why when I run the GY521_pitch_roll_yaw example code, all the numbers slowly increase 0 -> 359 -> 0?

[mw66]

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.

[mw66]

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:

jrowberg/i2cdevlib#763

[RobTillaart]

@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?

[mw66]

I got ~+/- 0.xx at zero degrees.

[RobTillaart]

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();

[mw66]

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

[mw66]

looks like:

setThrottle(false)

fix the drifting problem, now the numbers are stable.

[RobTillaart]

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.

[mw66]

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?

[mw66]

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

[RobTillaart]

Looks stable to me now, some random noise.

[mw66]

Looks stable to me now, some random noise.

PITCH 358.378 => 347.883

~10 degree, still random noise?

[RobTillaart]

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.

[RobTillaart]

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.