DJI Mavic, Air and Mini Drones
Friendly, Helpful & Knowledgeable Community
Join Us Now

DIY rear compass cabling fix

How do you guys work out the "units PP" I see all these graphs, but where are you getting the number from?
 
How do you guys work out the "units PP" I see all these graphs, but where are you getting the number from?

They come from the .DAT file which consist of your Flight Log Data. In order to view the data, you must use the program DatCon. The program used to display the data by graph is CsvView. And here is the link to the Magic Kingdom: CsvView/DatCon :)
 
They come from the .DAT file which consist of your Flight Log Data. In order to view the data, you must use the program DatCon. The program used to display the data by graph is CsvView. And here is the link to the Magic Kingdom: CsvView/DatCon :)


Hello, yes I have done that and I have all the graphs and lines set out with the axis etc. But how are you determining "200 PP" are you deducting the lowest plot point from the highest and getting the figure that way. There is no specific figure that shows what they have been talking about.


i.e This user says 70 p2p? but 517 - 424 is 93? what is the magic to work out p2p or pp?

upload_2017-2-27_15-46-40.png
 
Hello, yes I have done that and I have all the graphs and lines set out with the axis etc. But how are you determining "200 PP" are you deducting the lowest plot point from the highest and getting the figure that way. There is no specific figure that shows what they have been talking about.

I.e the mag.z etc is negative -500 etc.

I asked the same a few posts up! :)
 
Just finished my mavic and the before and after pictures look the same as those of @pitchmagic. What I found was this shielded HF wire directly underneath the motor wires. You can see it in the "Before" picture of @pitchmagic on the lower right corner. I'm not sure, but it could the antenna wire of the right front arm. If so, it would be another reason for twisting the whole length of the wires after removing the top cover and not just the rear section by only removing the rear legs.

BTW, bought mine on 7th of December in Germany.

Thanks to all who described the problem and the solution!
That antenna wire is shielded so It should not be affected. However, people have noted disturbance in the range and all wires that handle current should be shielded/twisted. That problem might not be relevant but better to supress as much noise as possible.
 
Still doesn't make sense, he says 130 for both, but yet the back is 130 and the top is 260 to 390??

View attachment 7257
The pk2pk means peak-to-peak of one signal. The front compass varies between 390 and 260. The back between 0 and -130. Each having a peak-to-peak of 130

upload_2017-2-26_6-9-2-png.7192

magZ.Front and magZ.Back come from different magnetometers each with their own gain and offset. Here you can see the offset difference is about 390 (the peak value magZ.Front - peak value of magZ.Back).

There isn't an absolute reference to determine the relative gain of the two sensors. But, there are entries in the eventLog that suggest what the gains could be. For my Mavic those entries look like
4.453 : 20036417 : 1 [L-COMPASS][read app cali data]fmu:0,user:0,scale:[ 3.681 3.613 3.495]
4.453 : 20036612 : 1 [L-COMPASS][read app cali data]fmu:0,user:0,bias:[-72.854 158.315 -658.393]
4.453 : 20036799 : 1 [L-COMPASS][read app cali data]fmu:1,user:1,scale:[ 3.619 3.713 3.758]
4.453 : 20036986 : 1 [L-COMPASS][read app cali data]fmu:1,user:1,bias:[165.180 492.723 515.798]
This and the eventLogs for other Mavics I've seen have these scale values (i.e. gain) always around 3.5.

When pk2pk noise comparisons are made between different AC there is the tacit assumption made that the sensor gains are the same. I'm a little uneasy about this. But, it's been pointed out to me that knowing more about the actual gain numbers would only improve the answer by 5-10%. And, AFAIK, there hasn't been a reason to suggest otherwise.
 
Last edited:
  • Like
Reactions: Stealthmatt
Ok, that twisting made a difference - no more errors and when reading the log p2p is down to about 50 - on full speed!! Thats an impressive difference.. Comparing from the old value that was p2p 257 its about an 80% difference?

upload_2017-2-27_17-21-52.png
 
Last edited:
  • Like
Reactions: Mr Spock
The pk2pk means peak-to-peak of one signal. The front compass varies between 390 and 260. The back between 0 and -130. Each having a peak-to-peak of 130

upload_2017-2-26_6-9-2-png.7192

magZ.Front and magZ.Back come from different magnetometers each with their own gain and offset. Here you can see the offset difference is about 390 (the peak value magZ.Front - peak value of magZ.Back).

There isn't an absolute reference to determine the relative gain of the two sensors. But, there are entries in the eventLog that suggest what the gains could be. For my Mavic those entries look like
4.453 : 20036417 : 1 [L-COMPASS][read app cali data]fmu:0,user:0,scale:[ 3.681 3.613 3.495]
4.453 : 20036612 : 1 [L-COMPASS][read app cali data]fmu:0,user:0,bias:[-72.854 158.315 -658.393]
4.453 : 20036799 : 1 [L-COMPASS][read app cali data]fmu:1,user:1,scale:[ 3.619 3.713 3.758]
4.453 : 20036986 : 1 [L-COMPASS][read app cali data]fmu:1,user:1,bias:[165.180 492.723 515.798]
This and the eventLogs for other Mavics I've seen have these scale values (i.e. gain) always around 3.5.

When pk2pk noise comparisons are made between different AC there is the tacit assumption made that the sensor gains are the same. I'm a little uneasy about this. But, it's been pointed out to me that knowing more about the actual gain numbers would only improve the answer by 5-10%. And, AFAIK, there hasn't been a reason to suggest otherwise.

Thank you for explaining it, this makes much more sense now. :)
 
Mine had to go back to DJI for an unrelated warranty repair. Just received an exchange unit back and it looks pretty good.

upload_2017-2-28_11-54-9.png

If I read it correctly Front ~ 30 P2P, whilst the rear is noisier at ~ 60 P2P . Flew it flat out in sport and no compass messages. I don't actually know the wiring config, as I have not pulled the top off yet..


My original unit had a significantly noisier aft compass ~ 140 P2P, so I expect the issue is fixed for me.
upload_2017-2-28_12-15-0.png
 

Attachments

  • upload_2017-2-28_12-11-55.png
    upload_2017-2-28_12-11-55.png
    22.2 KB · Views: 51
Last edited:
  • Like
Reactions: Mr Spock
Would wrapping the wires with either copper or aluminum sticky tape accomplish the same thing?


Sent from my iPhone using MavicPilots
Hard to say. It's much easier to twist the bundle and it's clear that works.
 
^^^Yup - If you leave the wires straight and add this additional metal near the compass it potentially could even make the problem worse through the induction of a magnetic field under load. Hard to know what the outcome would be, however, twisting the wires is much simpler and effective as has been proven. I'm all for over-engineering solutions, however, it just isn't worth it in this case... For all we know the next firmware update will set the front compass as primary and this thread will slide into obscurity
 
^^^Yup - If you leave the wires straight and add this additional metal near the compass it potentially could even make the problem worse through the induction of a magnetic field under load. Hard to know what the outcome would be, however, twisting the wires is much simpler and effective as has been proven. I'm all for over-engineering solutions, however, it just isn't worth it in this case... For all we know the next firmware update will set the front compass as primary and this thread will slide into obscurity

I went into ATTI mode this morning with 17 satellites showing, almost lost it, went into sport mode to have enough authority to bring it back, lasted about 3 minutes and still ATTI when I landed. Could this be a product of the rear compass problem?


Sent from my iPhone using MavicPilots
 
I went into ATTI mode this morning with 17 satellites showing, almost lost it, went into sport mode to have enough authority to bring it back, lasted about 3 minutes and still ATTI when I landed. Could this be a product of the rear compass problem?


Sent from my iPhone using MavicPilots

Do you have the new firmware?
 
  • Like
Reactions: Mr Spock
Do you have the new firmware?

Yes, flew it yesterday for quite a while and no problems, this morning problems. This isn't the first time, usually happens when I go into sport mode, which I didn't this morning until it happened, but it was a bit windy above so maybe the motors were working harder than usual, worries me.


Sent from my iPhone using MavicPilots
 
Would wrapping the wires with either copper or aluminum sticky tape accomplish the same thing?
Sent from my iPhone using MavicPilots
Nope. Seeing as it's a magnetic compass that is being disturbed the most likely culprit is stray magnetic fields from the high alternating current engine wiring. You can't effectively shield from magnetic fields without special high mu (magnetic permeability) materials. It is rarely done and not on wiring. Regular metallic shielding is primarily for EM fields (RF) or to some extent E fields (adjacent alternating voltages) but the shield must be grounded.

The high current travelling to and from the motor generates magnetic fields and the + and - wires going to the motor will have equal and opposite current. This causes the magnetic fields to cancel out but only if the wires are right on top of each other. The twisting itself doesn't help this situation, it's the fact that twisting causes the wires to be close together that solves the magnetic field transmission problem. The twist actually guards from field reception.
 
  • Like
Reactions: DerekG
Yes, flew it yesterday for quite a while and no problems, this morning problems. This isn't the first time, usually happens when I go into sport mode, which I didn't this morning until it happened, but it was a bit windy above so maybe the motors were working harder than usual, worries me.


Sent from my iPhone using MavicPilots
Are you sure you are on FW 01.03.4000 ? If you are then i am not sure that you should be having this issue as you would be getting a warning and the redundant compass would have kicked in. There are other here far more experienced than me but this is my understanding so far for the fix that is provided via the latest FW.

Sent from my SM-T810 using MavicPilots mobile app
 
I went into ATTI mode this morning with 17 satellites showing, almost lost it, went into sport mode to have enough authority to bring it back, lasted about 3 minutes and still ATTI when I landed. Could this be a product of the rear compass problem?


Sent from my iPhone using MavicPilots
This sounds like the infamous gpsHealth-drops-from-5-to-1 issue. It's being discussed here and a few other threads.
 
  • Like
Reactions: DerekG
Lycus Tech Mavic Air 3 Case

DJI Drone Deals

New Threads

Members online

Forum statistics

Threads
130,599
Messages
1,554,244
Members
159,603
Latest member
refrigasketscanada