SparkFun Electronics Forum

Hello I am working on a rocket model and my budget is tight and limited. Thuss, I need help to make the adequate choice regarding an accelerometer.

The rocket maximum acceleration is 250m/s^2. I hesitate between choosing :

• ADXL377 3axis +- 200g (since the rocket's maximum acceleration is 250m/s^2, the resolution of this accelerometer worries me)

or,

• ADXL78 one axis +-50g with H3LIS331DL 3axis +-100g (the ADXL technology is the most suitable for rocket so using one axis ADXL can be a bad idea).

Note : The accelerometer will be used to detect the apogee using kalman filter.

Be more precise in what worries you about the (over-)scaling in the former. And why a single axis is a bad idea for the latter. You can always mount 3 in perpendicular axis if you can manage it spacewise. Besides, both have analog outputs. So 'resolution' doesn't really apply, you probably mean sensitivity. Amplifiers can fix that.

And why is the second option with the additional H3LIS331DL so much different from the first. (in respect to resolution/sensitivity) It's only half as much overscale. Aside from the 2 chips being redundant on 1 axis.

If the only purpose is determining apogee, then one axis is all you need.

If the only sensor is the accel then the Kalman filter is wasted.

@UhClem: Don't you agree that for apogee detection, one would want to use a +/- 2g (or similar) accelerometer? I suspect that with a 100 or 200 g scale, the acceleration near apogee would be lost in the noise.

jremington wrote: ↑Thu Jun 07, 2018 5:55 pm @UhClem: Don't you agree that for apogee detection, one would want to use a +/- 2g (or similar) accelerometer?

Absolutely not. What you are doing is integrating the acceleration and waiting for the velocity to go from positive to negative.

I still have a BlackSky AltAcc2 which uses an ADXL50 accel with an 8 bit ADC. While it had trouble in some conditions (that would cause trouble with any accel based system) it worked well.

What you are doing is integrating the acceleration and waiting for the velocity to go from positive to negative.

Agreed. How is that better done with a 100 g accelerometer, near apogee when the engine is not firing?

I think you would want both, to more accurately measure the two acceleration regimes.

jremington wrote: ↑Thu Jun 07, 2018 7:19 pm I think you would want both, to more accurately measure the two acceleration regimes.

If you use one then scale factor (slope) errors do not matter since you start at zero velocity and end at zero. But if you use one accel under thrust and a different accel near apoge, their scale factor errors matter. Plus you have to worry about aligning the two packages.

Alignment of two 3D accelerometers is not required. Just measure a single acceleration (e.g. g) as a 3D vector on each, to get the relative orientation of one to the other.

Valen wrote: ↑Thu Jun 07, 2018 5:21 am Be more precise in what worries you about the (over-)scaling in the former. And why a single axis is a bad idea for the latter. You can always mount 3 in perpendicular axis if you can manage it spacewise. Besides, both have analog outputs. So 'resolution' doesn't really apply, you probably mean sensitivity. Amplifiers can fix that.

And why is the second option with the additional H3LIS331DL so much different from the first. (in respect to resolution/sensitivity) It's only half as much overscale. Aside from the 2 chips being redundant on 1 axis.

thank you so much for your answer, what worries me about the ADXL377 (+/- 200G) is the range of the sensor, the range on apogee of static calibration is too small to assure good accurancy over the +/- 200g! i can't risk losing such information so i thought about using +/-50g, the sensor of the second option isn't recommanded in rockets so i'm not sure about using it!

UhClem wrote: ↑Thu Jun 07, 2018 10:20 am If the only purpose is determining apogee, then one axis is all you need.

If the only sensor is the accel then the Kalman filter is wasted.

hello, thank you for your answer
i need to collect the acceleration's DATAs as well, so i must use three axis.
the kalman filter will help to detect apogee ( the sensor detect the burnout ) so why would it be wasted?

so why would it be wasted?

Because a primary function of the Kalman filter is to merge information from two or more sources. Used with only one source, a Kalman filter reduces to a simple low pass filter.

meriem wrote: ↑Thu Jun 14, 2018 7:36 pm i need to collect the acceleration's DATAs as well, so i must use three axis.
the kalman filter will help to detect apogee ( the sensor detect the burnout ) so why would it be wasted?

The extra two axis will pick up vibration and not much else. So unless you are trying to measure that, it is a waste. Here is an example:
(Captured at 4,000SPS. )


Absent a pressure sensor, applying a Kalman filter to the data would be equivalent to integrating the acceleration. But at greater computational cost.

Some work I have done on the subject:
http://home.earthlink.net/~david.schult ... index.html
http://home.earthlink.net/~david.schult ... index.html

UhClem wrote: ↑Fri Jun 15, 2018 1:35 pm

meriem wrote: ↑Thu Jun 14, 2018 7:36 pm i need to collect the acceleration's DATAs as well, so i must use three axis.
the kalman filter will help to detect apogee ( the sensor detect the burnout ) so why would it be wasted?

The extra two axis will pick up vibration and not much else. So unless you are trying to measure that, it is a waste. Here is an example:

xvibration.png
(Captured at 4,000SPS. )


Absent a pressure sensor, applying a Kalman filter to the data would be equivalent to integrating the acceleration. But at greater computational cost.

Some work I have done on the subject:
http://home.earthlink.net/~david.schult ... index.html
http://home.earthlink.net/~david.schult ... index.html

thank you so much for your help .

jremington wrote: ↑Thu Jun 14, 2018 8:05 pm

so why would it be wasted?

Because a primary function of the Kalman filter is to merge information from two or more sources. Used with only one source, a Kalman filter reduces to a simple low pass filter.

got it, thank you so much .