MMA8451Q accessing FIFO and xyz samples concurrently

[tom_ferree]

My question has two parts:

1. We have ODR set to 400 Hz.  We have configured the FIFO to trigger (FMODE = 11) on pulse event with watermark in the middle of the 32-sample FIFO buffer.  I think this implies that the 80 msec wide buffer has 40 msec from the pulse event (and watermark) to the time when the FIFO buffer is full and the interrupt is triggered.  Is that correct, i.e., in FMODE = 11 is the interrupt generated not at the watermark but rather 40 msec later when the buffer is full?

2.  For separate purpose, running concurrently, our firmware polls the most recent xyz-data sample at 50 Hz or every 20 msec (by reading the usual xyz-data register 0x01, not the FIFO buffer).  When we have this code in place, and we read the FIFO buffer following the pulse interrupt, we see FIFO values of -8160 filling one or more samples at the end of the FIFO, and the data spike that generated the pulse event is shifted that number of samples from the watermark.  Is it possible that reading 0x01 is corrupting the FIFO?  Is there a way to configure and use the accelerometer so we can get both FIFO for each pulse event and also poll ordinary data stream from 0x01 concurrently?

[david_diaz]

Hello Thomas,

Thank you for writing.

In this case, I created a sample project in which the MMA8451Q is set as follow: ODR=400Hz, FIFO Full Buffer Mode, interrupt enabled, I2C baud rate = 125KHz. You may refer to the images below in order to get a better understanding of the watermark:

1)

2)

The first image shows the time that passed, about 15ms, in order to fill the FIFO buffer using a watermark of 32. The second image shows the time that passed, about 7.8ms, with a watermark of 16. In both cases, an interrupt is generated (overflow).

So we can conclude that, indeed, your assumption is correct.

Please let me know if I misunderstood your question.

Regards,

David

[tom_ferree]

Hi David,

Thanks very much for your reply.  I really appreciated your taking time to set up and run that experiment.  I think that answered part 1 of my question.  Do you have any thoughts on part 2?  Does the -8160 value look familiar to you?  Is there a way to configure and use the accelerometer so we can get both FIFO for each pulse event and also poll ordinary data stream from 0x01 concurrently?

Thanks again!

Tom.

[david_diaz]

Hello Thomas,

In regard with the second question, please note that the sample data output registers (0x01 to 0x06), store the current sample data if the FIFO data output register driver is disabled, but if the FIFO data output register driver is enabled (F_MODE > 00) the sample data output registers point to the head of the FIFO buffer (register 0x01 X_MSB) which contains the previous 32 X, Y, and Z-data samples.

So the Register 0x01 points to the FIFO buffer and is used to access 8-bit or 14-bit data. You may want to wait until the FIFO buffer is full in order to read the data.

Please let me know if my answer is clear. I am glad to help.

Regards,

David