"steve" <bungalow_steve@yahoo.com> wrote in message
news:1162578014.718252.248070@i42g2000cwa.googlegroups.com...
>
> klman wrote:
> > I want to sample a gyro and then integrate the angular rate to get the
> > angle.
> >
> >
> >
> > When sampling the signal from the gyro, how important is it to filter
away
> > frequencies higher than half the sampling rate?
> >
> >
> >
> > I should have at least enough attenuation at the sampling frequency (Fs)
so
> > that a signal at Fs Hz doesn't alias back to DC (0Hz). But what bout the
> > other frequencies that alias, can they cause any problem when
integrated?
> >
>
> Yes, you will find it very helpful to model the system  (gyro as a
> signal + noise source, filters, integral) using a modeling system like
> scicos (it's free). You just draw the system with the built in blocks
> and run it and it plots everything out. It will only take a few minutes
> and you will get a feel of what is important and what isn't, what
> levels of noise are tolerable, what sampling is required etc much
> quicker then anyone telling you.
>
> see
> http://www.scicos.org/demos.html
>

Also - the instrument can drift - so an absolute reference should be
considered.

Vladimir Vassilevsky wrote:
> 
> 
> Jerry Avins wrote:
> 
> 
>>> klman wrote:
>>>
>>>> I want to sample a gyro and then integrate the angular rate to get the
>>>> angle.
>>>> When sampling the signal from the gyro, how important is it to 
>>>> filter away
>>>> frequencies higher than half the sampling rate?
>>>
>>> It depends on what kind of noise do you have, how much of noise do 
>>> you have, how much bandwidth do you have at the output and what are 
>>> your accuracy requirements.
>>>
>>>>
>>>> I should have at least enough attenuation at the sampling frequency 
>>>> (Fs) so
>>>> that a signal at Fs Hz doesn�t alias back to DC (0Hz). But what bout 
>>>> the
>>>> other frequencies that alias, can they cause any problem when 
>>>> integrated?
> 
>>
>> Klman's question has so far been avoided. If the result is to be 
>> integrated, are non-DC aliases really harmful? If their frequencies 
>> are high enough, it would seem not.
> 
> What does it mean "really harmful" ? High enough compared to what? 
> Nonsense.
> 
> Jerry,
> 
> When I was at school, they put me in trouble every time I said a word 
> like "obvious", "many", "little", "really", "sufficient" and so on. 
> Those words are bullshit, they mean nothing. The engineers are operating 
> with the numbers and proving their point with the logic.
> 
> 
> The integration creates a 6dB rolldown frequency slope. The output will 
> be affected according to this. Is it a problem or not depends on the 
> particular application.

Isn't that pretty much what I wrote?: "If their frequencies are high 
enough ...." If their frequencies are high enough, the instantaneous 
disturbance will be small enough. In any case, the long-term effect will 
be zero.

Jerry
-- 
        "The rights of the best of men are secured only as the
        rights of the vilest and most abhorrent are protected."
            - Chief Justice Charles Evans Hughes, 1927
?????????????????????????????????????????????????????????????????????

klman wrote:
> I want to sample a gyro and then integrate the angular rate to get the
> angle.
>
>
>
> When sampling the signal from the gyro, how important is it to filter away
> frequencies higher than half the sampling rate?
>
>
>
> I should have at least enough attenuation at the sampling frequency (Fs) so
> that a signal at Fs Hz doesn't alias back to DC (0Hz). But what bout the
> other frequencies that alias, can they cause any problem when integrated?
>

Yes, you will find it very helpful to model the system  (gyro as a
signal + noise source, filters, integral) using a modeling system like
scicos (it's free). You just draw the system with the built in blocks
and run it and it plots everything out. It will only take a few minutes
and you will get a feel of what is important and what isn't, what
levels of noise are tolerable, what sampling is required etc much
quicker then anyone telling you.

see
http://www.scicos.org/demos.html

Jerry Avins wrote:


>> klman wrote:
>>
>>> I want to sample a gyro and then integrate the angular rate to get th=
e
>>> angle.
>>> When sampling the signal from the gyro, how important is it to filter=
=20
>>> away
>>> frequencies higher than half the sampling rate?
>>
>> It depends on what kind of noise do you have, how much of noise do you=
=20
>> have, how much bandwidth do you have at the output and what are your=20
>> accuracy requirements.
>>
>>>
>>> I should have at least enough attenuation at the sampling frequency=20
>>> (Fs) so
>>> that a signal at Fs Hz doesn=92t alias back to DC (0Hz). But what bou=
t the
>>> other frequencies that alias, can they cause any problem when=20
>>> integrated?

>=20
> Klman's question has so far been avoided. If the result is to be=20
> integrated, are non-DC aliases really harmful? If their frequencies are=
=20
> high enough, it would seem not.

What does it mean "really harmful" ? High enough compared to what? Nonsen=
se.

Jerry,

When I was at school, they put me in trouble every time I said a word=20
like "obvious", "many", "little", "really", "sufficient" and so on.=20
Those words are bullshit, they mean nothing. The engineers are operating =

with the numbers and proving their point with the logic.


The integration creates a 6dB rolldown frequency slope. The output will=20
be affected according to this. Is it a problem or not depends on the=20
particular application.



Vladimir Vassilevsky

DSP and Mixed Signal Design Consultant

http://www.abvolt.com

Vladimir Vassilevsky wrote:
> 
> 
> klman wrote:
>> I want to sample a gyro and then integrate the angular rate to get the
>> angle.
>> When sampling the signal from the gyro, how important is it to filter 
>> away
>> frequencies higher than half the sampling rate?
> 
> 
> It depends on what kind of noise do you have, how much of noise do you 
> have, how much bandwidth do you have at the output and what are your 
> accuracy requirements.
> 
>>
>> I should have at least enough attenuation at the sampling frequency 
>> (Fs) so
>> that a signal at Fs Hz doesn�t alias back to DC (0Hz). But what bout the
>> other frequencies that alias, can they cause any problem when integrated?
> 
> It depends.
> 
>>
>> And a sharper cut of frequency (higher order filtering) gives more phase
>> delay to the sampled signal.
>>
> 
> Mr. Wescott has a paper on basics of sampling. You should apply the 
> principles from there:
> 
> http://www.wescottdesign.com/articles/Sampling/sampling.html

Klman's question has so far been avoided. If the result is to be 
integrated, are non-DC aliases really harmful? If their frequencies are 
high enough, it would seem not.

Jerry
-- 
        "The rights of the best of men are secured only as the
        rights of the vilest and most abhorrent are protected."
            - Chief Justice Charles Evans Hughes, 1927
?????????????????????????????????????????????????????????????????????

klman wrote:
> I want to sample a gyro and then integrate the angular rate to get the
> angle.
> When sampling the signal from the gyro, how important is it to filter a=
way
> frequencies higher than half the sampling rate?


It depends on what kind of noise do you have, how much of noise do you=20
have, how much bandwidth do you have at the output and what are your=20
accuracy requirements.

>=20
> I should have at least enough attenuation at the sampling frequency (Fs=
) so
> that a signal at Fs Hz doesn=92t alias back to DC (0Hz). But what bout =
the
> other frequencies that alias, can they cause any problem when integrate=
d?

It depends.

>=20
> And a sharper cut of frequency (higher order filtering) gives more phas=
e
> delay to the sampled signal.
>=20

Mr. Wescott has a paper on basics of sampling. You should apply the=20
principles from there:

http://www.wescottdesign.com/articles/Sampling/sampling.html


Vladimir Vassilevsky

DSP and Mixed Signal Design Consultant

http://www.abvolt.com

klman wrote:
> I want to sample a gyro and then integrate the angular rate to get the
> angle.
> 
> 
> 
> When sampling the signal from the gyro, how important is it to filter away
> frequencies higher than half the sampling rate?
> 
> 
> 
> I should have at least enough attenuation at the sampling frequency (Fs) so
> that a signal at Fs Hz doesn�t alias back to DC (0Hz). But what bout the
> other frequencies that alias, can they cause any problem when integrated?
> 
> 
> 
> And a sharper cut of frequency (higher order filtering) gives more phase
> delay to the sampled signal.
> 
> 
Ideally you find out what the spectral content is of the noise coming 
out of the gyro, and you decide what (if any) anti-aliasing filtering 
you need.

Fortunately numerical integration of this type is way fast -- you're 
just doing a sum, after all.  Given that, you may be able to sample and 
integrate pretty darn fast compared to other things your processor will 
do for you.

I recently wrote a white paper on sampling that may help: 
http://www.wescottdesign.com/articles/Sampling/sampling.html.

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google?  See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

I want to sample a gyro and then integrate the angular rate to get the
angle.



When sampling the signal from the gyro, how important is it to filter away
frequencies higher than half the sampling rate?



I should have at least enough attenuation at the sampling frequency (Fs) so
that a signal at Fs Hz doesn�t alias back to DC (0Hz). But what bout the
other frequencies that alias, can they cause any problem when integrated?



And a sharper cut of frequency (higher order filtering) gives more phase
delay to the sampled signal.