Hi guys,
As way of contributing to the development of AC, I have decided to try and document possible solutions to the compassmot problems I have, and their effects on compassmot.
In summary - I have a compassmot score of 148% (without modifications), and it toilet bowled. For those unfamiliar with this expression, it is when loiter is engaged, the MR starts to circle in an ever increasing circle (while maintaining the same heading) until a loss of control event. This is caused by magnetic interference on the compass. I won't cover this in too much detail, as there are threads dedicated to this.
The most effective method is to move the APM away from the source of the interference - most likely the PDB and ESC's directly below it. However, there are reasons why this is not a practical solution.
For me, my reasons are:
1. Safety features.
I want to keep my protective hardware to protect the electronics in the event of a flip. This is a clear dome, and a FG cage underneath (you can see the cage in the pic above). If I raise the APM board, I will need to raise these too. The cage I could probably do with 6-7cm risers but I've not found any. I can get them in metal, but that will cause more mag problems! But the dome needs to stay where it is, forming an effective seal when velcro'ed down to the top plate. It's very effective at dust and grass exclusion, and I don't want to lose that.
2. Cost.
To raise the APM board 2cm I need to purchase the following (admittedly not expensive) parts:
- 8 new male-male servo cables, 5cm longer than the current ones
- 4 x 3cm rubber risers, or 4 x conventional plastic ones. Obviously I want to preserve my vibe damping so I want those…but they're like hens teeth
- 4 x servo extension for the motor servo leads
- a 5cm longer DF13 for the telemetry
- a 5cm longer DF13 for the power module feed
- a 5cm longer DF13 for the GPS
- 4 x 6cm risers for the cage (assuming I want to sacrifice the protective qualities of the dome)
- 8 x plastic screws to replace the metal ones close to the APM
- I already have the mumetal and ferrous rings but they're still a cost
So what to do?
There are 3 alternative options to airframe modification. Two of which I will try and detail here - I have already covered the effects of the first one (addition of mumetal and ferrous rings here: http://www.diydrones.com/profiles/blogs/how-i-got-from-148-to-43-on-compassmot). I will try to document via pictures, video and log data their respective performance improvements, and where I got the components from.
Option one - addition of 3rd party external mag to existing APM 2.5+
For this I have purchased:
a) the external compass: http://www.goodluckbuy.com/hmc5883l-triple-axis-compass-magnetometer-sensor-module-for-arduino-mwc-m49.html
b) a DJI-style riser: http://www.goodluckbuy.com/bracket-holder-set-for-dji-wkm-gps-autopilot-system.html
c) a 30cm or 45cm DF13 4 wire. Except...I can't find any! Only place is 3DR...and they want to charge me $38 for shipping *one* cable! If anyone can help with the sensible purchase or construction of this cable, I'd be most grateful, and would happily cover your costs.
Option two - addition of new 3DR external GPS/compass to 2.5+.
In all cases, the "baseline" airframe will be the one that flew perfectly on 2.9.1. ie:
quadframe heavy: (http://quadframe.com/collections/quadcopter-frames/products/heavy-weight-quadcopter-frame-with-21-5mm-carbon-arms). The one useful feature is the external DJI pylons...perfect for mounting external, mag sensitive stuff away from the main components.
quadframe 155mm legs
quadframe DJI board (for mounting APM on)
quadframe PDB
Tmotor 3508-29's 380kv
Tmotor (clones) 1455 CF props
Tmotor 30A OPTO PRO ESC's
Spektrum Rx + telem
APM 2.5+, incl ublox GPS, power module, 433 telem
Flight pack is either a 10900 6S or 2 x 4000 6S. Electronic pack attached to 3DR power module is 890 3S.
UAW: 3.8kg
Stay tuned - I'll post the first results in this thread when the parts arrive in a few days time.
Comments
I have a longer frame with two center plates. The back one contains the Power Distsribution Board and the hind legs are where all the ESCs are attached (six in this case). But as you may be able to deduct from the image (http://www.diydrones.com/photo/img-1661) the center arms could also be removed for a quad frame. The front center plate houses the APM 2.5 and I have the GPS in between. This layout means there is protection of the onboard equipment (except GPS which needs still some housing) and there should be less magnetic interference due to the greater distance between PDB, ESC and APM. I have never had toiletbowling effect, but do ensure to always recalibrate the compass before flying just as a precaution.
I will then be as scientific as possible in testing each one. I hope, by the end, that fellow sufferers will have a good idea about what options they might need, based on their starting compassmot scores.
Ok, something went horribly wrong in tonights baseline test on 2.9.1. After loitering and RTL'ing like the old days, it did a toilet bowl then decided to do a very aggressive RTL landing (almost as if it thought it was higher than it should be), and tipped over. After than, vibes went up, yaw and roll were off, it refused to take off smoothly, tipping over twice again, and when I stopped, motors 1 and 2 were noticably warmer than 3 & 4.
Naturally, I suspect an arm has moved when tipping over. But the toilet bowl is a strange one. Works perfectly well at the beginning of the pack, but not from the middle. And it's doubly strange because the airframe is right back to it's 2.9.1 config to the last bolt and screw, and where it would loiter and RTL for an entire pack.
The tale of woe in the logs is here.
2013-07-08%2021-11%201.log
2013-07-08%2020-33-00.tlog