<patrick.melet@dmradiocom.fr> wrote in message 
news:1159949415.970392.141570@k70g2000cwa.googlegroups.com...
Hello,

I've got after the FM demodulation (by derivative : (IdQ-QdI) /
(I&#4294967295;+Q&#4294967295;)) a low pass filter.

This filter is a IIR Butterwoth 4th order. But I know that I can use a
box car filter.

But what is a box car filter ?

******************


It *could* mean all 1's in the time domain.  In this case, the frequency 
domain looks like sinx/x or the Dirichlet kernel.

It *could* mean all 1's in the frequency domain.  In this case, the time 
domain looks like sinx/x or the Dirichlet kernel.

Either one is a lowpass filter in frequency - one "better" than the other.

"Box car" has another name: a "gate" function.  You still have to decide 
which domain you're talking about.

In statistics an averager may be a boxcar averager which would most 
typically be analogous to the time domain.

In communications a boxcar filter may just be a shorthand term for a perfect 
lowpass filter - in the frequency domain.

Fred 

patrick.melet@dmradiocom.fr wrote:
> Hi,
>
> How long must be the box car filter register ? How many taps ?
>
> If I want to recover a 20 kHz sinus with 400 kSamples/s ?
>
> thanks ?

Hello Patrick,

Actually you don't want the filter to be too long. If it is very long,
then it will knock out your 20kHz signal.  What signal(s) are you
trying to reject with this filter?

Clay

Hi,

How long must be the box car filter register ? How many taps ?

If I want to recover a 20 kHz sinus with 400 kSamples/s ?

thanks ?

Hello Patrick,

Comments below:

patrick.melet@dmradiocom.fr wrote:
> Hello,
>
> I've got after the FM demodulation (by derivative : (IdQ-QdI) /
> (I=B2+Q=B2)) a low pass filter.
>
> This filter is a IIR Butterwoth 4th order. But I know that I can use a
> box car filter.
>
> But what is a box car filter ?
>
> Is it a FIR filter with rectangular window ? Or FIR filter with all the
> coefficients to value one ?


Yes, it has a sin(x)/x freq response. IT is formed by just summing the
last n samples of your data together. You may also think of it as a
moving average where the scaling by N (# of samples) is done last.

>
> I think that I just need to make an integration of my FM demodulated
> signal...

Do you need phase instead of frequency?

>
> I tried this but don't work very well : output[n] =3D FM_demodulated[n] +
> FM_demodulated[n-1] I though it was the box car filter, but I don't
> know if it is...


Make sure in your (IdQ-QdI) / (I=B2+Q=B2)) that the I(t) and Q(t) signals
are delayed so as to be properly time aligned with your dQ(t) and dI(t)
signals. If not, your results won't be very good.


IHTH,

Clay


>=20
> Best regards...

Hello,

I've got after the FM demodulation (by derivative : (IdQ-QdI) /
(I=B2+Q=B2)) a low pass filter.

This filter is a IIR Butterwoth 4th order. But I know that I can use a
box car filter.

But what is a box car filter ?

Is it a FIR filter with rectangular window ? Or FIR filter with all the
coefficients to value one ?

I think that I just need to make an integration of my FM demodulated
signal...

I tried this but don't work very well : output[n] =3D FM_demodulated[n] +
FM_demodulated[n-1] I though it was the box car filter, but I don't
know if it is...

Best regards...