DIY rear compass cabling fix

Starz · Mar 3, 2017

Maybe if more owners would help out and go to the other thread listed in post # 128, and provide serial numbers, more solid info could be a benefit to others.

Logger · Mar 3, 2017

bernie124 said:
My non twist, that I've got DEC 7 has a serial with 08QD DB2.

So about a week later than @Randy Nov 2 build. So the switch to twist is likely in the DB3 - DC1 somewhere. Just need to narrow it down a bit more.

Logger · Mar 3, 2017

Starz said:
... Sounds like my late December order date may put mine in the twisted group. Does that sound correct?

More than likely. Whats 3 characters on your serial no after 08QD?

bernie124 · Mar 3, 2017

Yep, I know it takes longer to get to Germany.

Starz · Mar 3, 2017

Logger said:
More than likely. Whats 3 characters on your serial no after 08QD?

DCV listed in the other thread. Per your chart, should be 16 model December 30. Ordered 12/27/16 ship notice date of 1/12/17

Randy · Mar 3, 2017

Logger said:
So about a week later than @Randy Nov 2 build. So the switch to twist is likely in the DB3 - DC1 somewhere. Just need to narrow it down a bit more.

I've flown it 3 times since the twist, no problems, GPS stays in the high numbers and stays green, these are short flights though because it's still pretty cold here and windy too, first semi-warm day (20 or above) I'll run it through the paces and keep you posted.

Mr Spock · Mar 4, 2017

I could not resist... I opened it.
Here it is mine: SN 08QDDAF0xxxxxx
Mavic Pro combo bought 21 Nov 2016 from the UK.
They look a bit twisted to me what do you think?
Never had a GPS warning issue but i flew in Sport M. Only 2 times and i have my Gains, Yaw etc down to around 85% so it could be that. I have not yet used BudWalker's magical SW to see that the Zmag front and back values look like P2P. Any way i am tempted to break the black Goop stuff, undo the 2 screws at each rear leg and twist some more now that the bird is open.
By the way for those that wonder there are 12 screws in total involved here, 6 on top and 4+2 at the bottom in the front side you have to undo the leg cover that are hold with a screw each and underneath there is one screw at each side.

Mr Spock · Mar 4, 2017

Update: Definitely not twisted...

Mr Spock · Mar 4, 2017

And done.
I have also replaced the black Goop with a little bit of transparent Goop i had.
Also noticed that my rear compass/GPS module has an induction ring in the cabling to avoid EMI.
Legs are hold with 3 screws each, 2 inside and one outside.

Strafe1 · Mar 4, 2017

nice work! I thought that after twisting the wires, it was best to reroute the wires OVER the leg housing and skip that first "hook". Not so?

Mr Spock · Mar 4, 2017

Strafe1 said:
nice work! I thought that after twisting the wires, it was best to reroute the wires OVER the leg housing and skip that first "hook". Not so?

Thanks. I don't see why is that, i just put back everything the way it was and also there was a lot of wire slack as before so i put them in the hocks. They are not hocks really more like forks.

Logger · Mar 4, 2017

Mr Spock said:
I could not resist... I opened it.
Here it is mine: SN 08QDDAF0xxxxxx
Mavic Pro combo bought 21 Nov 2016 from the UK.

Good stuff Mr Spock. An excellent how to guide for others. Your serial snippet is by the way, the perfect portion for us to use to work out when the change occurred. Those first 8 characters (actually only 5,6,7 & 8) tell us what we need to know I think. If anyone wants to help out, please answer the 5 Questions over at the serial no thread

Mr Spock · Mar 4, 2017

I will fly tomorrow and test the wire twister mod. I managed to get one of my Sport Mode DAT files where i did around 70km/h and thanks to @BudWalker's CSV SW tool i was able to compare the magZfront and magZback values and yes i saw that i had large p2p deltas when i was in motor high speed. I just managed to use the assistant 2 and the CSV tool for first time today so apologies for the print screens and i did not know how to optimise the CSV display to get the values on the side and to scroll left and right on the graph (maybe one of the more clever fellas here can tell me at some point). So i got Magz back high peak 328 low peak 163
Mag z front high peak 199 low peak 186. The difference on magZ back is very large =165 p2p where as magZ front is only = 13 p2p.
I am now curious to see what tomorrow's Sport Mode test run with the wire twist mode will show.
By the way i am still on FW 1.3.0200 intentionally but i intent to give 1.3.0500 a go after the test run tomorrow.

Mr Spock · Mar 4, 2017

Logger said:
Good stuff Mr Spock. An excellent how to guide for others. Your serial snippet is by the way, the perfect portion for us to use to work out when the change occurred. Those first 8 characters (actually only 5,6,7 & 8) tell us what we need to know I think. If anyone wants to help out, please answer the 5 Questions over at the serial no thread

Ok Logger will put the data up on that thread. Sure.

agilealtitude · Mar 7, 2017

Just wondering if this fix has NOT solved compass redundancy switch / sport mode GPS failures for anyone?

Randy · Mar 7, 2017

agilealtitude said:
Just wondering if this fix has NOT solved compass redundancy switch / sport mode GPS failures for anyone?

I've flown mine 4 times since the twist, and I believe it would have lost GPS under the conditions I have flown it in since, really too early to tell yet because it's been brutally cold here since, but there's a reason why DJI started twisting them.

BudWalker · Mar 8, 2017

Randy said:
I've flown mine 4 times since the twist, and I believe it would have lost GPS under the conditions I have flown it in since, really too early to tell yet because it's been brutally cold here since, but there's a reason why DJI started twisting them.

Just to be clear are you referring to the situation where noise on the back compass causes problems, sometimes a switch to ATTI mode? Or, are you referring to the situation where gpsHealth-drops-from-5-to-1 and that causes a switch to ATTI? The reason that I'm asking is that twisting the wires fixes the first problem. AFAIK untwisted wires aren't the cause of the gpsHealth-drops-from-5-to-1 issue. It'd be very significant if your data suggests otherwise.

Randy · Mar 8, 2017

BudWalker said:
Just to be clear are you referring to the situation where noise on the back compass causes problems, sometimes a switch to ATTI mode? Or, are you referring to the situation where gpsHealth-drops-from-5-to-1 and that causes a switch to ATTI? The reason that I'm asking is that twisting the wires fixes the first problem. AFAIK untwisted wires aren't the cause of the gpsHealth-drops-from-5-to-1 issue. It'd be very significant if your data suggests otherwise.

I thought it was determined that the GPS and compass are integrated, that if the compass was affected by the untwisted wires and high magnetic field when in sport mode, it affected the GPS health as well and would switch to ATTI mode. Maybe I'm deceiving myself, but I thought they were related. Most people on this forum claim that when switching to sport mode, they're MP would switch to ATTI mode due to compass, and therefore loss of GPS even though they're still showing up to 17 satellites. Somebody correct me if I'm wrong, maybe my problem is going away after all.

DroneTone · Mar 8, 2017

I think the wire twisting simply fixes compass error issues with the Rear compass.
With untwisted rear motor wires, the rear compass at times had enough error,
(electrically induced interference from the rear motor wires, usually in Sport Mode during times of high current loads),
that a message would appear in DJI GO saying the secondary Front compass was being used as the primary compass.
The Rear compass is integrated with the GPS module and acts as the primary compass.
In the event, it becomes unusable, the Mavic uses the Front compass as a backup and for comparison.
This meant that at times of high current load in Sport Mode, the Mavic might be operating with only one useable compass.
Twisting the rear motor wires is simply a solution to prevent that situation from occurring.
I suppose the GPS Health issue could be related if the rear compass had a large amount of error and was still been used as the primary compass,
but the Mavic should default to the Front compass when large errors occur with the Rear compass.

BudWalker · Mar 8, 2017

Randy said:
I thought it was determined that the GPS and compass are integrated, that if the compass was affected by the untwisted wires and high magnetic field when in sport mode, it affected the GPS health as well and would switch to ATTI mode. Maybe I'm deceiving myself, but I thought they were related. Most people on this forum claim that when switching to sport mode, they're MP would switch to ATTI mode due to compass, and therefore loss of GPS even though they're still showing up to 17 satellites. Somebody correct me if I'm wrong, maybe my problem is going away after all.

Unless I've missed it there are no incidents or data to suggest that the rear compass noise caused by high currents in the rear motor wires is related to the gpsHealth-drops-from-1-to-1 issue. That's not to say there isn't a relationship, just saying that, so far, there has been no data to suggest they are related. That's why I'm interested to know if you have data to suggest they are related. In particular, do you have the .txt logs from flights before twisting the wires that had a gpsHealth-drops-from-1-to-1 incident?