0
$\begingroup$

I am trying to use a FIR filter making software to create a moving average filter and I'm a little confused on what I should put in the coefficients part? I read that the coefficients are always 1 for moving average so this is what I did - is that correct? (Picture below)

Other than the coefficients, I just wanted to double check (because I am very new to working with filters): Moving average filters are symmetrical, have no L-th band filter, and are single rate filters right?

coefficients

$\endgroup$
  • $\begingroup$ Yes a moving average filter has all 1's (or all the same value if scaled higher). If you divided the output by the number of bins, that would truly be a moving average, right? $\endgroup$ – Dan Boschen Aug 3 at 0:46
  • $\begingroup$ @DanBoschen what do you mean by number of bins? would I be able to just have 1 coeffiecient (value = 1) instead of 36 and would that be the same? $\endgroup$ – Tarashi99 Aug 3 at 0:53
  • $\begingroup$ The number of coefficient - a moving average of two samples is current plus previous sample (that would be 2 bins each equal to 1. $\endgroup$ – Dan Boschen Aug 3 at 1:03
1
$\begingroup$

I read that the coefficients are always 1 for moving average so this is what I did - is that correct?

Technically no. For a moving average filter of length $N$ all the coefficients are $1/N$. In practice most people use $1$ for all coefficients and then scale the output with $1/N$ It gives the same result and it's typically more efficient, but if you were to look at the difference equation, the coefficients would show up as $1/N$

Moving average filters are symmetrical,

Depends a bit exactly how exactly you define symmetry. You can certainly time reverse it without changing anything.

have no L-th band filter,

I have no idea what you are mean by "L-th band filter"

and are single rate filters right?

The concepts of multi-rate and moving average are mostly orthogonal. You can use the together , individually or not all.

| improve this answer | |
$\endgroup$
  • $\begingroup$ when you say "a moving average filter of length N", would N be the length of the data going through the filter? Or would it be the data in the filter at any given time? My data is 1000 bytes long, but im only ever averaging 2 at a time, so would I make the scaled value 1/2 or 1/1000? $\endgroup$ – Tarashi99 Aug 3 at 14:33
  • $\begingroup$ Also, for a moving average filter, would the number of taps affect the efficacy of the filter (given that they're all the same)? $\endgroup$ – Tarashi99 Aug 3 at 14:43
  • $\begingroup$ N is the number of taps of the filter. The number of taps affects the efficacy greatly.. In essence the moving average filter is a low pass filter (although a pretty bad one) and the length determines the cutoff frequency . Shorter filters react more "quickly" but tend to be "noisier". So it really depends on your application. $\endgroup$ – Hilmar Aug 3 at 19:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.