Why doesn't the MMA8451Q accelerometer readings from the FIFO indicate the Motion Threshold Trigger value?

[seanbeatty]

Greetings - I'm puzzled by the results I'm getting from the MMA8451Q accelerometer. I've set the device up in Motion Detection mode with a threshold of approx. 1.5g's. But when the device indicates it's reached the Motion Detection Threshold and I read the acceleration values, they are often much lower than the specifiedThreshold!

Here are details of my setup -

// Set the range, -2g to 2g with High Pass Filter on, 800Hz ODR

	// Set to High Resolution mode ... and take the other defaults for CTRL_REG2

// Set to detect Motion on any axis - Debounced

  // Set motion threshold to 1.5g. Leave Debounce Counter Mode in Inc/Dec

	// Setup FIFO to Trigger Mode, set Watermark = 16

	// Setup FIFO to Trigger on Motion Detection exceeding Threshold

	//Set FIFO interrupt only to pin INT1

   //Enable FIFO interrupt only in MMA8451

XYZ_DATA_CFG: 0x10

FF_MT_CFG: 0xF8     //High Pass filter on, Fullscale is -2g to +2g

FF_MT_THS:   24     // 24 x 0.0625g = 1.5g when fullscale is set to 2g

FF_MT_COUNT: 8

FIFO_SETUP: 0xD0

FIFO_TRIG_CFG: 0x4

CTRL_REG1: 0x5

CTRL_REG2: 0x2

CTRL_REG3: 0x0

CTRL_REG4: 0x40

CTRL_REG5: 0x40

This all works fine - I route the device to interrupt my KE06128Z (on a FRDM-KE06Z board) via KeyBoardInterrupt0. In the ISR I read all 192 Acceleration values in I2C burst mode (2 each for X, Y, and Z axis - 14 bit mode - times 32 buffer locations). Then later, in my main routine, I dump them all out, (see below), converted into "g's".

The data shows the device triggers far below the 1.5 g threshold! Why is that? The 16th reading, the Watermark (WM), should be at the 1.5g threshold, shouldn't it? I've also dumped the raw accelerometer value for the WM reading, so you can see that my conversion to "g's" is correct.

I've tried setting the Debounce count to 0, to insure I don't lose any readings - the results are the same: the readings are all below the Motion Detection Threshold many times.

I also tried reading the accelerometer data directly, once I got a Motion Detected interrupt. I got the same results: many times, the acceleration readings were far below the specified Motion Detection Threshold. Why is that? Thanks for your help!

Sean

FIFO 00: data:  x= -0.012   y= -0.009   z=  0.045

FIFO 01: data:  x= -0.005   y= -0.017   z=  0.047

FIFO 02: data:  x= -0.009   y= -0.016   z=  0.043

FIFO 03: data:  x= -0.010   y= -0.025   z=  0.044

FIFO 04: data:  x= -0.010   y= -0.030   z=  0.049

FIFO 05: data:  x= -0.016   y= -0.036   z=  0.056

FIFO 06: data:  x= -0.018   y= -0.037   z=  0.064

FIFO 07: data:  x= -0.020   y= -0.044   z=  0.068

FIFO 08: data:  x= -0.026   y= -0.040   z=  0.077

FIFO 09: data:  x= -0.024   y= -0.038   z=  0.084

FIFO 10: data:  x= -0.030   y= -0.033   z=  0.090

FIFO 11: data:  x= -0.030   y= -0.028   z=  0.103

FIFO 12: data:  x= -0.032   y= -0.024   z=  0.107

FIFO 13: data:  x= -0.034   y= -0.021   z=  0.107

FIFO 14: data:  x= -0.031   y= -0.015   z=  0.114

FIFO 15: data:  x= -0.027   y= -0.009   z=  0.111

                Raw WM x= 0xFFFFFF82    y= 0xFFFFFFD6   z= 0x1B2

FIFO WM: data:  x= -0.031   y= -0.010   z=  0.106

FIFO 17: data:  x= -0.026   y=  0.000   z=  0.101

FIFO 18: data:  x= -0.021   y=  0.006   z=  0.100

FIFO 19: data:  x= -0.016   y=  0.006   z=  0.091

FIFO 20: data:  x= -0.012   y=  0.008   z=  0.079

FIFO 21: data:  x= -0.006   y=  0.006   z=  0.071

FIFO 22: data:  x= -0.000   y=  0.011   z=  0.064

FIFO 23: data:  x=  0.003   y=  0.009   z=  0.053

FIFO 24: data:  x=  0.001   y=  0.006   z=  0.043

FIFO 25: data:  x=  0.004   y=  0.007   z=  0.025

FIFO 26: data:  x=  0.001   y=  0.005   z=  0.005

FIFO 27: data:  x=  0.004   y=  0.004   z= -0.005

FIFO 28: data:  x=  0.008   y=  0.004   z= -0.021

FIFO 29: data:  x=  0.016   y=  0.003   z= -0.046

FIFO 30: data:  x=  0.011   y=  0.003   z= -0.056

FIFO 31: data:  x=  0.016   y=  0.003   z= -0.070

[seanbeatty]

Thanks, Tomas -

Sean

[seanbeatty]

Thanks, Tomas - that did it!

I'm curious as to why it works .... I would expect a high-pass filter to pass quick changes in motion/acceleration on, not block them. High values of acceleration should be "passed through" ... at least that's what I assumed. This seems to work just the opposite. Any comment? Can you direct me to where I might find more information on this feature?

Thanks again for your help!

Sean

[TomasVaverka]

Hi Sean,

The high-pass filter eliminates the DC tilt offset (effect of gravity) and low frequency acceleration changes, allowing only the higher frequencies to pass. The high-pass filtered data are used by the transient detection function by default (HPF_BYP bit = 0) whereas the motion detection function, which you have decided to use, uses “unfiltered” data. Have a look at our app notes AN4071 and 4070 for more information.

Regards,

Tomas

[TomasVaverka]

Hi Sean,

Looking at your settings, I noticed that you have the high-pass filter enabled (XYZ_DATA_CFG = 0x10). It explains why you are seeing such small values - the output data is high-pass filtered. If you disable the high-pass filter (XYZ_DATA_CFG = 0x00), you should get the expected acceleration readings.

Let me know if you have any other questions.

Regards,

Tomas

PS: If my answer helps to solve your question, please mark it as "Correct". Thank you.