DIY rear compass cabling fix

[BadBob]

What if you take some solder wick that is big enough to put all three wires through it, then ground it out to true ground. Not a floating ground. Then cover that with heat shrink? If one of the wires is true ground then the solder wick or shielding as it is then may be used with only two wires going through it and the solder wick shield as the ground and third wire? Just don't isolate the signal antennas at the same time. I believe the control signal antenna are also in the legs? I have not opened mine. I have made mini two conductor coax this way before though. You can twist a short piece of the solder wick to extend out with smaller heat shrink around it for the connection at both ends.


Sent from my iPhone using MavicPilots

 

[Egika]

Possible - but a lot more effort than twisting the wires which obviously is good enough to solve the problems.

 

[npmullins]

Welcome @npmullins . So many Questions...
Have you had a chance to test it yet? Sounds like you did the full process too which is a first in this thread.
By removing the top - did you take advantage of being wind the cable all the way back to the circuit board?
In what state did you find the cables upon popping the top? Were the wires tucked into grey the loops near the rear arms.
Were they unwound as one might expect, if you were getting compass switching?

Interesting about the difficulty you had with the from clips. Did you start by opening the rear end first as @Thunder-drones has recommended?

Hi Logger
yes i removed all the screws then popped the clips at the rear first. The piece was fully free apart from the front nose leading edge, these clips where very tight and a bugger to pop open.
The wires where fairly neat down the sides of the body (unwound) but they where not inside the grey plastic loops at the base of the arm and became fairly splayed before they entered the arm.

Still not had a chance to have a test flight as we now have gail force winds in the south of england

 

[Xcel]

Here's another sample for you guys. Bought mid Jan from a local dealer, who knows how old the batch is.

 

[BudWalker]

Although I've never had the noisy back compass issue I decided to do this fix. @Logger and I had looked at some flights where a marginally noisy magZ.Back could cause a problem under the right circumstances. The geomagnetic field itself is noisy and a combination of the right geomagnetic inclination, heading and pitch angle can add enough noise to magZ.Back to cause a problem.

Rather than look just at magZ.Back I compared it to magZ.Front. In Sport mode drawing about 20 amps this is what it used to be.

The magZ.Back is about 3.6 times noisier than magZ.Front.

After twisting the back motor wiring it looks like this

Now the ratio is about 1.1.

My Mavic was delivered in late October. The photo below was taken after the left motor cables had been twisted 6 times. The right motor cables had not been twisted yet. For illustration purposes, the right motor bundle had been brought out a little so that it can be seen it wasn't twisted.


@npmullins I agree those front clips are a bugger. I found that starting from one side helped some, but it takes more force than I was comfortable with.

 

[Mr Spock]

Could someone explain the noise and any effects other then noise in detail?

You can find it all here
Firmware v01.03.0000 Compass Error.

 

[pitchmagic]

Hey guys!

I also got the switch message the two times I have flown the Mavic in sports mode. When analyzing my logs I can see that the Mag.Back.Z fluctuates alot. The wave varies from 300 to 550 when pulling +18amps and 8000rpm at the back rotors. Then it switches to the front compass. That never goes above 330. When you say "p2p" do you mean the difference in the wave at a given time or do you compare front and back mag at a given time?

 

[Mavicsailor]

P2P=peak to peak

The highest measurement in a given cycle (top of peak) - the lowest measurement in that cycle (bottom of peak) when measuring mag.back.Z. (Or any other frequency)

Given that the noise frequency is pretty high when viewing it, it's the average high-low closest to the time curser you can move around.

Hope this Helps!

 

[pitchmagic]

P2P=peak to peak

The highest measurement in a given cycle - the lowest cycle in that cycle when measuring mag.back.Z. (Or any other frequency)

Given that the noise frequency is pretty high when viewing it, it's the average high-low closest to the time curser you can move around.

Hope this Helps!

Thanks! I need to do a few clean runs to measure this better and to get a more proper log. I can tell that I am affected however.

For the ones that have twisted the wires, can anyone confirm how this is done without opening the Mavic? Can you take out the single torx at the rear arms, pull out and twist clockwise? Then reverse the process to refit?

I really want to open it and do this properly but im no too keen on loosing the guarantee.

 

[DerekG]

Is there any reason DJI can't just prioritize using the Front Compass as the primary one? It seems for almost everyone that the front one always gives the better reading with less variation

 

[pitchmagic]

Is there any reason DJI can't just prioritize using the Front Compass as the primary one? It seems for almost everyone that the front one always gives the better reading with less variation

No reason, I can bet money that they will do that for the next FW. No errors, happy customers, fewer returns = better revenue!

It could be so easy that rear is set as compass 1 and design wise you start with the first and switch to the secondary.

 

[flyNfrank]

You can find it all here
Firmware v01.03.0000 Compass Error.

@Mr Spock

Thanks, after 10 pages of reading I now understand how this thread came about. I'd have to think that the more recent Mavic production has had this issue addressed? I fly in strong winds a lot and I tend to get the Strong Wind message box which is understandable. I never get any type of compass warnings.

 

[harfordhawk]

Could someone explain just briefly, what to look for in these charts? What is good and what is bad? thanks

 

[flyNfrank]

Could someone explain just briefly, what to look for in these charts? What is good and what is bad? thanks

I suppose each person would differently. I have little experience with this, but some of what I read had decent descriptions making it easier to understand for someone researching this issue.
To answer your question partially, I went a borrowed the image below. I like it because it shows the name description by color at the bottom. If this was your DAT flight log file you were analyzing, what you would look for is the line with a sudden scribble section. I personally would refer to it as "Noise" instead of the scribble reference.

The teal and lime green lines show noticeable sections near the beginning. What would also be the same time the Compass warning messaged displayed in the upper toolbar in Flight view. What you would want to see from your A/C is the teal line to look just like the blue line seen in the below graph. The teal and blue line are both of your A/C's compasses.

The part I didn't explain was about the images in the thread that show different wire routing. I hope this helps you and any others understand what they're looking at about the issue. Others may also have a better description then what I gave as well.....

EDIT: I missed this before posting, but look at BudWalkers #45 post. It gives the details you need to look for if you're concerned about your A/C.

 

[flyNfrank]

Although I've never had the noisy back compass issue I decided to do this fix. @Logger and I had looked at some flights where a marginally noisy magZ.Back could cause a problem under the right circumstances. The geomagnetic field itself is noisy and a combination of the right geomagnetic inclination, heading and pitch angle can add enough noise to magZ.Back to cause a problem.

Rather than look just at magZ.Back I compared it to magZ.Front. In Sport mode drawing about 20 amps this is what it used to be.
View attachment 7004
The magZ.Back is about 3.6 times noisier than magZ.Front.

After twisting the back motor wiring it looks like this
View attachment 7005
Now the ratio is about 1.1.

My Mavic was delivered in late October. The photo below was taken after the left motor cables had been twisted 6 times. The right motor cables had not been twisted yet. For illustration purposes, the right motor bundle had been brought out a little so that it can be seen it wasn't twisted.
View attachment 7007

@npmullins I agree those front clips are a bugger. I found that starting from one side helped some, but it takes more force than I was comfortable with.

@BudWalker

I'm wondering if adding shielding sleeve over each set of the compass wire bundles possibly help with less noise or any upcoming related issues?

 

[BudWalker]

@BudWalker

I'm wondering if adding shielding sleeve over each set of the compass wire bundles possibly help with less noise or any upcoming related issues?

Sounds like a lot of trouble. I'm not sure it's even possible. I'd bet that the benefit of a shielding over twisting can't be measured. Ethernet cable is twisted for the same reason; actually it's to prevent interference being coupled into the ethernet wires. But, the same physics applies. The twisting will prevent the coupling.

 

[flyNfrank]

Sounds like a lot of trouble. I'm not sure it's even possible. I'd bet that the benefit of a shielding over twisting can't be measured. Ethernet cable is twisted for the same reason; actually it's to prevent interference being coupled into the ethernet wires. But, the same physics applies. The twisting will prevent the coupling.

It appeared to me that the straight wires on your setup were laying over the board and absorbing some current leaks. In this case using a wrap cover with a split seam down the middle, might help? The Mavic I have was built at the end of Nov. Looking at both MagZ.Front & MagZ.Back using CsvView, they have the appearance of not being twisted on both sides for whatever the reason. As I mentioned I haven't had any compass related warnings of any kind. So that has me wondering if they didn't program the firmware to run both compasses at the same time.

 

[BudWalker]

From post #45
For illustration purposes, the right motor bundle had been brought out a little so that it can be seen it wasn't twisted.

 

[Logger]

Is there any reason DJI can't just prioritize using the Front Compass as the primary one?

Logically there must be a reason. We just do not know what it is yet

I have yet to be sight the rear compass, whilst the front one is totally obvious and shown in the first post of this thread. There is speculation that the problematic rear compass is tucked up in the rear left of the top cover. However it is probably not a standalone unit like the front one. There is just the single "GPS" connector to the top cover. I am thinking the rear compass is part of an integrated GPS/Compass module. If this is in fact the case, there are likely sound reasons for making this the primary compass and the remote standalone compass up front one the auxiliary. Could be to do with the modules startup process or some such.

In any case, twisting the rear motor power cables seems to be a decent fix to the issue. This along with DJI's existing firmware solution (compass switching).

 

[Rich Sugden M.D.]

Something strange here ... you mention a "front" compass, and a "rear" compass ... I doubt seriously there are two compasses on the Mavik, or any device, for that matter ... unless for redundancy if one fails. No reason to have more than one.