MedicineMan4004
Well-Known Member
'"Compass report wrong data, don't take off, without checking"' Taking this to heart. Great advice.
DJI Spark Compass Calibration
- Thread starter JoeCec
- Start date
Just got around to viewing this YouTube. Unfortunately, it promotes the misconception that the compass calibration detects and then compensates for magnetic effects external to the AC. This is not correct. In fact, it's mathematically impossible for a calibration to detect and compensate for magnetic effects external to the AC. Whether it's declination, deviation, or whatever it's labelled, if it's external to the AC then it can't be compensated for by a calibration.
This misconception has been around for a long time. I suspect it probably stems from this thread
Compass Calibration, A Complete Primer
Lately, I've been attempting to address this starting about here
Compass Calibration, A Complete Primer
In addition, the video doesn't explain what a calibration actually does. I wrote this sometime back
The purpose of the "dance" is to expose each of the three magnetometers to a wide range of magnetic values. After this the "gain" is set for each of the three magnetometers so that there will be a uniform response to the geomagnetic field. To determine what the gain needs to be each of the magnetometers needs to be exposed to a wide range of magnetic values.
After rotating 360 degrees in the XY plane the X and Y magnetometers have both seen positive and negative values. By orientating the AC so that the Z axis is now perpendicular to it's original orientation and then rotating the AC 360 degrees the Z axis sees the positive and negative values. The usual nose down (X axis down) brings the Z axis perpendicular to it's original orientation. But, so does side down (Y axis down).
The "dance" is just a procedure to insure that the three magnetometers have each been exposed to a wide range of values. It would work equally well if the AC were "tumbled" provided it was placed in enough different orientations. I once did an experiment where the calibration was done upside down. I.e., first rotation was done top side down and the second rotation was done nose up. Didn't make any difference.
Nice video, thanks. And good feedback in the post above too. I was curious to find the compass module in the Spark but wasn't sure where it is. Since you suggested it's part of the IMU assembly I've attached is a pic of this chip with the thermal paste removed, which looks like a 6-axis MPU-6500, so perhaps it's not located there? I do believe it's near the front of the drone somewhere though because unlike the rest of the drone the screws at the top of the 3D vision sensor assembly were non-magnetic. I might have a closer look later on.
Attachments
You could observe the compass interference display in the Go App while probing various parts of the Spark with a screwdriver.
It's on the 3D vision sensor circuit board at the front, behind the camera. I confirmed it by disconnecting the flex connector cable to this unit, which gave me a "compass disconnected" error (see attached). I'm a bit surprised DJI didn't try and better isolate the compass from other electronic circuitry in order to minimise electromagnetic interference, which they did with previous models (Mavic, P3, P4 etc). I suppose the Spark presented an engineering challenge to fit everything into a small footprint.
Attachments
I'm a new Spark owner. Does it matter in which direction (counterclockwise or clockwise) you rotate the spark when you do the compass dance.
Thanks,
Bobby
Thanks,
Bobby
DesertWindAero
Well-Known Member
- Join
- May 28, 2017
- Messages
- 526
- Age
- 74
Thanks for your reply. I tried to fly it yesterday, but kept getting Compass Calibration errors. I think the problem was my prosthesis. I'm a right leg, above the knee amputee, and figured the problem was that I was wearing my computerized prosthetic leg (I have to charge it every night.)
So, the next time I try it, I'm gonna be sitting in my wheelchair, without my prosthesis on.
DesertWindAero
Well-Known Member
- Join
- May 28, 2017
- Messages
- 526
- Age
- 74
Yes, that occurred to me. So today, I did the compass calibration in the kitchen of my home, in my wheelchair, and it worked (I got a check mark). But darn it, I can't fly it at home due to me living in a No-Fly Zone (Columbus airport and Ft Benning), so, I'll take it up to my hunting property, It's outside the no-fly zone. That's where I tried to fly it for the first time. Hopefully, it'll work out next time.
Last edited:
The biggest sources of magnetic interference are high amp moving electricity. This means battery and power cables. In the Spark, the main power distribution board is near the rear, as is the battery, so having the compass up front makes sense. The vision sensors wouldn't be a significant amount of interference, and even then it would be constant, so you can program around it. Usually the gimbal motors would be somewhat noisy, but I suspect they are so small it doesn't matter much.
I calibrate the compass before every flight. Also ditch any metallic objects like watch, rings and stuff. I even take my Leatherman of me. Any metal object within 1.5m will affect your compass.
Don't recalibrate every flight, if is asking for trouble. Just do a calibration in a known good area, then dont touch it.
Read the guidlines DJI has set up, if you have a good calibration no need to redo unless you fit the criteria specified by DJI. Otherwise you might introduce errors from calibration at a bad site.
Lots of people say this, but I've never seen evidence that calibrating the compass before every flight can introduce errors (in any DJI drone). If either of you know of a case we can take a look at, then please link it here.
Spark user manual indicates not to calibrate unless prompted.
If you calibrate near ground level interference, it may affect flight results at altitude.
The debate stems from those like myself who used a Phantom, which advised calibration if you travelled any great distance. This seems to no longer be an issue.
Preflight, just check your compass for interference, and if very low (there will always be some interference) then fly safe.
Many fly aways have been caused by people taking off near magnetic metal objects. I've never seen one caused by someone calibrating the compass near a magnetic metal object.
Have you ever tried that? I have. The calibration always fails after the second rotation. Furthermore, if one knows not to attempt to calibrate near a magnetic metal object, then that makes it ever harder to accomplish such a thing.
Yep, I've seen this. Have you ever seen the DJI GO app prompt you to calibrate the compass? I haven't. Nor have I seen anyone else mention it has ever prompted them to calibrate. So, I'm not sure it's wise to follow that advise.
In this article from December 21st, DJI recommends pilots should calibrate the compass before every flight because it's essential for safety. Again, I don't think it's necessary in most cases. I'm just pointing it out because DJI does not always recommend what you see in the current Spark manual.
You're right but it's more than that. In the past year or so there have been several threads in this and the other forums where it's been claimed that a flawed calibration can cause a fly away. Members have been asked (or challenged) to come up with just one incident where there was at least some evidence to support the claim that a flawed calibration caused a fly away. So far not one incident has been presented. And I think they tried really hard hoping to prove me wrong.
@msinger is right. A calibration will be rejected if the geomagnetic field is distorted within the confines of the compass dance. In this case the "interference detected, move location" message will be issued. A calibration can only detect and compensate for distortions that rotate with the AC. External distortions may be detectable but the data is ambiguous and, therefore, can not be compensated for. In fact, this ambiguity is the criteria by which the calibration is rejected.
Part of the reason, IMHO, that the flawed-calibration-causes-fly-away misconception exists is that it's easy to confuse it with the launch-from-geomagnetically-distorted-site-cause-fly-away. The latter has happened many times. And, usually without any error or warning being issued. No amount of calibration, in any location, or not calibrating, will have an effect on this type of fly away. The single most effective way to prevent these fly aways is to check before launch that the red triangle heading indicator on the map display is consistent with the AC's actual orientation.
Similar threads
