Movement Filter

[speedyreidy]

Hello all,

I'm working on a robotic hand that mimics the user's movement. I am trying to use SE-08606 flex resistors (as a voltage divider) with an Arduino to detect finger curling into a fist. My problem is that there is a lot of noise and the servos that curl the fingers become really jittery. I'm thinking that I could use a low pass filter to get rid of the noise (my thought process is it will filter out the noise into a smooth movement of the servos). I put the circuit below with the filter I was trying to use (which doesn't work). Do you think there is any low pass that would work for this? Or if I'm even approaching this noise reduction issue properly?

Much appreciated

 

[speedyreidy]

To clarify, the current filter is charging and discharging the capacitor, causing the Vout to fluctuate.

 

[steveB]

There is a great lack of critical information in your question. You did not provide component values for your circuit and you did not provide any specifications about what cutoff frequency (or even general range) you need. Also, the filter will interact with any control feedback loop you have, assuming you have one. If you do have one, then information about that is important because the entire feedback loop must be designed as a unit, including the measurement filter.

 

[Cicero]

Why not just use a simple RC low pass, and ditch the op amp? Be aware an LM324 isn't rail-to-rail as well.

 

[RichTheDude]

Steve makes many good points, more information is needed.

Many times you can reduce the noise just by averaging in the micro-controller, I do this by taking a number of samples N as of 2^N. You can perform a simple binary shift to the division quickly. No idea how fast you can sample on an arduino mind, never used one in anger.

 

[4pyros]

Or if I'm even approaching this noise reduction issue properly?

What is making all the noise?

 

[MikeMl]

... Or if I'm even approaching this noise reduction issue properly?

To answer that, you need to record the signal you are attempting to isolate with no hand movement. Store that in a file called A.
Then record the signal again with normal hand movement>Store that in a file called B.

Now use spectrum analysis software to compute the FFT first of file A, and then separately for file B. Now compare the spectra of A and B. If the noise lands on top of the spectrum of the signals attributed to hand movement, you cannot use frequency domain filtering to separate the two...