GY521
GY521 copied to clipboard
RobTillaart
why when I run the GY521_pitch_roll_yaw example code, all the numbers slowly increase 0 -> 359 -> 0?
I didn't move my chip, but:
...
CNT PITCH ROLL YAW
97940 68.228 70.984 113.046
97941 68.247 71.046 113.051
97942 68.247 71.046 113.051
97943 68.247 71.046 113.051
97944 68.247 71.046 113.051
97945 68.263 71.107 113.057
97946 68.263 71.107 113.057
97947 68.263 71.107 113.057
97948 68.263 71.107 113.057
97949 68.279 71.166 113.061
CNT PITCH ROLL YAW
97950 68.279 71.166 113.061
CNT PITCH ROLL YAW
130150 215.906 233.981 152.385
130151 215.906 233.981 152.385
130152 215.924 234.042 152.390
130153 215.924 234.042 152.390
130154 215.924 234.042 152.390
130155 215.924 234.042 152.390
130156 215.942 234.107 152.395
130157 215.942 234.107 152.395
130158 215.942 234.107 152.395
130159 215.955 234.159 152.399
it goes on and on.
I also tried another library, at lease the output numbers are stable, does not auto-incremental over time when the chip is not moving at all.
But I have another data conversion issue:
https://github.com/jrowberg/i2cdevlib/issues/763
@mw66 Thanks for reporting this issue, not seen this drift in my tests so far. I have not enough information to analyze this problem As the other library seems to work well - can you provide a link? Might give a clue what they do differently.
In the data you provided I see a repeating pattern that every 3 or 4 measurements the data changes a bit. This is not strange as e.g. getYaw() returns an internal variable that only is updated after a call to read(). As read() is throttled, not all reads are factual reads to keep the load on the device within certain limits. This GY521_THROTTLE_TIME is default 10 milliseconds.
You might try to switch of the throttling by setThrottle(false) and see if that improves your readings.
But I have another data conversion issue:
Your other issue, could it be that you need to calibrate ? line 200-210 You state you have output 28 at 90 degrees. What value do you get at zero degrees?
Can you run - GY521_test_1.ino for a few minutes? you may change these line to be all float's to get a bit more detail (I'll make an issue to patch the example)
int ax = sensor.getAccelX();
int ay = sensor.getAccelY();
int az = sensor.getAccelZ();
int gx = sensor.getGyroX();
int gy = sensor.getGyroY();
int gz = sensor.getGyroZ();
int t = sensor.getTemperature();
Capture another try, number going down this time:
CNT PITCH ROLL YAW
1040 0.268 2.047 0.001
1041 0.268 2.047 0.001
1042 0.636 2.528 0.001
1043 0.636 2.528 0.001
1044 0.636 2.528 0.001
1045 0.636 2.528 0.001
1046 0.636 2.528 0.001
1047 359.995 3.084 0.001
1048 359.995 3.084 0.001
1049 359.995 3.084 0.001
CNT PITCH ROLL YAW
1050 359.995 3.084 0.001
1051 0.336 2.654 0.001
1052 0.336 2.654 0.001
1053 0.336 2.654 0.001
1054 0.336 2.654 0.001
1055 0.948 2.108 0.000
1056 0.948 2.108 0.000
1057 0.948 2.108 0.000
1058 0.948 2.108 0.000
1059 0.948 2.108 0.000
CNT PITCH ROLL YAW
1060 0.706 2.335 359.999
1061 0.706 2.335 359.999
1062 0.706 2.335 359.999
1063 0.706 2.335 359.999
1064 0.611 2.540 359.998
1065 0.611 2.540 359.998
1066 0.611 2.540 359.998
1067 0.611 2.540 359.998
1068 0.620 2.600 359.998
1059.996
1074 0.306 2.285 359.996
1075 0.306 2.285 359.996
1076 0.722 1.748 359.995
1077 0.722 1.748 359.995
1078 0.722 1.748 359.995
1079 0.722 1.748 359.995
looks like:
setThrottle(false)
fix the drifting problem, now the numbers are stable.
The raw measurements look indeed stable - maybe longer run needed to be sure.
looks like:
setThrottle(false)fix the drifting problem, now the numbers are stable.
It might be that the default throttle time is too large (10 milliseconds), need to investigate how it would affect the math.
Actually I concluded to quickly, it still drift slowly, but not as steady in one direction as not setting setThrottle(false).
with setThrottle(false), longer output:
CNT PITCH ROLL YAW
15300 358.378 359.556 0.346
15301 358.414 359.785 0.345
15302 358.751 359.422 0.346
15303 358.763 359.415 0.346
15304 358.566 359.638 0.346
15305 358.440 359.743 0.345
15306 358.672 359.528 0.346
15307 358.598 359.509 0.346
15308 358.286 359.817 0.346
15309 358.459 359.618 0.347
CNT PITCH ROLL YAW
15310 358.452 359.706 0.347
15311 358.498 359.341 0.347
15312 358.498 359.636 0.347
15313 358.679 359.504 0.348
15314 358.401 359.779 0.348
15315 358.670 359.529 0.348
15316 358.573 359.623 0.348
15317 358.595 359.394 0.348
15318 358.783 359.413 0.348
15319 358.867 359.310 0.348
...
CNT PITCH ROLL YAW
48170 348.432 357.415 1.723
48171 348.701 357.050 1.723
48172 348.600 357.196 1.723
48173 348.437 357.374 1.723
48174 347.974 357.733 1.722
48175 348.146 357.647 1.722
48176 348.115 357.695 1.722
48177 348.110 357.199 1.722
48178 348.486 357.243 1.722
48179 348.973 356.692 1.721
CNT PITCH ROLL YAW
48180 348.947 356.909 1.721
48181 348.215 357.329 1.722
48182 348.255 357.405 1.722
48183 348.592 357.262 1.722
48184 348.044 357.813 1.723
48185 347.883 357.908 1.723
PITCH 358.378 => 347.883 ROLL 359.556 => 357.908 YAW 0.346 => 1.723
Is there any caching going on? why the math error accumulate?
Can you run - GY521_test_1.ino for a few minutes? you may change these line to be all float's to get a bit more detail (I'll make an issue to patch the example)
int ax = sensor.getAccelX(); int ay = sensor.getAccelY(); int az = sensor.getAccelZ(); int gx = sensor.getGyroX(); int gy = sensor.getGyroY(); int gz = sensor.getGyroZ(); int t = sensor.getTemperature();
here is the output with floats:
GY521_LIB_VERSION: 0.5.1
start...
ACCELEROMETER GYROSCOPE TEMPERATURE
ax ay az gx gy gz T
0 0.06 -0.01 -1.07 -5.73 -1.62 -0.71 17.19
1 0.06 -0.01 -1.06 -5.48 -1.58 -0.50 17.09
2 0.06 -0.00 -1.06 -5.91 -1.53 -0.50 17.09
3 0.06 -0.01 -1.06 -5.76 -1.66 -0.55 17.19
4 0.06 -0.01 -1.06 -5.83 -1.73 -0.66 17.19
5 0.06 -0.01 -1.06 -5.89 -1.61 -0.57 17.24
6 0.06 -0.01 -1.06 -5.76 -1.37 -0.67 17.24
7 0.06 -0.01 -1.07 -5.91 -1.68 -0.45 17.19
8 0.06 -0.01 -1.07 -5.80 -1.34 -0.83 17.19
9 0.06 -0.01 -1.05 -5.59 -1.63 -0.65 17.14
ACCELEROMETER GYROSCOPE TEMPERATURE
ax ay az gx gy gz T
10 0.06 -0.01 -1.06 -5.77 -1.47 -0.56 17.19
11 0.06 -0.01 -1.07 -5.85 -1.66 -0.60 17.14
12 0.06 -0.01 -1.06 -5.73 -1.53 -0.53 17.14
13 0.06 -0.01 -1.05 -5.88 -1.48 -0.53 17.19
14 0.06 -0.00 -1.06 -5.80 -1.56 -0.73 17.19
15 0.06 -0.01 -1.07 -5.80 -1.68 -0.50 17.09
16 0.06 -0.01 -1.06 -5.82 -1.44 -0.67 17.19
17 0.06 -0.01 -1.06 -5.84 -1.53 -0.27 17.19
18 0.06 -0.02 -1.06 -5.98 -1.65 -0.62 17.19
19 0.07 -0.01 -1.07 -5.69 -1.60 -0.52 17.19
...
ACCELEROMETER GYROSCOPE TEMPERATURE
ax ay az gx gy gz T
360 0.06 -0.01 -1.06 -5.93 -1.60 -0.46 17.89
361 0.06 -0.01 -1.07 -5.93 -1.54 -0.76 17.94
362 0.06 -0.01 -1.07 -5.86 -1.54 -0.54 17.89
363 0.06 -0.01 -1.06 -6.00 -1.48 -0.72 17.99
364 0.06 -0.01 -1.06 -5.79 -1.44 -0.47 17.99
365 0.06 -0.01 -1.07 -5.80 -1.60 -0.56 17.94
366 0.07 -0.01 -1.07 -5.72 -1.56 -0.61 17.94
367 0.06 -0.01 -1.06 -5.83 -1.82 -0.56 17.94
368 0.06 -0.01 -1.06 -5.81 -1.41 -0.51 17.89
369 0.06 -0.01 -1.06 -5.71 -1.47 -0.46 18.04
ACCELEROMETER GYROSCOPE TEMPERATURE
ax ay az gx gy gz T
370 0.06 -0.01 -1.06 -5.70 -1.45 -0.57 17.94
371 0.06 -0.01 -1.06 -5.78 -1.73 -0.32 17.94
372 0.06 -0.00 -1.06 -5.84 -1.63 -0.63 17.99
373 0.06 -0.01 -1.08 -5.85 -1.62 -0.40 17.99
374 0.06 -0.01 -1.07 -5.79 -1.63 -0.60 17.99
375 0.06 -0.01 -1.06 -5.63 -1.36 -0.52 17.85
376 0.06 -0.01 -1.07 -5.78 -1.66 -0.82 17.89
377 0.07 -0.01 -1.07 -5.84 -1.84 -0.53 17.89
378 0.06 -0.00 -1.06 -5.54 -1.60 -0.61 17.94
379 0.06 -0.00 -1.06 -5.77 -1.64 -0.50 17.85
ACCELEROMETER GYROSCOPE TEMPERATURE
ax ay az gx gy gz T
380 0.07 -0.01 -1.06 -5.93 -1.59 -0.73 17.85
381 0.06 -0.01 -1.06 -5.80 -1.52 -0.53 17.94
382 0.05 -0.01 -1.06 -5.66 -1.75 -0.68 17.99
Looks stable to me now, some random noise.
PITCH 358.378 => 347.883
~10 degree, still random noise?
Looks stable to me now, some random noise.
PITCH 358.378 => 347.883
~10 degree, still random noise?
Sorry, my comment was on the Accelerometer gyroscope temperature data from the - GY521_test_1.ino. I should have been more explicit on that.
I need to investigate why the raw values are stable and the PITCH ROLL YAW are drifting. Hope to find some time coming days or next week to recreate the issue.
Note: Combine GY521_test_1.ino with pitch roll yaw demo, ==> put the data in a spreadsheet format. new demo sketch GY521_debug.ino or so.
@mw66
Created a debug sketch that combines the raw accel + gyro + temp with the derived angle and pitch, roll yaw values.
Can you generate e.g. 10-20 lines ?
//
// FILE: GY521_test_debug.ino
// AUTHOR: Rob Tillaart
// PURPOSE: output raw and cooked data from sensor for analyzing.
// URL: https://github.com/RobTillaart/GY521
#include "GY521.h"
GY521 sensor(0x68);
uint32_t counter = 0;
void setup()
{
Serial.begin(115200);
Serial.println();
Serial.println(__FILE__);
Serial.print("GY521_LIB_VERSION: ");
Serial.println(GY521_LIB_VERSION);
Wire.begin();
delay(100);
while (sensor.wakeup() == false)
{
Serial.print(millis());
Serial.println("\tCould not connect to GY521: please check the GY521 address (0x68/0x69)");
delay(1000);
}
sensor.setAccelSensitivity(0); // 2g
sensor.setGyroSensitivity(0); // 250 degrees/s
sensor.setThrottle();
Serial.println("start...");
// set calibration values from calibration sketch.
sensor.axe = 0;
sensor.aye = 0;
sensor.aze = 0;
sensor.gxe = 0;
sensor.gye = 0;
sensor.gze = 0;
Serial.println("\n\tACCELEROMETER\t\tGYROSCOPE\t\tTEMP\tPRY\t\t\tANGLE");
Serial.println("\tax\tay\taz\tgx\tgy\tgz\tT\tPITCH\tROLL\tYAW\tX\tY\tZ");
}
void loop()
{
sensor.read();
float ax = sensor.getAccelX();
float ay = sensor.getAccelY();
float az = sensor.getAccelZ();
float gx = sensor.getGyroX();
float gy = sensor.getGyroY();
float gz = sensor.getGyroZ();
float t = sensor.getTemperature();
float pitch = sensor.getPitch();
float roll = sensor.getRoll();
float yaw = sensor.getYaw();
float x = sensor.getAngleX();
float y = sensor.getAngleY();
float z = sensor.getAngleZ();
Serial.print(counter);
Serial.print('\t');
Serial.print(ax, 3);
Serial.print('\t');
Serial.print(ay, 3);
Serial.print('\t');
Serial.print(az, 3);
Serial.print('\t');
Serial.print(gx, 3);
Serial.print('\t');
Serial.print(gy, 3);
Serial.print('\t');
Serial.print(gz, 3);
Serial.print('\t');
Serial.print(t, 3);
Serial.print('\t');
Serial.print(pitch, 3);
Serial.print('\t');
Serial.print(roll, 3);
Serial.print('\t');
Serial.print(yaw, 3);
Serial.print('\t');
Serial.print(x, 1);
Serial.print('\t');
Serial.print(y, 1);
Serial.print('\t');
Serial.print(z, 1);
Serial.println();
counter++;
// delay(1000);
}
// -- END OF FILE --
@mw66 Did some investigation of the status of the GY521 code and the Pitch Roll Yaw part is not correct and needs a lot of work. I will make this explicit in the readme file of the 0.5.2 version. Must find a long timeslot to dive into and redo the math needed.
Sorry for the inconvenience.
Can you generate e.g. 10-20 lines ?
Here you are:
12344 0.080 -0.027 -1.069 -5.824 -1.389 -0.420 16.954 216.632 166.603 298.401 -1.4 -4.3 -85.5
12345 0.083 -0.016 -1.063 -5.794 -1.473 -0.679 17.095 216.602 166.535 298.390 -0.9 -4.5 -85.5
12346 0.083 -0.016 -1.063 -5.794 -1.473 -0.679 17.095 216.602 166.535 298.390 -0.9 -4.5 -85.5
12347 0.090 -0.019 -1.072 -5.702 -1.427 -0.695 17.095 216.567 166.443 298.379 -1.0 -4.8 -85.1
12348 0.078 -0.020 -1.064 -5.847 -1.588 -0.802 17.001 216.577 166.387 298.372 -1.1 -4.2 -85.7
12349 0.078 -0.020 -1.064 -5.847 -1.588 -0.802 17.001 216.577 166.387 298.372 -1.1 -4.2 -85.7
12350 0.080 -0.021 -1.062 -5.756 -1.405 -0.511 17.048 216.551 166.299 298.364 -1.1 -4.3 -85.5
12351 0.080 -0.021 -1.062 -5.756 -1.405 -0.511 17.048 216.551 166.299 298.364 -1.1 -4.3 -85.5
12352 0.085 -0.030 -1.065 -5.779 -1.733 -0.626 17.001 216.515 166.194 298.354 -1.6 -4.5 -85.2
12353 0.087 -0.021 -1.056 -5.893 -1.282 -0.557 17.048 216.497 166.160 298.349 -1.1 -4.7 -85.1
12354 0.081 -0.023 -1.062 -5.901 -1.573 -0.603 16.954 216.498 166.103 298.343 -1.2 -4.3 -85.5
12355 0.081 -0.023 -1.062 -5.901 -1.573 -0.603 16.954 216.498 166.103 298.343 -1.2 -4.3 -85.5
12356 0.081 -0.017 -1.072 -5.649 -1.443 -0.511 17.001 216.477 166.032 298.335 -0.9 -4.3 -85.6
12357 0.081 -0.017 -1.072 -5.649 -1.443 -0.511 17.001 216.477 166.032 298.335 -0.9 -4.3 -85.6
12358 0.088 -0.021 -1.061 -5.840 -1.656 -0.702 16.954 216.436 165.936 298.324 -1.1 -4.7 -85.1
12359 0.088 -0.021 -1.061 -5.840 -1.656 -0.702 16.954 216.436 165.936 298.324 -1.1 -4.7 -85.1
12360 0.077 -0.021 -1.061 -5.939 -1.458 -0.603 16.954 216.438 165.846 298.315 -1.1 -4.1 -85.7
12361 0.077 -0.021 -1.061 -5.939 -1.458 -0.603 16.954 216.438 165.846 298.315 -1.1 -4.1 -85.7
12362 0.080 -0.023 -1.064 -5.802 -1.573 -0.664 16.906 216.408 165.754 298.304 -1.2 -4.3 -85.5
12363 0.080 -0.023 -1.064 -5.802 -1.573 -0.664 16.906 216.408 165.754 298.304 -1.2 -4.3 -85.5
12364 0.088 -0.022 -1.058 -5.863 -1.359 -0.618 17.001 216.370 165.667 298.295 -1.2 -4.7 -85.1
12365 0.088 -0.022 -1.058 -5.863 -1.359 -0.618 17.001 216.370 165.667 298.295 -1.2 -4.7 -85.1
12366 0.081 -0.024 -1.064 -5.924 -1.405 -0.863 16.954 216.362 165.571 298.281 -1.3 -4.4 -85.4
12367 0.081 -0.024 -1.064 -5.924 -1.405 -0.863 16.954 216.362 165.571 298.281 -1.3 -4.4 -85.4
12368 0.079 -0.023 -1.065 -5.962 -1.527 -0.656 17.048 216.344 165.480 298.270 -1.3 -4.2 -85.6
12369 0.081 -0.021 -1.068 -5.855 -1.305 -0.725 16.954 216.329 165.433 298.263 -1.1 -4.3 -85.5
12370 0.081 -0.021 -1.068 -5.855 -1.305 -0.725 16.954 216.329 165.433 298.263 -1.1 -4.3 -85.5
12371 0.085 -0.021 -1.069 -5.863 -1.473 -0.580 16.954 216.298 165.346 298.254 -1.1 -4.5 -85.3
12372 0.085 -0.021 -1.069 -5.863 -1.473 -0.580 16.954 216.298 165.346 298.254 -1.1 -4.5 -85.3
12373 0.078 -0.019 -1.061 -5.786 -1.443 -0.611 17.001 216.290 165.262 298.245 -1.0 -4.2 -85.7
12374 0.080 -0.024 -1.071 -5.832 -1.366 -0.504 17.095 216.276 165.200 298.240 -1.3 -4.3 -85.6
12375 0.080 -0.024 -1.071 -5.832 -1.366 -0.504 17.095 216.276 165.200 298.240 -1.3 -4.3 -85.6
12376 0.084 -0.020 -1.063 -5.832 -1.427 -0.603 16.954 216.243 165.122 298.231 -1.1 -4.5 -85.3
12377 0.081 -0.019 -1.067 -5.977 -1.573 -0.603 17.001 216.236 165.070 298.225 -1.0 -4.4 -85.5
12378 0.081 -0.019 -1.067 -5.977 -1.573 -0.603 17.001 216.236 165.070 298.225 -1.0 -4.4 -85.5
12379 0.073 -0.018 -1.061 -5.908 -1.466 -0.756 17.001 216.231 164.985 298.213 -0.9 -3.9 -86.0
12380 0.073 -0.018 -1.061 -5.908 -1.466 -0.756 17.001 216.231 164.985 298.213 -0.9 -3.9 -86.0
12381 0.085 -0.023 -1.055 -5.840 -1.458 -0.626 17.001 216.183 164.886 298.204 -1.2 -4.6 -85.2
12382 0.085 -0.023 -1.055 -5.840 -1.458 -0.626 17.001 216.183 164.886 298.204 -1.2 -4.6 -85.2
12383 0.084 -0.019 -1.074 -5.969 -1.435 -0.733 17.048 216.167 164.805 298.192 -1.0 -4.5 -85.4
12384 0.084 -0.019 -1.074 -5.969 -1.435 -0.733 17.048 216.167 164.805 298.192 -1.0 -4.5 -85.4
12385 0.080 -0.018 -1.067 -5.954 -1.374 -0.634 17.001 216.152 164.718 298.182 -0.9 -4.3 -85.6
12386 0.080 -0.018 -1.067 -5.954 -1.374 -0.634 17.001 216.152 164.718 298.182 -0.9 -4.3 -85.6
Thanks for looking into this issue.
@RobTillaart Just want to let you know, according to the discussion of the issue in other library, I may have a faulty chip, the data and conversation is here:
https://github.com/ElectronicCats/mpu6050/issues/74
@mw66 Thanks for the table, looking per column the standard deviation is small which indicates the values are stable. Within the accuracy of the decimals printed I could reconstruct the three angle columns (last three).
For PRY I need more time to dive into the math, expect it will take a lot of time to understand. (as there are many other issues and work etc to be done). alas it is on my todo list for 2024.
@RobTillaart Just want to let you know, according to the discussion of the issue in other library, I may have a faulty chip, the data and conversation is here:
The words of jrowberg is about the math I need to dive into.
To be continued
@RobTillaart Just want to let you know, according to the discussion of the issue in other library, I may have a faulty chip, the data and conversation is here: ElectronicCats/mpu6050#74
The words of jrowberg is about the math I need to dive into.
To be continued
You may also check this library:
https://github.com/gabriel-milan/TinyMPU6050
where I can get stable ~80 degree reading for 90 degree chip position, so I think there maybe a bug in i2cdevlib (and mpu6050 since it uses the former).
Checking other peoples code is good to do after studying the math, just copying it without understanding is risky.