Help need for mixing table

[DUKE40]

Even trying to compile only l1Tool, I run into an error which clearly points to missing libraries.
But from the path It looks like own libraries of the developper from the namespace name:
duke40@mail:~/autoquad-read-only/ground$ make l1Tool
g++ -c -g -O3 l1Tool.cc -o ../build/l1Tool.o -I../../../lib/eigen
l1Tool.cc:3:19: error: expat.h: Datei oder Verzeichnis nicht gefunden
l1Tool.cc:11:22: error: Eigen/Core: Datei oder Verzeichnis nicht gefunden
l1Tool.cc:12:20: error: Eigen/LU: Datei oder Verzeichnis nicht gefunden
l1Tool.cc:13:21: error: Eigen/SVD: Datei oder Verzeichnis nicht gefunden
l1Tool.cc:15: error: ‘Eigen’ is not a namespace-name
l1Tool.cc:15: error: expected namespace-name before ‘;’ token
l1Tool.cc:91: error: ‘Vector3d’ does not name a type
l1Tool.cc:92: error: ‘MatrixXd’ does not name a type
l1Tool.cc:93: error: ‘MatrixXd’ does not name a type
l1Tool.cc:94: error: ‘VectorXd’ does not name a type
l1Tool.cc:95: error: ‘VectorXd’ does not name a type
l1Tool.cc:96: error: ‘MatrixXd’ does not name a type
l1Tool.cc:97: error: ‘MatrixXd’ does not name a type
l1Tool.cc:99: error: ‘VectorXd’ does not name a type
l1Tool.cc:101: error: ‘VectorXd’ does not name a type
l1Tool.cc:103: error: ‘VectorXd’ does not name a type
l1Tool.cc:104: error: ‘VectorXd’ does not name a type
l1Tool.cc:107: error: ‘MatrixXd’ does not name a type
l1Tool.cc:108: error: ‘MatrixXd’ does not name a type
l1Tool.cc:109: error: ‘MatrixXd’ does not name a type
l1Tool.cc:110: error: ‘MatrixXd’ does not name a type
l1Tool.cc:111: error: ‘Matrix3d’ does not name a type
l1Tool.cc:134: error: ‘numeric_limits’ is not a member of ‘std’
l1Tool.cc:134: error: expected ‘(’ before ‘>’ token
l1Tool.cc:134: error: ‘::epsilon’ has not been declared
l1Tool.cc: In function ‘bool pseudoInverse(const _Matrix_Type_&, _Matrix_Type_&, double)’:
l1Tool.cc:135: error: ‘Eigen’ has not been declared
l1Tool.cc:135: error: expected primary-expression before ‘>’ token
l1Tool.cc:135: error: ‘svd’ was not declared in this scope
l1Tool.cc:135: error: ‘Eigen’ has not been declared
l1Tool.cc:135: error: ‘Eigen’ has not been declared
l1Tool.cc: In function ‘int l1ToolFindPort(int)’:
l1Tool.cc:147: error: ‘struct l1Data_t’ has no member named ‘ports’
l1Tool.cc:148: error: ‘struct l1Data_t’ has no member named ‘ports’
l1Tool.cc: At global scope:
l1Tool.cc:154: error: ISO C++ forbids declaration of ‘XML_Char’ with no type
l1Tool.cc:154: error: expected ‘,’ or ‘...’ before ‘*’ token
l1Tool.cc: In function ‘int l1ToolConfigTypeByName(int)’:
l1Tool.cc:158: error: ‘name’ was not declared in this scope
l1Tool.cc:158: error: ‘strlen’ was not declared in this scope
l1Tool.cc:158: error: ‘strncasecmp’ was not declared in this scope
l1Tool.cc: At global scope:
l1Tool.cc:164: error: ISO C++ forbids declaration of ‘XML_Char’ with no type
l1Tool.cc:164: error: expected ‘,’ or ‘...’ before ‘*’ token
l1Tool.cc: In function ‘int l1ToolElementIdByName(int)’:
l1Tool.cc:168: error: ‘name’ was not declared in this scope
l1Tool.cc:168: error: ‘strlen’ was not declared in this scope
l1Tool.cc:168: error: ‘strncasecmp’ was not declared in this scope
l1Tool.cc: At global scope:
l1Tool.cc:174: error: expected initializer before ‘*’ token
make: *** [../build/l1Tool.o] Fehler 1

This make it very difficult for an outstander to get the tools compiled.

[DUKE40]

I really would like to calculate the mixing table myself, but with what is currently available to me, this is getting a steeplechase.
If anybody could help me out, by calculating the mixing table, just let me know which measurements are needed.
I will provide them.

[LPR]

Duke

Ivo is flying a quad similar to yours.

viewtopic.php?f=36&t=2279&start=20

Larry

[simon]

Duke,
I think we cannot use circle CG approach in this case (just a feeling though;). If you average the x and y axes separately, You will get this numbers for Octave calculations:

A = [
-0.26 0.26 0.215 -0.215
0.155 0.155 -0.155 -0.155
0.1 -0.1 0.1 -0.1
];

And the results look promising:

Btilde =
roll pitch yaw
-100 100 83
100 100 -83
100 -100 100
-100 -100 -100

Pitch and Roll axis must be balanced separately (If my reasoning is correct:). The required engine power is proportional to the distance from the axis (x and y), not the geometric center of those motors (circle).
If you consider CoG as a defined by circle (geometric), front motors will have to perform harder to maintain balance in the pitch axis.

[DUKE40]

Hi Simon,

this sounds good, but will need me changing the setup to allow me to get it right where it should be.
This will mean that my quad will get higher than it currently is.

I think I need to think how to get the best of it.

[simon]

Not necessarily. Maybe moving the battery to the back a little will be enough to balance the quad on pitch axis.

I think that the AQ board can be offset a little? Or not - different acc readings during yawing? But this question should be answered by someone with a better knowledge of AQ than me:)

And maybe it will just fly right without any modification and moving the battery - if there is enough engine power headroom, a little shift in CoG shouldn't be a problem. And if it is intended mainly for fast-forward flying - it can even be an advantage.

[DUKE40]

I have changed now the configuration and the AQ is now over the crossing of the X and Y axis as suggested by Simon.
This is also the CG, thus I just need now to change my mixing table as provided by Simon and I should have setup my Quad the best possible way on the physical part.
The next step will be tuning the parameters (PID and others).
I was already flying OK in the previous configuration, but I want to have it setup the best possible way, thus making everything possible to be as close as possible to the theoretical optimum.

[DUKE40]

Unfortunately it is too late for flying now
But the changes I made to get the AQ on the CoG have resulted in making my Quad a little higher, and now the AQ is almost on the same level as the props which is in theory even better.

I expect once it is fine tuned to have a Quad which will be near optimum.

[simon]

Just to clarify:
I do not know whether the AQ board needs to be in the CoG - this may not really matter.
I was thinking of balancing the real weight of the copter along 2 axis - just like balancing props. If you grab the copter at the intersection of these axes it should not lean in any direction - then, to hover, all the engines will require the same amount of power.
But if the frame is designed mainly for fast-forward flying (like some fpv frames), the shift of the real weight of the copter towards the front is perfectly ok - the copter will require the same amount of power for all the engines to fly forward (it will require a little more power directed to the front in hovering). But no matter how the copter is balanced in reality, the mixing-table should rather be the same.