How to read DJI IOSD logs (How to Instructions)

[Quinton]

OK, I have just started using DJI IOSD and want to understand exactly what all the options tell us so have started this thread so hopefully we can collate as much information as possible to understand this unexplained world.

Here is the Instructions that DJI uploaded, which is not exactly great and helping decipher anything, so maybe with everyone's input we can start to understand things a bit better.

The only video from DJI explaining how it works is shown below, so if others can share any more information it would be much appreciated..

I understand x is time, but for example they say the Y axis is "corresponding data" so why does a motor sit at 17000 when not started and goto 35000 when flying, (what is this amount) any other good resources out there?


Here is an example DAT file if users want to use it in any screenshots or to help explain what exactly each item is..
https://www.dropbox.com/s/90ri83845iqkgt6/FLY004.DAT

 

[Bartman]

the 17000 and 35000 might be the PWM intervals defining the motor throttle positions....not sure, sort of a WAG actually.

 

[Tahoe Ed]

Quinton, thank you for starting this. The DJI flight data files are not the easiest to interpret. That is one of the reasons that I send the files that I receive to DJI HQ for review. There will be some enhancements coming but they are not here yet. One of the things that I will say is that the flight files should be fairly large. About 3mb per minute recorded. If they are less than that they may be corrupt.

 

[kloner]

your able to pull up input commands like what your radio told it to do and compare them to motor outputs. if a motor output goes to 100% power and that arm is the one tht dropped in a crash you can make a pretty decent educated guess that the motor or esc failed..... however, if you see a bunch of input and your not doing it, will show interference or incompatibility. when you turn on the motor outputs and there never close to being even esp in a windless hover your motors are crooked. it also shows what percent throttle is being sent to the motors for prop sizing. I had a hard time dealing with iosd files till i got ahold of a superX, then it all started making since.

looking at voltage can show a battery leave,,,,

you can tweak the gains based off of accelerometer data,,,,,

can make educated decisions on when to throw it all away during a crash by looking at g-force data, also shows vibrations like motor or prop problems,,,

the way they use the numbers is pretty confusing, but voltages are volts and this will show if you have an issue with battery or regulators....

theres a bunch of stuff i have no idea what are

the biggest prob with iosd is it must be usb1 cause downloading the file takes forever.

 

[Quinton]

It is starting to sink in a bit the more I look at it.
If you had a dead calm day, it could actually teach you a lot about your craft, the biggest problem is that they are not time stamped, so you don't really know when or which ones they are, as I seem to also get a few DAT files along with the flight ones with no data in them.
If they had time stamps included you could for example just hover for 1 minute with no stick movements and take a look at each motor to see which ones are working harder, then throw it into manual and try and do the same, but matching up the times is the hardest part.
It could be a good tool if it was easier to use, and we understood it more.

The biggest thing I do not get however is the Y axis...

in_R starts at 28000 goes upto around 33000 and down to 25000
in_U goes from 0 to -800
in_T -11000 to 5000
in_A starts at 0 and goes -1000 to +1000 (which makes sense)
in_E starts at 0 and goes -1000 to +1000 (which makes sense)

So maybe if it can be broke down a small bit at a time, and explained it may make more sense, but like I said the Y axis is just strange to me at the moment.

 

[kloner]

_r = rudder
_t = throttle
_a = ailerons
_e = elevator

I believe there just using a refrence number. pwm signals are 2000 wide, it is showing +-1000. It shows your _a at full deflection at 550 time

When i pull up your motor outputs, in particular i just hit m3, see how many times it goes down to really low, almost arming level, that looks to me like a oscilation or a high gain.

this is what a superX log looks like, you can see it explained a little simpler
http://log.xaircraft.com/report.htm#2014_07_26_073556.500_75.1.06.04

 

[Quinton]

_r = rudder
_t = throttle
_a = ailerons
_e = elevator

I believe there just using a refrence number. pwm signals are 2000 wide, it is showing +-1000. It shows your _a at full deflection at 550 time

When i pull up your motor outputs, in particular i just hit m3, see how many times it goes down to really low, almost arming level, that looks to me like a oscilation or a high gain.

this is what a superX log looks like, you can see it explained a little simpler
http://log.xaircraft.com/report.htm#2014_07_26_073556.500_75.1.06.04

Surely it would just depend what you are doing at the time, I was flying pretty hard without a any payload to try and sort a problem, sometimes motors almost stop to counteract the other motor do they not?

Now that Superx log, THAT is a data log, maybe DJI should take a page out of that book, thanks for sharing.

 

[kloner]

same thing was done on that superx log, no payload 78kmh....see my motor outputs.... no where near zero ever..... i'd expect it to shake looking at yours. i'll get my skyjib back this week and we will have a look at it to see if mine does that too

Its pretty cool except you need internet service to look at them.... it also has this
http://log.xaircraft.com/#2014_07_26_073556.500_75.1.06.04

 

[Quinton]

same thing was done on that superx log, no payload 78kmh....see my motor outputs.... no where near zero ever..... i'd expect it to shake looking at yours. i'll get my skyjib back this week and we will have a look at it to see if mine does that too

Its pretty cool except you need internet service to look at them.... it also has this
http://log.xaircraft.com/#2014_07_26_073556.500_75.1.06.04

Brilliant, just looking at your first log again, look at the battery drop when you hit 60% motor output.
Black Boxes should be made compulsory when flying if they give you the info that Xaircraft gives you, maybe then people would see problems before they happen.

 

[kloner]

it's also included in the flight controller, don't need the osd to get to it

 

[Quinton]

Is that a Skyjib Hex? Would love you to get hold of the longest retracts to see if you can get the same problem as I had with CW yaw when going fast forward.

Right, was having a little play just on the bench and the IOSD and you can learn quite a bit from it actually.
Here are the images of the 6 motors at 50% throttle then going upto 100% (without props) and the throttle data.

Notice the CCW motors (M1,M3,M5) at 100% compared to the CW ones, wonder why that would be?



Here is the log file for reference.
https://www.dropbox.com/s/1ljam9pe9p0lp94/FLY014.DAT

 

[kloner]

if that wasn't in manual mode that coulda been the algorithym doing it's thing

 

[Tahoe Ed]

Notice the CCW motors (M1,M3,M5) at 100% compared to the CW ones, wonder why that would be?

Yaw to the right?

 

[Quinton]

Yaw to the right?

No Idea, funny how all motors had it at 50% and only the CCW at 100%
Def a lot more to be learned here, and the bench without props may be the best place to try and understand what is going on.

 

[ZAxis]

A question... are people running the IOSD assistant under Windows 7 or 8 and 32 bit or 64 bit. I'm running either 7( mac/Parallels) or 8(MS Surface Pro) but always 64bit, the X axis shows a very small number and data plots don't seem sensible. Putting a multiplier on the x axis makes no difference. This happens with my own files and the examples posted in this and other posts.

any ideas what to do? ... other than get another computer!

andy

 

[Quinton]

I have mine on a W8 64 bit machine.
Do you want to upload a DAT file and we can see how it looks in our assistants, to see if we get the same, maybe its your file and not your computer?

 

[ZAxis]

I have mine on a W8 64 bit machine.
Do you want to upload a DAT file and we can see how it looks in our assistants, to see if we get the same, maybe its your file and not your computer?

Thanks for the offer.
https://www.dropbox.com/s/8wrju3dh8jlg1w5/FLY052.DAT

Its a bit big at 60mb. Actually all the files are pretty big, it took several hours for them to download so big opportunity for corruption.

andy

 

[kloner]

on mine it shows less than a second?

 

[Quinton]

Thanks for the offer.
https://www.dropbox.com/s/8wrju3dh8jlg1w5/FLY052.DAT

Its a bit big at 60mb. Actually all the files are pretty big, it took several hours for them to download so big opportunity for corruption.

andy

The good news is, you don't need a new computer.
The bad news is there is something wrong with your IOSD files.
I take it its the IOSD MK2 and not the mini or something.
I have read that it has an sd card built into them, so it may be bad or something.

How is it set up, you do have a canbus cable going from it into your GPS..right.
You are correct though there is something wrong with your x timeline it is only showing .016 when 1 is usually 1 second I think.
I see it is also the A2 which looks diffferent than the Wookong files.
Maybe TahoeEd could shed some help, but for sure you have a problem with your files and not your PC.

 

[ZAxis]

Thanks for the effort guys.
I was pretty sure it was the files. I'll check the iOSD connections but since we get the correct displays on screen I'd assume all is well there. Duff memory card is a possibility but I can't get any file to open correctly. The A2 started on v2.1, missed out v2.2 and is now v2.3. It doesn't matter what version the files were created under they act the same. A possibility was that the download thru the USB from the IOSD was bad. I did it on a Mac under Parallels which is usually OK. It was the first download and contained 4 months of data. The file sizes range from an average of around 50mb to a maximum of over 200mb. Not sure if there's any significance in that.
It would be interesting what DJI would make of it all. TahoeEd maybe will take it up?

andy