Self test Gyro and Accel is fail

Hi, Kris.

I bought IC MPU9250 30 pieces,
I soldered them into the board,
Оnly 5 chips passed the self test. Mainly gyroscope test is failed, but accel test sometimes is fail too.
Result of test:
1.
Compass = 0x71 // whoami
Accel test: 2.113, -5.867, 6.430 // x, y, z respectively
Gyro test: 35.538, -1.239, -3.729 // x, y, z respectively
2.
Compass = 0x71
Accel test: 3.196, -1.056, 4.287
Gyro test: 54.333, 7.460, 0.413
3.
Compass = 0x71
Accel test: 10.754, -1.104, 0.790
Gyro test: 46.511, 1.630, -2.275
4.
Compass = 0x71
Accel test: -90.022, -92.128, -90.965
Gyro test: 64.147, 38.558, -11.374
5.
Compass = 0x71
Accel test: 16.004, 52.897, 63.220
Gyro test: 912.197, 17.676, -96.470
6.
Compass = 0x71
Accel test: 9.040, 3.397, -1.560
Gyro test: 85.267, 9.598, -99.190

Soldering temperature did not exceed 270 degrees Celsius.
My init algorithm is:
` resetMPU9250(); // Reset registers to default in preparation for device calibration
Delay_us(20000);

MPU9250SelfTest(selfTestResult);
printf("Accel test: %.3f, %.3f, %.3f\r\n", selfTestResult[0], selfTestResult[1], selfTestResult[2]);
printf("Gyro test: %.3f, %.3f, %.3f\r\n", selfTestResult[3], selfTestResult[4], selfTestResult[5]);

Delay_us(20000);
calibrateMPU9250(NULL, NULL);
Delay_us(3000);
initMPU9250(); 
Delay_us(20000);
res = initAK8963(magCalibration);`

What is the problem, any considering?

This code is meaningless without the method detail. I wouldn't worry too much about the self test results since almost no one implements this properly. You can compare the method to the documentation <https://github.com/kriswiner/MPU9250/blob/master/Documents/AN-MPU-9250A-03%20MPU-9250%20Accel%20Gyro%20and%20Compass%20Self-Test%20Implementation%20v1%200_062813.pdf> . I would simply calibrate the sensors and see if they return reasonable values.

…

On Thu, Jan 21, 2021 at 2:17 AM iamnot-v ***@***.***> wrote: Hi, Kriss. I bought IC MPU9250 30 pieces, I soldered them into the board, Оnly 5 chips passed the self test. Mainly gyroscope test is failed, but accel test sometimes is fail too. Result of test: 1. Compass = 0x71 // whoami Accel test: 2.113, -5.867, 6.430 // x, y, z respectively Gyro test: 35.538, -1.239, -3.729 // x, y, z respectively 2. Compass = 0x71 Accel test: 3.196, -1.056, 4.287 Gyro test: 54.333, 7.460, 0.413 3. Compass = 0x71 Accel test: 10.754, -1.104, 0.790 Gyro test: 46.511, 1.630, -2.275 4. Compass = 0x71 Accel test: -90.022, -92.128, -90.965 Gyro test: 64.147, 38.558, -11.374 5. Compass = 0x71 Accel test: 16.004, 52.897, 63.220 Gyro test: 912.197, 17.676, -96.470 6. Compass = 0x71 Accel test: 9.040, 3.397, -1.560 Gyro test: 85.267, 9.598, -99.190 Soldering temperature did not exceed 270 degrees Celsius. My init algorithm is: resetMPU9250(); // Reset registers to default in preparation for device calibration Delay_us(20000); MPU9250SelfTest(selfTestResult); printf("Accel test: %.3f, %.3f, %.3f\r\n", selfTestResult[0], selfTestResult[1], selfTestResult[2]); printf("Gyro test: %.3f, %.3f, %.3f\r\n", selfTestResult[3], selfTestResult[4], selfTestResult[5]); Delay_us(20000); calibrateMPU9250(NULL, NULL); Delay_us(3000); initMPU9250(); Delay_us(20000); res = initAK8963(magCalibration); What is the problem, any considering? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#452>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKTTCS44KLXKWDT7GFDS275M7ANCNFSM4WMTIZPA> .

Thank you for your answer.

My selftest function was copyed from this project:

MPU9250/Dual_MPU9250/MPU9250.cpp

Line 463 in 72038b0

void MPU9250::calibrateMPU9250(uint8_t MPUnum, float * dest1, float * dest2)

I worry about the self test results because this have direct correlation with bechaviour of MPU9250.
If gyroscope test is fail i have slow change yaw at fast rotate of device.
if accelerometer test is fail then i have errors yaw at tilt of device.

You linked to the calibration function, which is quite different. If this is what you are using as your "self test" function, maybe there are other, um, errors in your code too.

…

On Mon, Jan 25, 2021 at 5:09 AM iamnot-v ***@***.***> wrote: Thank you for your answer. My selftest function was copyed from this project: https://github.com/kriswiner/MPU9250/blob/72038b040bef3cf072612cd8a8ee8f26e3c87158/Dual_MPU9250/MPU9250.cpp#L463 I worry about the self test results because this have direct correlation with bechaviour of MPU9250. If gyroscope test is fail i have slow change yaw at fast rotate of device. if accelerometer test is fail then i have errors yaw at tilt of device. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#452 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKXQQZJZU2IFYADQZKLS3VUSJANCNFSM4WMTIZPA> .

this is my error.

that is selftest function what i use.

MPU9250/Dual_MPU9250/MPU9250.cpp

Line 610 in 72038b0

    
           void MPU9250::SelfTest(uint8_t MPUnum, float * destination) // Should return percent deviation from factory trim values, +/- 14 or less deviation is a pass

But of course, this is wrong, because the self test only uses the LSByte. Check the documentation. In any case, if you get crappy results, lots of jitter or poor response it doesn't matter what the self test says, the chips are bad. If you buy only high-quality breakouts or genuine MPU9250 not fakes less chance of these kinds of problems....

…

On Mon, Jan 25, 2021 at 10:34 PM iamnot-v ***@***.***> wrote: this is my error. that is selftest function what i use. https://github.com/kriswiner/MPU9250/blob/72038b040bef3cf072612cd8a8ee8f26e3c87158/Dual_MPU9250/MPU9250.cpp#L610 — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#452 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABTDLKRIK6G3IXH5EHVQWDLS3ZO53ANCNFSM4WMTIZPA> .

Thank, Kris.

This is bad ICs,
I think my issue can be close.