MultiWii

Alexinparis wrote:...
On MEGA boards: the first 4 motors are now driven by timers configured in a 16 bit mode.
The difference is noticeable in flight for a quad. It's a really nice improvement for MEGA boards running a quad for instance.
...

dongs wrote:You forgot the STM32 port jk

nhadrian wrote:Hi,

please don't forget the ultrasonic range finder implementation!
It looks like coding for SRF08 is done, calculations should be tested.

BR
Adrian

Magnetron wrote:Could you consider my mod: moving average on gyro and acc freely configurable via config.h?

ronco wrote:

Alexinparis wrote:...
On MEGA boards: the first 4 motors are now driven by timers configured in a 16 bit mode.
The difference is noticeable in flight for a quad. It's a really nice improvement for MEGA boards running a quad for instance.
...

Hi Alex,

a small correction the first 6 motors are with 16bit PWM on MEGA boards.

regards

felix

Bledi wrote:One more think : the state machine in the main loop to reduce the time loop. The GPS is not in this "case" list

Hamburger wrote:We achieved even more. Have a look at the wishlist thread first pozt. viewtopic.php?f=7&t=882&p=5366#p5366

ronco wrote:Hi,

found one bug (dev 20120225) .. tri yaw servo dont moves in GUI.
the output signal of it works well.


regards felix

jyrcorp wrote:Hello,

I found a bug in dev 20120219 for my octo with arduino mini.
I'm using pins A0, A1, A2 and 12 to connect the 4 last motors.
I defined #define A0_A1_PIN_HEX and #define OCTOFLATX but motors 5 and 6 are not initialized.
I copied/paste some code in output.ino so that with 8 motors the motors 5 and 6 and now initialized:

Code: Select all

...
    #elif (NUMBER_MOTOR == 8) // PIN A2 & 12
      initializeSoftPWM();
// modif start     
      #if defined(A0_A1_PIN_HEX) || (NUMBER_MOTOR > 6)
        pinMode(5,INPUT);pinMode(6,INPUT);     // we reactivate the INPUT affectation for these two PINs
        pinMode(A0,OUTPUT);pinMode(A1,OUTPUT);
      #endif
// modif end     
    #endif
  #endif
...
 

EOSBandi wrote:Hi !
Error found in output.pde. in line 617

Code: Select all

 servo[5] = constrain(tri_yaw_middle + YAW_DIRECTION * axisPID[YAW], TRI_YAW_CONSTRAINT_MIN, TRI_YAW_CONSTRAINT_MAX); //REAR

should be

Code: Select all

servo[5] = constrain(TRI_YAW_MIDDLE + YAW_DIRECTION * axisPID[YAW], TRI_YAW_CONSTRAINT_MIN, TRI_YAW_CONSTRAINT_MAX); //REAR

On the other hand, it's amazing how much it developed during the past 12 months....

Alexinparis wrote:I've completed the first post

Alexinparis wrote:

Magnetron wrote:Could you consider my mod: moving average on gyro and acc freely configurable via config.h?

Hi,
I'm not totally convinced by the interest of this mod:
- for gyro, I would really like to see a video which shows an improvement on a multi before->after. From my tuning experience, any delay after 2 or 3 cycles on gyro makes the situation worse.
I think it's however important for flying wing or servo heli configs, but we have something for this per axis in the shared rep based on LPF. So convince me and I will adopt it
- for ACC, a smoothing a naturally done via LPF filter.
- for Baro, the algo uses already a moving average of 20 samples spaced each by 25ms

• starts with M as it does now.
• 1 byte count of segments.
• 1 byte version number, and IMPORTANT NOTE, this is the version of this data segment, not the code!
• Length of this segment (1 byte?)
• Binary data almost exactly like now, but segment 1 contains only "core" stuff. PID gains, RC values, whatever we decide is core.
• Checksum byte.

Hamburger wrote:it was already under discussion to split the current M data into two:
- new M for the sensor data etc. (only what is neccessary to update the GUI)
- new X for the static data like PID settings etc. - would be transmitted only upon request (the READ button)
Motivation was to reduce transmission time for the M stream - currently each transmission exceeds the 650us limit. This may be bad because people use this to tune their PIDs which they do with an exceptionally prolonged cycle time.

Whatever we want to do - probably too late for 2.0

EOSBandi wrote:MavLink is unfortunately out of the question due memory constraints...

dongs wrote:What do you mean "not working"? It's working in both cases.

Tifani wrote:Thanks Alex and all !!!
Just little sad than You abandoned Ultrasonic Range Finder - Maxbotix LV-EZ1
Regards
Tom

Magnetron wrote:

Alexinparis wrote:

Magnetron wrote:Could you consider my mod: moving average on gyro and acc freely configurable via config.h?

Hi,
I'm not totally convinced by the interest of this mod:
- for gyro, I would really like to see a video which shows an improvement on a multi before->after. From my tuning experience, any delay after 2 or 3 cycles on gyro makes the situation worse.
I think it's however important for flying wing or servo heli configs, but we have something for this per axis in the shared rep based on LPF. So convince me and I will adopt it
- for ACC, a smoothing a naturally done via LPF filter.
- for Baro, the algo uses already a moving average of 20 samples spaced each by 25ms

Alex, moving average implemented by me was on gyro and acc on a 8-lenght vector. I tryed personally on my quad and other users had tryed it on italian forum (baronerosso) and the feedback were positive. The average is costumizable via config.h file so if user want a little average he could have or can be fully disabled. I have no video but you can try directly. I think that this mod can be usefull for all type of flying object. For baro I ipotize about 6 months ago the low/high moving average algorithm but I have no implemented due to other problem (I speak this idea on italian forum). Alex, I want only help development for multiwii, consider my mod.

mbrak wrote:hi

a change for setting new home position during flight with AUX4 (High). without the secound term in the if-structure you would not get the FIX LED pattern automaticly.

Line 736 in DEV 20120225

Code: Select all

 if (rcData[AUX4]>1800 || GPS_fix_home == 0) {

EOSBandi wrote:

Hamburger wrote:it was already under discussion to split the current M data into two:
- new M for the sensor data etc. (only what is neccessary to update the GUI)
- new X for the static data like PID settings etc. - would be transmitted only upon request (the READ button)
Motivation was to reduce transmission time for the M stream - currently each transmission exceeds the 650us limit. This may be bad because people use this to tune their PIDs which they do with an exceptionally prolonged cycle time.

Whatever we want to do - probably too late for 2.0

Agree twice :
1.Serial protocol revamping is necessary. The current version is rather a quick hack and lacking reliability and robustness. MavLink is unfortunately out of the question due memory constraints...
2.Current dev is feature freezed, we must concentrate on finding and fixing bugs...

Stalk wrote:Altitude hold in dev_20120225 only works well at temperatures above 20 degrees.
In a room works well, but outside no.
In my workshop I have 18 degrees.
Sometimes I get this in the GUI:

Alexinparis wrote:

mbrak wrote:hi

a change for setting new home position during flight with AUX4 (High). without the secound term in the if-structure you would not get the FIX LED pattern automaticly.

Line 736 in DEV 20120225

Code: Select all

 if (rcData[AUX4]>1800 || GPS_fix_home == 0) {

Hi,
I've corrected it. I'm sure EOSBandi will enhanced anyway this part dealing with I2C GPS

Alexinparis wrote:

Magnetron wrote:

Alexinparis wrote:Hi,
I'm not totally convinced by the interest of this mod:
- for gyro, I would really like to see a video which shows an improvement on a multi before->after. From my tuning experience, any delay after 2 or 3 cycles on gyro makes the situation worse.
I think it's however important for flying wing or servo heli configs, but we have something for this per axis in the shared rep based on LPF. So convince me and I will adopt it
- for ACC, a smoothing a naturally done via LPF filter.
- for Baro, the algo uses already a moving average of 20 samples spaced each by 25ms

Alex, moving average implemented by me was on gyro and acc on a 8-lenght vector. I tryed personally on my quad and other users had tryed it on italian forum (baronerosso) and the feedback were positive. The average is costumizable via config.h file so if user want a little average he could have or can be fully disabled. I have no video but you can try directly. I think that this mod can be usefull for all type of flying object. For baro I ipotize about 6 months ago the low/high moving average algorithm but I have no implemented due to other problem (I speak this idea on italian forum). Alex, I want only help development for multiwii, consider my mod.

ok,
So I keep the moving average for gyro (to let more people to test it easier via a simple #define)
And I keep also the moving average for gimbal as I see here something really useful.

//#######################################################
// FUNCTIONS BY MAGNETRON1 (Michele Ardito)
// Moving Average Gyros by Magnetron1
#define MMGYRO                         // Active Moving Average Function for Gyros
#define MMGYROVECTORLENGHT 8           // Lenght of Moving Average Vector
// Moving Average Accelerometers by Magnetron1
#define MMACC                          // Active Moving Average Function for Accelerometers
#define MMACCVECTORLENGHT 8            // Lenght of Moving Average Vector
// Trusted AccZ with PIDVEL Enable
#define TRUSTED_ACCZ_PID               // Active function Trusted AccZ with PIDVEL

// ****************
// GYRO common part
// ****************
void GYRO_Common() {
  static int16_t previousGyroADC[3] = {0,0,0};
  static int32_t g[3];
  uint8_t axis;
#if defined MMGYRO       
  // Moving Average Gyros by Magnetron1
  //---------------------------------------------------
  static int16_t mediaMobileGyroADC[3][MMGYROVECTORLENGHT];
  static int32_t mediaMobileGyroADCSum[3];
  static uint8_t mediaMobileGyroIDX;
  //---------------------------------------------------
#endif
  if (calibratingG>0) {
    for (axis = 0; axis < 3; axis++) {
      // Reset g[axis] at start of calibration
      if (calibratingG == 400) g[axis]=0;
      // Sum up 400 readings
      g[axis] +=gyroADC[axis];
      // Clear global variables for next reading
      gyroADC[axis]=0;
      gyroZero[axis]=0;
      if (calibratingG == 1) {
        gyroZero[axis]=g[axis]/400;
        blinkLED(10,15,1+3*nunchuk);
      }
    }
    calibratingG--;
  }
#ifdef MMGYRO       
  //Moving Average Gyros by Magnetron1
  mediaMobileGyroIDX = ++mediaMobileGyroIDX % MMGYROVECTORLENGHT;
  for (axis = 0; axis < 3; axis++) {
    gyroADC[axis]  -= gyroZero[axis];
   //new implementation of Moving Average for fast response
   mediaMobileGyroADCSum[axis] -= mediaMobileGyroADC[axis][mediaMobileGyroIDX];
    //anti gyro glitch, limit the variation between two consecutive readings
    mediaMobileGyroADC[axis][mediaMobileGyroIDX] = constrain(gyroADC[axis],previousGyroADC[axis]-800,previousGyroADC[axis]+800);
   mediaMobileGyroADCSum[axis] += mediaMobileGyroADC[axis][mediaMobileGyroIDX];
   gyroADC[axis] = mediaMobileGyroADCSum[axis] / MMGYROVECTORLENGHT;
#else 
  for (axis = 0; axis < 3; axis++) {
    gyroADC[axis]  -= gyroZero[axis];
    //anti gyro glitch, limit the variation between two consecutive readings
    gyroADC[axis] = constrain(gyroADC[axis],previousGyroADC[axis]-800,previousGyroADC[axis]+800);
#endif    
    previousGyroADC[axis] = gyroADC[axis];
  }
}

// ****************
// ACC common part
// ****************
void ACC_Common() {
  static int32_t a[3];
  uint8_t axis;
#if defined MMACC       
  // Moving Average Accelerometers by Magnetron1
  //---------------------------------------------------
  static int16_t mediaMobileAccADC[3][MMACCVECTORLENGHT];
  static int32_t mediaMobileAccADCSum[3];
  static uint8_t mediaMobileAccIDX;
  //---------------------------------------------------
#endif
  if (calibratingA>0) {
    for (axis = 0; axis < 3; axis++) {
      // Reset a[axis] at start of calibration
      if (calibratingA == 400) a[axis]=0;
      // Sum up 400 readings
      a[axis] +=accADC[axis];
      // Clear global variables for next reading
      accADC[axis]=0;
      accZero[axis]=0;
    }
    // Calculate average, shift Z down by acc_1G and store values in EEPROM at end of calibration
    if (calibratingA == 1) {
      accZero[ROLL]  = a[ROLL]/400;
      accZero[PITCH] = a[PITCH]/400;
      accZero[YAW]   = a[YAW]/400-acc_1G; // for nunchuk 200=1G
      accTrim[ROLL]   = 0;
      accTrim[PITCH]  = 0;
      writeParams(); // write accZero in EEPROM
    }
    calibratingA--;
  }
  #if defined(InflightAccCalibration)
      static int32_t b[3];
      static int16_t accZero_saved[3]  = {0,0,0};
      static int16_t  accTrim_saved[2] = {0, 0};
      //Saving old zeropoints before measurement
      if (InflightcalibratingA==50) {
         accZero_saved[ROLL]  = accZero[ROLL] ;
         accZero_saved[PITCH] = accZero[PITCH];
         accZero_saved[YAW]   = accZero[YAW] ;
         accTrim_saved[ROLL] = accTrim[ROLL] ;
         accTrim_saved[PITCH] = accTrim[PITCH] ;
      }
      if (InflightcalibratingA>0) {
        for (axis = 0; axis < 3; axis++) {
          // Reset a[axis] at start of calibration
          if (InflightcalibratingA == 50) b[axis]=0;
          // Sum up 50 readings
          b[axis] +=accADC[axis];
          // Clear global variables for next reading
          accADC[axis]=0;
          accZero[axis]=0;
        }
        //all values are measured
        if (InflightcalibratingA == 1) {
          AccInflightCalibrationActive = 0;
          AccInflightCalibrationMeasurementDone = 1;
          blinkLED(10,10,2);      //buzzer for indicatiing the start inflight
        // recover saved values to maintain current flight behavior until new values are transferred
         accZero[ROLL]  = accZero_saved[ROLL] ;
         accZero[PITCH] = accZero_saved[PITCH];
         accZero[YAW]   = accZero_saved[YAW] ;
         accTrim[ROLL] = accTrim_saved[ROLL] ;
         accTrim[PITCH] = accTrim_saved[PITCH] ;
        }
        InflightcalibratingA--;
      }
      // Calculate average, shift Z down by acc_1G and store values in EEPROM at end of calibration
      if (AccInflightCalibrationSavetoEEProm == 1){  //the copter is landed, disarmed and the combo has been done again
        AccInflightCalibrationSavetoEEProm = 0;
        accZero[ROLL]  = b[ROLL]/50;
        accZero[PITCH] = b[PITCH]/50;
        accZero[YAW]   = b[YAW]/50-acc_1G; // for nunchuk 200=1G
        accTrim[ROLL]   = 0;
        accTrim[PITCH]  = 0;
        writeParams(); // write accZero in EEPROM
      }
  #endif
#ifdef MMACC
  //Moving Average Accelerometers by Magnetron1
  mediaMobileAccIDX = ++mediaMobileAccIDX % MMACCVECTORLENGHT;
  for (axis = 0; axis < 3; axis++) {
    accADC[axis]  -= accZero[axis];
   //new implementation of Moving Average for fast response
   mediaMobileAccADCSum[axis] -= mediaMobileAccADC[axis][mediaMobileAccIDX];
    mediaMobileAccADC[axis][mediaMobileAccIDX] = accADC[axis];
   mediaMobileAccADCSum[axis] += mediaMobileAccADC[axis][mediaMobileAccIDX];
   accADC[axis] = mediaMobileAccADCSum[axis] / MMACCVECTORLENGHT;
  }
#else 
  accADC[ROLL]  -=  accZero[ROLL] ;
  accADC[PITCH] -=  accZero[PITCH];
  accADC[YAW]   -=  accZero[YAW] ;
#endif    
}

  #if defined(TRUSTED_ACCZ_PID)
     temp32 = D8[PIDALT]*(BaroHigh - BaroLow) / 400 + accADC[YAW]*D8[PIDVEL] / 30;
  #else
     temp32 = D8[PIDALT]*(BaroHigh - BaroLow) / 400;
  #endif

nhadrian wrote:Please note that in preversion 2.0 in multiwii.ino the complete I2C_GPS code is missing from the end.
BTW, excellent job!!! Many thanks for your job in behalf of us!

Are you planning to put in the implementation of SRF08 range finder together with the modified alt code?

BR
Adrian