ADXL372 Peak detection questions

Hi Yukio, 

PdelCorro 

Below is a capture of the datasheet section explaining some of the issues. 

Below are my answers to the questions: 

Q1.The longer the duration of the shock pulse and the lower the shock acceleration, the more unstable the peak detection becomes.

　　Can you give any advice on this?

MM - This is mostly due to the difference between the threshold set and the shock value. As explained in the datasheet section above, the peak detection feature requires at least 2 points above the threshold to give an accurate reading. By having a shock way higher than the threshold (50g shock to 15g threshold), even if the duration of the shock pulse is shorter, there was still enough of a rising edge above the threshold to get an accurate reading. In the first case, the shock is lower in magnitude (17g): 

• 10 g threshold case: The difference between the shock and threshold is only 7g, the result will be less accurate as there will be less datapoints above the threshold. 
• 15g threshold case: the shock is too close to the threshold chosen, as the maximum sensor error in peak detection is +/- 3g.

Q2.Peak detection operation does not occur when the threshold installation is 1G or 3G.　Can you give any advice on this, also?

MM - As stated in the datasheet section above, the resolution in peak detection is lower than when the sensor is in normal operation. Because of that, the maximum error is approximatively +/- 3g. The noise floor being in the peak detection range, there will be no detection possible. 

Regards, 

Marine 

Hi PdelCorro,

Thank you for your prompt response.

I would like to set the HPF_CORNER Settings from 11 to 10 to lower the noise floor.

I have three additional questions.

1. I understand that there is an error of ±3g in the output peak value. By the way, is the activity threshold triggered with an accuracy of ±0.1g?

2. I have the inactivity threshold set to 1/2 of the activity threshold. Is this setting appropriate?

3. I set TIME_ACT to 3.3ms (6400HzODR) and TIME_INACT to 13ms x 5 = 65ms. Are these settings reasonable?

Best regards,

Yukio

Hi Yukio ,

I will answer every question below. 

"I would like to set the HPF_CORNER Settings from 11 to 10 to lower the noise floor."

• We do not recommend setting the HPF_CORNER from 11 to 10 to lower the noise floor, as the low pass activity filter has a cutoff frequency of 10Hz.  At 6400Hz ODR, If HPF_CORNER is set to b'10, the -3dB corner is 7.88Hz. It is a bit too close to detect a shock event. Below is the Low pass activity filter section of the DS. 

"1. I understand that there is an error of ±3g in the output peak value. By the way, is the activity threshold triggered with an accuracy of ±0.1g?"

• That is correct. The activity inactivity is triggered with an accuracy of 0.1g as it is a digital comparison between the threshold level and the LSB output of the accelerometer (12-bit resolution)

"2. I have the inactivity threshold set to 1/2 of the activity threshold. Is this setting appropriate?"

• The inactivity threshold set to 1/2 of the activity threshold is appropriate, providing that there is at least 2 datapoints are within the rise part of the shock. 

"3. I set TIME_ACT to 3.3ms (6400HzODR) and TIME_INACT to 13ms x 5 = 65ms. Are these settings reasonable?"

• Those are valid settings, but whether they are reasonable depends on the application.

Thanks, 

Marine ( )

Hi Marine,

Thank you for the advice. HPF_CORNER Settings should be left at 11.

I don't have much time before our product release so I was wondering

if you could answer additional questions in pdf.

PDF

Best regards,

Yukio

Hi Yukio ,

I will answer the following questions below: 

"I would like to change the circuit to supply power (VS,VDDIO) of ADXL372 from GPIO of CPU. Is there any problem if the power supply is started up in the sequence shown in the left figure?" 

• I would refer to the Power Supply Requirements section of the datasheet. Below is the rise time subsection. The GPIO power circuit must have a rise time ≤200 μs. Make sure the GPIO is in push pull configuration. 

"Would an inductor of 6.8 μF be more appropriate than a 10 μF for connecting one cell?"

• This is a question about the LTC3525L-3 and the LTC525L-3.3, so this is out of our scope. I would recommend posting on the power management forum for more guidance here:  Power Management .
• From what I could gather from the datasheet of the LTC535L-3.3,
  • the AAA battery is a single cell 1.5V nominal battery, so 6.8 μH is more appropriate than a 10 μH inductor.
  • The D battery is also single cell and 1.5V, however it is Alkaline, so I am not sure if a 6.8 μH is more appropriate than a 10 μH inductor. 

Thank you, 

Marine 

Hi Marine,

Thank you for your kind advice. It was very helpful.

Thank you,

Yukio