monney on this....

So WHY DJI let us calibrate the IMU...?

Don´t they know that their IMU (inside the phantom) is not horizontal and calibrating WILL NOT fix nothing, instead will make it worst?
They should know... they are not users....

THANKS Cory and Roger !!!!!!

what size of a spirit level do i need (ie how tiny does it have to be and is the IMU obvious when i open it up, i didnt see photos of the inside and havent done any opening as of yet)...

Another satisfied customer!

Just wanted to say thanks Cory. I tried your method to correct a 10ft drift to the right and it's now fine. It took 13 goes and it's now recorded in case I have to do it again anytime.

Hey guys you're welcome! I too spent days finding threads about yaw drift that ended with no solution. (there are 100's) I figured there had to be something I could do, since mine was flying correctly out of the box until I did the IMU

I have a smile on my face that I was able to jump start the discussion for the yaw drift problem and get it fixed!

The spirit level on the flight controller is definitely the way to go. I'

would this be a good spirit level to use?

http://www.amazon.com/Circular-B ... words=spirit+levels

also, does anyone that has been doing this approach have a photo of the IMU with the case off by chance that I can use for reference?

gnixon2015 Posted at 2015-1-20 05:31
would this be a good spirit level to use?

http://www.amazon.com/Circular-Bubble-Spirit-8-25mm-Tripo ...

It is the "yellow" labeled NAZA unit with "Phantom" on it.  It has been assumed, by myself as well, that the "IMU" components are within this unit.  My shimming efforts for calibration leveled the unit.  My next step is to level the unit with the craft level, and re-calibrate.

Also is a picture of mine with a tungsten rod across it, and you can see how much it's out of level compared to the Phantom.

Jamie Hellmich Posted at 2015-1-20 05:57
It is the "yellow" labeled NAZA unit with "Phantom" on it.  It has been assumed, by myself as well, ...

Hey Guys,

I should have a youtube vid ready in a couple of days to show what I did and how, I actually used a normal 6 inch spirit level to start with but now I'm waiting for a 'bulls eye surface spirit level' to turn up which cost about 2 quid on ebay and is about 3 cm wide http://www.ebay.co.uk/itm/ORBIT- ... hash=item43b1f0c511

I've actually now thought of a couple more ways to do this, but I need to test something first before I explain them because they could be a red herring...

Roger

I plan on doing the sport level thing but not until I open my Phantom for my 10 hour flight maintenance check. What I did do was lower the gain setting on mine. I did that because I felt the default setting of 100% was to high and caused a jerky initial yaw.

The lowrr rate seems to help but it still drifts, and of course it rotates slower because of the lower gain setting. My controller already has the switch with the 10k resistors modification that slows the yaw rate even more when the switch is flipped.

I will measure the IMU, but I'm just not really in a real hurry to do it right now. My plan is to open my Phantom every 10 hours for inspection. But let's be clear lowering the yaw rate did not fix the problem it just lowered the effect of the problem.

Again kudos go to those who solved this issue.

kudos indeed, ill defn be looking into doing these things if/when i open mine up (and after ordering a spirit level).  great read.

one thing ill add to this overall, and this is in terms of diagnosing actual drift from perceived drift (so that some people dont mess with IMU inadvertently).  from what ive read, you can defn make things worse messing wtih IMU so make sure you do have an imbalance before tinkering.

when flying in wind (Even light wind like say 5-6mph), i can confirm that the phantom disengages GPS hold when doing nothing but yawing (that was referenced in this thread and a few others) but ive watched it drift from the wind while yawing, and then when i stop, it returns to the old gps position.  this proves that it knew you only spun and it just never compared the current to target gps coords WHILE you were yawing.  then, when i put it in a situation with NO wind AT ALL, it spun with no drift.  

i believe this is because while you are actively flying (sticks not at neutral) it is disengaged (like putting a clutch in) the GPS hold until you let up.  also, i think, while you are pressing, it is never recalc'ing or re-evaluating the gps coords if ONLY yaw is being pressed.  it seems when i press roll or pitch it knows that would be lateral movement so it keeps 'updating' my gps location, but when yawing, i think it seems to ignore that (because, well, it makes sense that yawing and nothing else, you'd not be changing location).  probably not a bad thing since if it were trying to fight wind and stay in position while i was also yawing, i could see stability potential there (maybe not just thinking it through, ohterwise, why make it do that).

now this certainly doesnt mean that anyone's is likely that versus real yaw drift, but evaluate it in low wind conditions as i think my perceived yaw drift is only happening in moderate wind situations so I think it is best i leave it alone for now.

but when it comes time, and im sure it will, this thread will be a total life saver.

gnixon2015 Posted at 2015-1-20 13:32
kudos indeed, ill defn be looking into doing these things if/when i open mine up (and after ordering ...

That all makes sense. I was going to say that we could test this 'clutch' theory by looking at the data presented by the iOSD whilst yawing etc but that I'm sure it works completely independently from the IMU.

GPS 100% disengages while it yaws....HOWEVER, it also reengages if you continue to spin and yaw.... Perhaps there is a bug...

For example, if its 25MPH winds outside and I start yawing, it will fly off in the direction of the wind...After about 5 seconds it regains its GPS  abilities and starts to fight the wind, even when yawing.

Its strange no doubt.

thanks again Cory for the solution you provided.
got me thinking about the whole process of initial calibration at DJI.
it seems to me that when setting this at dji they would not be doing it on a kitchen bench :-) but most likely would have a proper level surface (most likely to an international standard) ie. dead flat and dead level. If this is the case then any recalibration by a future user would almost certainly be not done on a dead level surface, and hence the problem we experience.
As I mentioned in an earlier post, I opened my phantom 2 prior to recalibration and noticed the sloppy job that was done in fitting the naza unit in the first place. With this in mind, I thought I would try the get as close to a dead level surface as possible. I had to assume a couple of things ... that because the phantom 2 from its design and manufacture, everything about the quad itself should be symmetrical and fairly level...there is also the possibilty of slightly more thrust from a particular motor or prop.... but nothing we can do about this.
My solution to the problem was to sit the phantom on a flat surface and then place a flat board (that I know is not warped or twisted) across all 4 motor spindles (shafts). I then sat my iphone using a compass app (which includes a level)  in the centre of the board (use an app that has a level with degrees to 2 decimal places) where I could clearly see that the quad was out of level. I then used cory's method to place a few business cards under the appropriate legs to get the iphone level. Once I got this close to dead level, I ran the advanced calibration and found that it was very stable on first flight. This method may well save time and effort as opposed to the trial and error method.

wow disco AWESOME suggestion

gnixon2015 Posted at 2015-1-22 10:44
wow disco AWESOME suggestion

Hi

Yes that's another good suggestion. I've come up with another one too but haven't had a chance to fly yet to see what happens. I'm not sure what result of this will produce....

Put your phantom on a table. Turn it on and plug it into the Assistance software. Go to the tab where you do the IMU calibration but don't perform it just yet. Note that you can see the accelerometer status for X and Y and that these values change real-time above and below zero as you lift the phantom in various directions. With this in view, add paper below the phantom until both of the X & Y values rest on 0.00. Now do the advanced calibration....

What do you think the result will be? Could it be a way of calibrating without a spirit level?

Roger

wow roger, i think you could be on to something there ... and after that, i could even do disco's method to confirm the spirit level SHOULD SHOW perfect when done with your idea... interesting....

roger, i went to look into this and found that the values that move around in the assistant while the craft is plugged in are for gyroscope, acceleration and compass.  none of those values show how level the craft is.  i dont think they expose that figure, they just read it during calibration.  therefore, i think you have to do the calibration the way others have stated.

I still plan to do disco's method because to me it seems like what is most important is how level the motors are in relation to one another not the internal imu itself.  guess we'll see...

Cory,

You got it buddy.  Mine was drifting on Yaw something terrible.  I had IMU calibrated on a pull out table from my desk.  I know that it wasn't level.  Then I put it on a leveled surface and then re-calibrated and now it's the difference between night and day.  Too bad the tech people at DJI didn't say anything about "did you calibrate on a level surface?".  Anyway, now I just have to tweak it.  Yay!

gnixon2015 Posted at 2015-1-24 03:50
roger, i went to look into this and found that the values that move around in the assistant while th ...

Hi

Whilst it says gyroscope, its the accelerometer figures that definitely line up something horizontally as I was able to get them to rest on zero by placing paper underneath. Unfortunately I won't be able to see what the effect until next week now, well perhaps tomorow. I'm going to compare all three methods though - spirit level on IMU, spirit level on motors and zeroing accelerometer and try and draw some conclusion

Roger

Roger

roger.allen.ema Posted at 2015-1-25 05:34
Hi

Whilst it says gyroscope, its the accelerometer figures that definitely line up something hori ...

Roger, Looking forward to your results ... should be very informative!!!!

michael.haws.em Posted at 2015-1-25 07:49
Roger, Looking forward to your results ... should be very informative!!!!

Hmm, so I tried the accelerometer method and disappointingly it hasn't produced a decent result yet. I then used the spirit level on the IMU and this worked really well again. I really wanted the accelerometer method to do something as this method does not rely on having to take the cover off. I may try it again though as I haven't had much time this weekend and it was a bit tricky getting it level.

I also want to try the spirit level across the motors. The span of the motors is quite large though and having thought about this I should really use a long spirit level to do this correctly. I know it is possible to use a large piece of wood or something but there is no guarantee that it will be straight so I'll try and get hold of a long spirit level this week.

With the cover off it will be possible to have both the spirit level across the motors and a surface spirit level on the IMU so just with the eye I should be able to see how they compare and hopefully see a correlation.

We shall see
Roger

roger.allen.ema Posted at 2015-1-25 05:34
Hi

Whilst it says gyroscope, its the accelerometer figures that definitely line up something hori ...

I don't see how zeroing based on the accelerometer will do any good as its current zeroes are what's making the drone drift.

I put a piece of glass large enough to cover all 4 motors and put one circular spirit level in the center and leveled the aircraft to that. This method produced an almost dead zero yaw (except for yaws in strong wind gusts, of course).

Javeee Posted at 2015-1-29 01:01
I don't see how zeroing based on the accelerometer will do any good as its current zeroes are what ...

Ah yes, glass would be good as it is less likely to bend and very likely to be flat. I don't have a piece unfortunately. I may just buy a long spirit level this weekend as I would really like to compare this method to the surface spirit level on the IMU and hoping this is the same if not better as it doesn't require taking the case off

Roger

javee's recommendation is fantastic.  i want to calibrate (even tho my yaw is very very minor i at least want to know how to do this if when it becomes more material) but i dont want to remove my cover just yet as the craft is only a few weeks old.  thanks, javee, for a fantastic suggestion.  ill defn be trying that.

gnixon2015 Posted at 2015-1-29 05:19
javee's recommendation is fantastic.  i want to calibrate (even tho my yaw is very very minor i at l ...

Anytime! That's what the forums are for!

Still haven't tried the spirit level across the motors yet, must admit now that the yaw is working again I took it out for a proper flight and shot some video, first time in months! Felt great So I added it to some other stuff and made this.



Anyway, I'll try the spirit level across the motors and report back the comparison

Big thanks to Cory for the insight!

I'm hesitant to take the lid off my Phantom and level the IMU so I'm taking the trial and error approach of shimmying the base using index cards, calibrating, test flying, and adjusting based on drift in my test flight.  While my drift is much improved, I'm concerned because my adjustment requires 48 index cards on the back left side and 52 cards on the front left side.  With these adjustments the Phantom is visibly leaning to the right.  Bottom line, this adjustment seems excessive - but it seems to be working.

Like so many others, I had absolutely no drift out of the box.

I'd appreciate any thoughts and comments and I'd also like to hear how you go about taking the lid off the Phantom.

Thanks everyone!  Ron

I had issues with drift at a hover when yaw input was made. I had several close calls with loss or crash. I attributed the lack of experience on my part as contributing.  I saw a video addressing the normal IMU readings and saw the Gyro Y was at 1.5 rather than close to zero.  An advanced IMU calibration worked for me.  The Phantom is new. I'm feeling fortunate after reading about some of the other issues that IMU advanced calibration did not help.

ok. I finally had the chance to try the spirit level on across the motors and I am pleased to say that it produces the same if not better result as a spirit level on the IMU but of course this method as the advantage in that there is no need to open up the phantom.

I would say that the best result would be by shimming paper because you can literally adjust it to your hearts content, however this will take a lot of time and flights. Something that I don't really want to do and can't do in a hurry.

The surface spirit level on the IMU and normal spirit level across the motors give you a pretty much spot on result first time so I'm going to stick with these. In the end though I'll use the spirit level across the motors because it does not involve having to take the lid off the phantom, I don't mind doing that as I have done it many times but this exercise is all about getting a decent result quickly.

I'll make a video at some point to explain this, hopefully this week

thanks roger for the update, look forward to the video!

This is how I balanced my Phantom at the motors with the props off:

This is the stuff I used.
-24" level
-caliper
-4 metalsleeve  things from Lowes ~1" tall .85" diameter
4 wire nuts 18-8 awg
-Digital Level


First I drilled out the metal spring threads inside the wire nuts. I then grinded the tops off so they were shorted than the metal sleeves. Using a rubber hammer and a dow I hammered the wire nuts inside the sleeves. The wire nuts keep the sleeves centered on the motor's prop shaft. The sleeves are use so the level stays on the phantom since it can't stay put on the motor threads.

Notice how the wire nuts are recessed so only the sleeve touches the top of the motor base and the level. The caliper was used to make sure the sleeves were the same height.

I put the levels across both axis of the Phantom. The digital level takes the guess work out. The Phantom is on one of the sections of marble floor that I know is level.



I also used a bulls eye level by turning the 24" level on its side. The white paper is a business card so you can see the bubble level. It is not touching the bubble level.


If your Phantom is not level you can use the business card method and read the digital scale to get it perfect.

cory@movecost.c Posted at 2015-1-8 11:11
Jamie, my desk is level as well..I leveled it with a bubble level. I believe the V3's IMU are not 10 ...

I'm stuck with the same yaw drift issue (so. annoying.), but your  success has really given me some optimism. Before you fixed it did your phantom always drift in the same direction whether you were yawing left or right?  Was the direction of the yaw determined from the phantom start (pre yaw) position? So if you say it drifted forward, is that facing forward from before the yaw?

Also, when I fly my Phantom it takes upwards of 15 minutes for it to cool enough so when I do the advanced IMU cali it doesn't give me the temperature error. Did this slow down the calibration process for you much?