Compassmot with external gps/compass, and e300/phantom props

Hi

I’ve built a quadcopter with a pixhawk, 3dr gps/compass module and 3dr telemetry on a F450 frame with E300 propulsion system (it’s a tuned/matched set of DJI motors,esc and props). The props are the standard DJI self-tightening props, so I can’t turn them upside down. How can I do the compassmot with these? Is it possible?

Is it necessary to do the compassmot if I have the external 3dr gps/compass mounted up on a stick mount? Or can I safely skip it?

I have made about 15 flights now, all have ended in disaster because my brain clearly can’t cope with the complexity of flying a quadcopter! Aside from single handedly contributing to the share prices of companies that supply spares - I’ve literally lost count of the amount of arms and props i’ve already been through - I did manage to get enough flight to see that althold is pretty good, and I got a single loiter flight which was pretty rock solid hover.

I’m having all sorts of other problems with althold/loiter/land modes which are contributing to crashes, but I think that’s still setup issues that I’m working through.

The only way I can help you with these questions and problems is if I have a log file. Especially one of a crash.

Can you describe one of your crashes?

You can do compassmot by restraining the copter. I recommend doing it with mid throttle, not at full throttle.

Thanks for the reply. I’ve rebuilt everything from scratch, reset and reconfigured the pixhawk, recalibrated everything, remounted the GPS on a mast and have got much, much better results. I can now fly it in stabilise mode with reasonable success and control (although my dogs have learnt to run for cover whenever the arming tone sounds), alt-hold works perfectly, and loiter is absolutely rock solid. RTL and landing even work perfectly and almost always lands in the 2ft x 2ft square takeoff area I cut in my long grass field which is amazing! And I’ve done a bunch of successful auto flights which is just awesome. I’ve had a couple of crashes testing throttle failsafe and auto mode which I’ll post separately.

For compassmot though, can it be done without props? If it’s for measuring the effect of EMI from ESCs/motors etc on the compass, does it make any difference if the props are on or not? I can try and tie it down somehow but it would be an awful lot safer to just take the props off. I’m interested from a practical and technical view.

Compassmot need to be done with the props on because your motor’s magnetic field will increase with the load the motor sees. A motor without any prop has just a very little load and thus no strong magnetic field.

I know the props you are using and the only way to execute a compassmot for you is to tie your quad down to something heavy. When I say “tie down” I mean tie down, the quad must have no possibility to take off. The object/ground etc. you are tying the quad to needs to be heavy enough (at least 10 times the maximum thrust of your engines) to ensure it will not move.

I have seen videos where people used large wood plates with holes in it to tie down the quad with cable ties and large stones on the wood plate.

Attention: Perform the compassmot only over the air and stay away from the quad! Cable ties are also no secure way to tie quad down. Use something that cannot open itself alone. And stay away when doing compassmot!!! And the last rule: Stay away!!! God damnit! :slight_smile:

I don’t like this procedure either but it increases the accuracy of the inertial navigation system tremendously.

Regards!

Sven

Why not simple swap props between motors to push air up instead of down and then measure compassmot?

Yes, you can flip props over. They have to be flipped and swapped, I believe. Like I said, I just restrain the copter. Zipties are plenty strong.

If you have a current sensor, it does not matter how the props are configured, as long as all the motors are drawing equal current.

I am trying to get compassmot learning done for 3.2. Compassmot is the most onerous part of our setup procedure, and I’d like to eliminate it. A side effect will be increasing accuracy - changing batteries, differences in current distribution between a hover and the test (which just sets all motors equal), etc. Compassmot learning will be able to correct a pretty large portion of the error on takeoff.

The DJI props come with a built-in tightening nut, if you like, which is molded into the prop and can’t be removed. So you can’t flip the props upside down, hence my original query.

Thanks SvenL for the suggestions. I’ll try and work out a safe way of tying it down with out decapitating myself or the local rabbits :slight_smile:

jschall: that would be amazing, looking forward to automated facility.

I’m still wondering - does compassmot really need to be done if you are using the external gps/compass module that sits six inches or so above the ESCs/motors? I haven’t rushed to do this because my quad with pixhawk and external gps/compass is pretty solid at hover/loiter/auto flights. I’m still new to all this and learning, and improving it all the time so there is definitely room for improvement, but i’m so impressed with it so far.

jschall: do you mean that if you change battery the compassmot has to be redone? I like to swap between 3s and 4s for different flying types/reasons.

The interference comes from current-carrying conductors. Some portion of those conductors are in and on the battery. Depending on how close those conductors are in comparison to other conductors on the copter, it stands to reason that moving, rotating, or swapping out the battery can change the compassmot calibration vector.

Compassmot does not need to be done when conductors are far away, though it can still make a difference. For example, I’ve done it on my y6, which has 90mm to the nearest ESC power wire (which is twisted). If you can send me a flight log with MAG enabled, I can tell you how much interference you have.

last time i did a compassmot it wrote NAN values in my param, and from that point on my artificial horizon went mental… had to do a erase eeprom and load all my params (minus compassmot) again… :frowning: was hoping this will be fixed in the 3.1.4 release of the firmware

@fluffyflyer: Perhaps a sim such as Phoenix or similar would get you more “familiar” without the stress and the crashes. In any case it can’t harm your aquired skills.
Just a thought… Learning to fly on models isn’t the best as too many parameters have to be dealt with…
Cheers
Henri

[quote=“fluffyflyer”]Hi

I’ve built a quadcopter with a pixhawk, 3dr gps/compass module and 3dr telemetry on a F450 frame with E300 propulsion system (it’s a tuned/matched set of DJI motors,esc and props). The props are the standard DJI self-tightening props, so I can’t turn them upside down. How can I do the compassmot with these? Is it possible?

Is it necessary to do the compassmot if I have the external 3dr gps/compass mounted up on a stick mount? Or can I safely skip it?

I have made about 15 flights now, all have ended in disaster because my brain clearly can’t cope with the complexity of flying a quadcopter! Aside from single handedly contributing to the share prices of companies that supply spares - I’ve literally lost count of the amount of arms and props i’ve already been through - I did manage to get enough flight to see that althold is pretty good, and I got a single loiter flight which was pretty rock solid hover.

I’m having all sorts of other problems with althold/loiter/land modes which are contributing to crashes, but I think that’s still setup issues that I’m working through.[/quote]