48 replies to this topic

[FredericG]

In my plane it would be good if I could mount the CC upside-down. In the "attitude-rotation" in the config gadget, I can set roll to 180 degrees, but I don't think this works...

Frederic

[peabody124]

I'll try and check that soon since I've never flown with it before.

So if you set roll to 180 degrees do the sensor change at all (specifically accels)

[dankers]

This is a feature we really need for v1, testing the rotation is important. This is how we + mode of course and also people will want to mount CC side on in a heli as well.

[peabody124]

dankers, on 01 May 2011 - 03:35 PM, said:

This is a feature we really need for v1, testing the rotation is important. This is how we + mode of course and also people will want to mount CC side on in a heli as well.

More specifically - this is how you define "forward and level" to not be coming out the center and front of the board.  You can do plus mode by mounting the flat edge forward too and it will take a bit less CPU.

[Kenn Sebesta]

Is there a reason why not to do an automatic rotation/transformation in the INS calibration gadget? If it's just a lack of manpower, I've already got code that does this. It requires only a small amount of user intervention, mostly to help it define which axis is "forward". That can either be done manually, or by tilting the vehicle.

[jes1111]

Kenn Sebesta, on 02 May 2011 - 11:47 AM, said:

Is there a reason why not to do an automatic rotation/transformation in the INS calibration gadget? If it's just a lack of manpower, I've already got code that does this. It requires only a small amount of user intervention, mostly to help it define which axis is "forward". That can either be done manually, or by tilting the vehicle.

This would be a marvellous feature - the ability to mount the board in any orientation would allow useful variations in multirotor frame design (and greater freedom in fixed wing and helicopter board positioning). Confining the allowable offset to  0/90/180/270 on each axis would make it simple to configure.

EDIT:
And, whilst we're in there, I'd still like to see configurable X/Y/Z offset from CoG as this would allow airframe design where the CoG point is in free space, i.e. not in a "mountable position". I agree with James' earlier posted calculation on angular error for a small offset, but allowing bigger X,Y,Z deviations would be cool.  

Edited by jes1111, 02 May 2011 - 12:10 PM.

[Kenn Sebesta]

jes1111, on 02 May 2011 - 12:02 PM, said:

This would be a marvellous feature - the ability to mount the board in any orientation would allow useful variations in multirotor frame design (and greater freedom in fixed wing and helicopter board positioning). Confining the allowable offset to  0/90/180/270 on each axis would make it simple to configure.

I don't believe there's any reason to confine it. The automatic configuration doesn't care, as it's just a rotation matrix defined by two orthogonal vectors, the third being the cross-product of the first two. It really comes down to leaving it level, and then tilting it on the forward axis. If this forward axis can't be well defined, then it could be tilted on the side axis.

[jes1111]

Kenn Sebesta, on 02 May 2011 - 12:05 PM, said:

I don't believe there's any reason to confine it. The automatic configuration doesn't care, as it's just a rotation matrix defined by two orthogonal vectors, the third being the cross-product of the first two. It really comes down to leaving it level, and then tilting it on the forward axis. If this forward axis can't be well defined, then it could be tilted on the side axis.

In theory, I agree - in practise it would be hard to tip the craft in its intended pitch axis without introducing any roll element.

[Kenn Sebesta]

jes1111, on 02 May 2011 - 12:16 PM, said:

In theory, I agree - in practise it would be hard to tip the craft in its intended pitch axis without introducing any roll element.

I wouldn't think any harder than mounting it perfectly flat, either.

Another possibility is just putting the vehicle on a slope, and then turning it  90 degrees and measuring a second time. That's something that should be very easy to do reliably.

[jes1111]

Kenn Sebesta, on 02 May 2011 - 12:45 PM, said:

I wouldn't think any harder than mounting it perfectly flat, either.

That's what spirit levels are for  

[jes1111]

Just another thought: allowing this freedom of mounting orientation would also facilitate using CC as a dedicated gimbal controller.

[Aerhead]

Jes

I have built and have been flying a ruggedized and much less dangerous then most multicopters, 24 motor Mikrokopter.  The flight controller is mounted offset about 20% off the CoG of the crafts radius. The offset does give the craft a little strange behavior but is still very controllable. The next multicopter I want to build would have two fight controllers each mounted offset either side of the CoG. With two receivers, at least eight motors and two batteries, true redundancy would be possible. The two sets of motors would fight each other a little if there isn't a way to adjust for CoG offset. Just pointing out one of the many ways flight controllers will probably be used.

Larry  

jes1111, on 02 May 2011 - 12:02 PM, said:

This would be a marvellous feature - the ability to mount the board in any orientation would allow useful variations in multirotor frame design (and greater freedom in fixed wing and helicopter board positioning). Confining the allowable offset to  0/90/180/270 on each axis would make it simple to configure.

EDIT:
And, whilst we're in there, I'd still like to see configurable X/Y/Z offset from CoG as this would allow airframe design where the CoG point is in free space, i.e. not in a "mountable position". I agree with James' earlier posted calculation on angular error for small offset, but allowing bigger X,Y,Z deviations would be cool.  

Edited by Aerhead, 02 May 2011 - 01:33 PM.

[peabody124]

Just to be clear, the current flight firmware (CC, I'll add this to INS later) supports any arbitrary angle.  In the attitude gadget right now you specify the rotation of the board in yaw, roll and pitch.

This hasn't been tested rigorously though so people should take care and Fred says it doesn't work.  Ivan was using it last night to go from + to X without remounting board and I believe that worked.

I'll be able to test it again when get new props.

[jes1111]

peabody124, on 02 May 2011 - 02:12 PM, said:

I'll be able to test it again when get new props.

Been pruning trees again, have we?  

[jes1111]

peabody124, on 02 May 2011 - 02:12 PM, said:

Just to be clear, the current flight firmware (CC, I'll add this to INS later) supports any arbitrary angle.  In the attitude gadget right now you specify the rotation of the board in yaw, roll and pitch.

This hasn't been tested rigorously though so people should take care and Fred says it doesn't work.  Ivan was using it last night to go from + to X without remounting board and I believe that worked.

I'll be able to test it again when get new props.

Trying this out (simply by holding the CC board in my hand) the board still assumes that pitch is in line with the arrow and roll is perpendicular to it, i.e. it works for certain "misorientations" of the board (like Kenn's simple inversion), but not others. If I understand correctly what's going on, there's no way the current config gadget can do the job, since there's no way to tell it which direction is "forwards" relative to the board.

[FredericG]

I must add that, after a short test on the bench, I had the impression it did not work. I was unsure if setting the roll to 180 was the way to go.

I was also wondering if I needed to perform the bias trim before or after I had put in the 180 degrees and flipped the CC. I tried several options and it did not seem to work out. This is when I decided to ask. I am not 100% sure that it does not work.

Fred

[jes1111]

FredericG, on 02 May 2011 - 03:59 PM, said:

...put in the 180 degrees...

Where are you entering this? Manually in the UAVObject Settings? If so, which field?

[FredericG]

In the INS tab in the config gadget

Fred

[jes1111]

FredericG, on 02 May 2011 - 04:22 PM, said:

In the INS tab in the config gadget

Fred

Is there a version difference here? I don't get any fields to enter an offset - this is the binary download "2011/04/10, Rev 3135 (3138), by os".

[FredericG]

Can't verify right now, but I think mine is different.

I use (and compiled) the latest from git

Fred