Feasibility of Analog Integrator Circuit to Integrate Accel'or o/p for Arduino.

Projects General Guidance
1

Context

I am building an IC to calculate finger tap velocity (And other control enabling data like Non-inertial acceleration, tilt etc.) and relay the data to my android. I am using an accelerometer over a speedometer: 1. I already had it from my UG project. 2. It is mulitipurpose and can be upgraded to an IMU with Gyroscope, giving more data and control. 3. It is easier to fit in a rubber glove, even with No-Glove-System, it doesn't hamper the finger movement as much as a speedometer would.(Or so is my assumption from online comparison.)

I am working on an Arduino Uno and using an ADXL 335 as my Accelerometer with a future outlook to upgrade the set-up with a Gyroscope to improve cancellation of Normal forces on the accelerometer, i.e. phase 2. If you feel this project will need better instruments, please recommend to your delight.

For now, I have cancelled the normal forces(Calculated the Gravitational acceleration in my locality) on my Z axis, as I am only looking out for Non-inertial forces i.e., Just wanna know my value bruh... not the value gravity and society exerts on me bruh. I'm limiting to one axis before upgrading to phase 2 for Testing simplicity and focussed debugging.

Crux of the Post

As a finger tap isn't a very long process (Unless done by Flash the Sloth from Zootopia), I reckoned Software Integrating wouldn't accumulate perceivable error. It is erratic with compounding errors but worked just about in a limited application. Then I had an epiphany(from my AIC classes) about using an Analog Integrated Circuit Integrator that we make using an Op-Amp and its viable use in my application.

Can I feed the Accelerometer Analog Output into the integrator(Which integrates and outputs the Integral of the Input)?

If so...

How can I convert the resultant velocity signal into applicable velocity format i.e m/s? I know the mV/G sensitivity and Zero G values of my sensor, but how can I convert them to use for the Velocity signal?

If conversion is not possible, is there any viable way to calculate those new Values required to convert the Velocity Voltage?

Possible Pitfalls

A problem I think this system will face is the Normal force/Gravity, as Integrator won't be able to make a difference between Gravitational acceleration and Inertial Acceleration. And, weeding out the Gravitational error after Integration would be as messy as a Pollock Painting. Or I am not qualified enough to come up with a way to fix it. So, can anyone suggest a way to remove the gravitational error before the integrator, while keeping the analog, non-discrete nature of the system?

Conclusion

This is my first Post here. I've read the "Before you post" post and searched this forum for similar ideas. Googled for it too. But, I did not find any viable results that use hardware for integrating. All of them either showcase How to integrate using software, thus in Discrete signal... Or showcase why NOT to integrate using software. So here I am, looking for a solution or work on creating one with your help.

PS. I am a fiction writer on the side so my language might be a bit odd and non-academic.

2

Shaurcasm:
I've read the "Before you post" post

Maybe you missed this part?

  1. Forum etiquette

Once you have made your post, here are a few other guidelines:

Don't cross-post!

We know you are anxious to get an answer to your problem, but re-posting the same question to multiple parts of the forum causes duplicate effort, and divides the focus on your problem. You'll get a better answer if everyone can discuss it in one place. Cross-posting is very annoying; the moderators will delete your extra posts, and you will get a warning. Save all the hassle, and just post the question once. It will be noticed.

If you aren't sure what section to post in (Programming Questions, Electronics, etc.) make a guess. Don't cross post "just in case".

I see two copies of your post in this forum section...

3

Shaurcasm:
Can I feed the Accelerometer Analog Output into the integrator(Which integrates and outputs the Integral of the Input)?

Sure, that will work, but the usualy integrator issues apply, drift, zeroing and infinite DC gain.

If so...

How can I convert the resultant velocity signal into applicable velocity format i.e m/s? I know the mV/G sensitivity and Zero G values of my sensor, but how can I convert them to use for the Velocity signal?

"applicable velocity format"? You mean convert to particular units?

You can figure out the accelerometer sensitivity in V/(m/s^2) from the mV/g value
You can determine the integrator constant from the RC values in s^-1 (simply 1/RC), multiply these
two constants and you get V/(m/s)
to get the velocity sensitivity.

4

If your acceleration sensor is not fast enough to catch details of a finger tip, there remains nothing to improve.

I wonder what you consider "the" velocity that you want to measure. A finger tip starts and ends at zero velocity, and in between you'll find a velocity curve. As the only meaningful value I can imagine the maximum velocity or acceleration, reached during the tip.

I also wonder why you bother with linear acceleration and velocity, instead of angular velocity (gyroscope). You can have both in one small chip, though.

5

No. You have to integrate after gravity is removed, which is on the digital side.

Usually intehration is the wrong method. Looking for sudden stops like taps is best done by looking at acceleration without integration. You can add reliability by looking at the gyros and making a threshold rotational speed that must be reached before the sudden stop.