High-pass filter without phase shift

Hi,

I'm trying to design a high-pass filter to filter out DC component (bias)
from a noisy accelerometer measurement. I've tried to go with a second
order Butterworth filter with coefficients from
http://www-users.cs.york.ac.uk/~fisher/mkfilter/trad.html. The filtering
works just fine, but the filter has a non-linear phase shift, so I don't
know if that's the right choice. I'd also like to go with a higher order
filter than this.
The problem is, higher order filters designed with the above site are not
stable, so they generally wind up. Is there a way to design a high-pass
filter with cut-off frequency of about 0,5-1Hz without phase shift (or with
small negative shift) in the pass band? I'd appreciate some practical
links/examples.

Thanks and regards,
Tom

On Nov 15, 4:27�pm, "futureignobel"
<futureignobel@n_o_s_p_a_m.gmail.com> wrote:
> Hi,
>
> I'm trying to design a high-pass filter to filter out DC component (bias)
> from a noisy accelerometer measurement. I've tried to go with a second
> order Butterworth filter with coefficients fromhttp://www-users.cs.york.ac.uk/~fisher/mkfilter/trad.html. The filtering
> works just fine, but the filter has a non-linear phase shift, so I don't
> know if that's the right choice. I'd also like to go with a higher order
> filter than this.
> The problem is, higher order filters designed with the above site are not
> stable, so they generally wind up. Is there a way to design a high-pass
> filter with cut-off frequency of about 0,5-1Hz without phase shift (or with
> small negative shift) in the pass band? I'd appreciate some practical
> links/examples.
>
> Thanks and regards,
> Tom

why not subtract the bias? Or does it vary with time?

On Mon, 14 Nov 2011 21:27:15 -0600, "futureignobel"
<futureignobel@n_o_s_p_a_m.gmail.com> wrote:

>Hi,
>
>I'm trying to design a high-pass filter to filter out DC component (bias)
>from a noisy accelerometer measurement. I've tried to go with a second
>order Butterworth filter with coefficients from
>http://www-users.cs.york.ac.uk/~fisher/mkfilter/trad.html. The filtering
>works just fine, but the filter has a non-linear phase shift, so I don't
>know if that's the right choice. I'd also like to go with a higher order
>filter than this.
>The problem is, higher order filters designed with the above site are not
>stable, so they generally wind up. Is there a way to design a high-pass
>filter with cut-off frequency of about 0,5-1Hz without phase shift (or with
>small negative shift) in the pass band? I'd appreciate some practical
>links/examples.
>
>Thanks and regards,
>Tom
>
Hello Tom,
   I'm not sure it will help you, but 
you might have a look at:

http://www.dsprelated.com/showarticle/58.php

Good Luck,
[-Rick-]

futureignobel wrote:

> I'm trying to design a high-pass filter to filter out DC component (bias)
> from a noisy accelerometer measurement. I've tried to go with a second
> order Butterworth filter with coefficients from
> http://www-users.cs.york.ac.uk/~fisher/mkfilter/trad.html.

First mistake: don't use some junk from the net.

> The filtering works just fine,

If it works fine, what else do you need?

> but the filter has a non-linear phase shift,

And this is a problem why?

> so I don't know if that's the right choice.

Then you have to decide what is the right choice.

> I'd also like to go with a higher order filter than this.

What for?

> The problem is, higher order filters designed with the above site are not
> stable, so they generally wind up.

Second mistake: don't use some junk from the net.

> Is there a way to design a high-pass
> filter with cut-off frequency of about 0,5-1Hz without phase shift (or with
> small negative shift) in the pass band? 

Yes. But I doubt you will like it.

> I'd appreciate some practical
> links/examples.

A practical example? Subtract a rolling average calculated over the span 
of 1 second from your initial signal.


> Thanks and regards,
> Tom

I'll send you a bill.



Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com