3D printed F450 Frame - Wonky flight and GPS lock

Mongoose · Jan 29, 2017

Hi all!

Have had my FC40 for a while but it recently met with a tree at high speeds, which absolutely devastated the body and two of the motors.

So playing around, i've 3D printed a frame based on an F450 and put all of my internals into it. It's came out looking pretty cool, but when I fly it, i'm having issues with it listing around, and biasing to the rear left motor in a tilt.

If i'm on the controls, i can keep it pretty OK, but GPS lock will not work, and the entire craft "toilet bowls" around, always beginning in the same rear-left tilt. I get solid green with the GPS but i am in a valley with trees and my house in the way, so i'm not exactly trusting of it, so i usually just fly in ATTI mode-- but the odd listing / wandering behavior still persists even there- it almost looks like it's off-balance and not leveling out.

Another concern i have is where i mounted the compass module, It seemed like a good place since it fit on the GPS mast pretty well, but not sure if that's a good place to really have it. I've uploaded some pictures to show my build-- excuse the wiring for now, as i haven't really messed with finalizing it all yet.

See anything that really stands out as wrong? How can I correct this odd biasing behavior? IMU is within spec and i've calibrated it multiple times.

Really wish there was a way to "trim" it, but i know the NAZA is supposed to work all of this out automatically.

Any thoughts?

rocketflyer · Jan 29, 2017

Possible reasons?
- damaged controller IMU from the crash.
- remaining motors may also have been damaged or the replacements are nor exactly the same
- FC not tuned for the new 'body'
- having printed frames before, I would say that the top/bottom plates won't be as rigid as they should be (even if printed at 100%) and would flex in flight

FunN4lo · Jan 29, 2017

The F450 is a fun machine. The NAZA units that you would normally buy for the 450 have the compass/gps that sits on a tall stalk. The toilet bowl effect (TBE) is cased by the alignment of that unit. You turn that unit a few degrees one way of the other to correct the TBE. Your pics are hard to make heads or tails of, but adjustment should correct the TBE.

I agree with rocket flier. The flexiness of the 3d printed frame may be part of the issue also

Mongoose · Feb 10, 2017

FunN4lo said:
The F450 is a fun machine. The NAZA units that you would normally buy for the 450 have the compass/gps that sits on a tall stalk. The toilet bowl effect (TBE) is cased by the alignment of that unit. You turn that unit a few degrees one way of the other to correct the TBE. Your pics are hard to make heads or tails of, but adjustment should correct the TBE.

I agree with rocket flier. The flexiness of the 3d printed frame may be part of the issue also

View attachment 74821

Sorry for late reply- so the alignment of the GPS disc, or the compass module?

Finally got another day to play around with it, and turned the GPS disc-- seemed to have some kinda effect, but i couldn't really be too sure of how finely i needed to turn it. Looking at how i did the print, the "stalk" is just barely turned, and i can't remember how they went onto the FC40. Another concern is that the FC40 had some shielding around the area.

I flew it in ATTI mode, and did just fine, but i guess i do kinda think i'm only getting half the machine if the GPS isn't working right-- but i'm doing a lot of action filming with it so most of the time, i don't need it to hover or stay in lock.

Mongoose · Feb 10, 2017

rocketflyer said:
Possible reasons?
- damaged controller IMU from the crash.
- remaining motors may also have been damaged or the replacements are nor exactly the same
- FC not tuned for the new 'body'
- having printed frames before, I would say that the top/bottom plates won't be as rigid as they should be (even if printed at 100%) and would flex in flight

Turning the FC..? Is this setting up your basic gains and all that?

FunN4lo · Feb 10, 2017

Mongoose said:
Sorry for late reply- so the alignment of the GPS disc, or the compass module?

Finally got another day to play around with it, and turned the GPS disc-- seemed to have some kinda effect, but i couldn't really be too sure of how finely i needed to turn it. Looking at how i did the print, the "stalk" is just barely turned, and i can't remember how they went onto the FC40. Another concern is that the FC40 had some shielding around the area.

I flew it in ATTI mode, and did just fine, but i guess i do kinda think i'm only getting half the machine if the GPS isn't working right-- but i'm doing a lot of action filming with it so most of the time, i don't need it to hover or stay in lock.

In the NAZA M combo the gps/compass was all in the same disc. But the rotation was for the compass. There is an arrow on the disc that points forward. Then you turn it right/left off of center for the magnetic declination where you fly. If you were off in your setting, you would get the TBE. Then you needed more/less deviation off of center to correct. I don't think there was any user adjustment in the FC40, but you may not have it aligned correctly in your conversion

Mongoose · Feb 12, 2017

FunN4lo said:
In the NAZA M combo the gps/compass was all in the same disc. But the rotation was for the compass. There is an arrow on the disc that points forward. Then you turn it right/left off of center for the magnetic declination where you fly. If you were off in your setting, you would get the TBE. Then you needed more/less deviation off of center to correct. I don't think there was any user adjustment in the FC40, but you may not have it aligned correctly in your conversion

Now, is there a way to really fine-tune it, say-- in NAZA assistant; some sort of value output that you can verify? Eyeballing it seems sort of hard to really detect minute changes.

FunN4lo · Feb 12, 2017

Mongoose said:
Now, is there a way to really fine-tune it, say-- in NAZA assistant; some sort of value output that you can verify? Eyeballing it seems sort of hard to really detect minute changes.

Are you talking about gains settings?

FunN4lo · Feb 12, 2017

I start here. Most of what I built would fly good with these settings and didn't need tweaking to far off of these settings. But every aircraft is different.

The Atittude gains control how the aircraft reacts to stick movements. The lower gains are like adding expo to a traditional transmitter. When gains are high it overreacts when the gains are low it underreacts.
Basic gains are how your multirotor levels itself, and the attitude gains are like stick scaling.

In the Naza software there's a loop, in short, there's two sets of values :
1 : how the multirotor really is, in position and angle
2 : how the multirotor should be, according to your stick inputs.

Basics gains controls how hard the multirotor will try to reach 2 from 1, by itself.
If the gain is too high, it's trying too hard, and will overshoot 2, so it comes back, once again too hard, overshoots again, and it's oscillating.
If the gain is too low it's slow to move from 1 to 2.

Attitude gains is how quickly Naza is reacting your stick movements. High gains it will react fast low gains it will react slow.
Your propellers size and pitch and also the weight of the aircraft will also affect how you should set your gains. Also the altitude where you fly from.

Mongoose · Feb 13, 2017

FunN4lo said:
Are you talking about gains settings?

Naw- was talking about the GPS orientation, to make sure it really is pointing where it should and to some degree of accuracy. Curious if there's some number that gives some inclination of where it's pointing.

Though in terms of gains, I think my current frame is lighter than the FC40, and i'm mostly flying at around 800-1000' ASL. I futzed around with them but it's a little too twitchy currently. Need a good day of sitting outside with a laptop.