third order filter ?
Hi,
I am trying to monitor an analog signal
from a sensor that is very noisy.
I had found that there is a way to derive
an average of the inputs and get a useable
value.
it may even be possible to do this in hardware.
Does anybody know where I can get more information
on a method of getting a stable value from
a noisy signal ?
Dave
I am trying to monitor an analog signal
from a sensor that is very noisy.
I had found that there is a way to derive
an average of the inputs and get a useable
value.
it may even be possible to do this in hardware.
Does anybody know where I can get more information
on a method of getting a stable value from
a noisy signal ?
Dave
Comments
Don Lancaster's Active Filter Cookbook is an excellent resource for quick
filter designs. Simple R/C networks are effective in many cases, but they
have limitations that active (Op-Amp) filters can eliminate.
National Semiconductor, Linear Technology, Burr-Brown (now part of TI),
Analog Devices, and others, sell filter chips that improve on Op-Amp
filters in various ways, but they'll cost you more. Design software and
application notes for these chips are often free at vendor web sites, and
they do sometimes treat sensor signal conditioning issues. Analog Devices
published a book titled "Transducer Interfacing Handbook, A Guide to Analog
Signal Conditioning".
>I am trying to monitor an analog signal
>from a sensor that is very noisy.
>
>I had found that there is a way to derive
>an average of the inputs and get a useable
>value.
>
>it may even be possible to do this in hardware.
>
>Does anybody know where I can get more information
>on a method of getting a stable value from
>a noisy signal ?
Mike Hardwick, for Decade Engineering -- <http://www.decadenet.com>
Manufacturer of the famous BOB-II Serial Video Text Display Module!
A while back I had to input an inexpensive pyrometer at about a 10 mS rate
into an analog input module on a micro PLC. The PLC manufacturer said it
couldn't be done. The sensor was so noisy that it gave output all over the
map, and the PLC actually followed it. I was pressed for time, so I made a
simple filter by increasing the rate so as to allow more samples in the
prescribed period, then simply discarding a number of the highest and lowest
samples from the average. I appreciate that there are more professional
methods, but this one worked quickly, and well. I seem to remember that I
used the values derived on only the second attempt. One key was being able
to visualize the data by displaying it in a simple excel graph (the PLC was
used to store it). Although there was no pattern to the noise, I was able to
determine what percentage of samples were good ones (close enough to the
curve to not adversely affect performance with the given PID) do come up
with simple numbers for averaging and discarding. Now I have the lowest cost
system in my industry group that works as well as my competitors' at about
20% of the cost. I recall doing something similar with a Stamp program for a
temporary test circuit also.
Chris
>
Original Message
> From: Dave Mucha [noparse]/noparse]mailto:[url=http://forums.parallaxinc.com/group/basicstamps/post?postID=KJx2cd-fWkBrj9Dxfdp7wJGPAUwV9YisGzc2GxyneA-333E95Pc2QA0Y1e3aIalInAjJH2Vzeg8MR20]davemucha@j...[/url
> Sent: Wednesday, May 09, 2001 12:41 PM
> To: basicstamps@yahoogroups.com
> Subject: [noparse][[/noparse]basicstamps] third order filter ?
>
>
> Hi,
>
> I am trying to monitor an analog signal
> from a sensor that is very noisy.
>
> I had found that there is a way to derive
> an average of the inputs and get a useable
> value.
>
> it may even be possible to do this in hardware.
>
> Does anybody know where I can get more information
> on a method of getting a stable value from
> a noisy signal ?
>
> Dave
>
>
>
>
> Your use of Yahoo! Groups is subject to
http://docs.yahoo.com/info/terms/
Good solutions are always worth listening to.
Dave
--- In basicstamps@y..., "Chris Loiacono (E-mail)" <chris01@t...>
wrote:
> I know this is a bit delayed - but....
>
> A while back I had to input an inexpensive pyrometer at about a 10
mS rate
> into an analog input module on a micro PLC. The PLC manufacturer
said it
> couldn't be done. The sensor was so noisy that it gave output all
over the
> map, and the PLC actually followed it. I was pressed for time, so I
made a
> simple filter by increasing the rate so as to allow more samples in
the
> prescribed period, then simply discarding a number of the highest
and lowest
> samples from the average. I appreciate that there are more
professional
> methods, but this one worked quickly, and well. I seem to remember
that I
> used the values derived on only the second attempt. One key was
being able
> to visualize the data by displaying it in a simple excel graph (the
PLC was
> used to store it). Although there was no pattern to the noise, I
was able to
> determine what percentage of samples were good ones (close enough
to the
> curve to not adversely affect performance with the given PID) do
come up
> with simple numbers for averaging and discarding. Now I have the
lowest cost
> system in my industry group that works as well as my competitors'
at about
> 20% of the cost. I recall doing something similar with a Stamp
program for a
> temporary test circuit also.
>
> Chris
>
> >
Original Message
> > From: Dave Mucha [noparse][[/noparse]mailto:davemucha@j...]
> > Sent: Wednesday, May 09, 2001 12:41 PM
> > To: basicstamps@y...
> > Subject: [noparse][[/noparse]basicstamps] third order filter ?
> >
> >
> > Hi,
> >
> > I am trying to monitor an analog signal
> > from a sensor that is very noisy.
> >
> > I had found that there is a way to derive
> > an average of the inputs and get a useable
> > value.
> >
> > it may even be possible to do this in hardware.
> >
> > Does anybody know where I can get more information
> > on a method of getting a stable value from
> > a noisy signal ?
> >
> > Dave
> >
> >
> >
> >
> > Your use of Yahoo! Groups is subject to
> http://docs.yahoo.com/info/terms/
>... I made a
>simple filter by increasing the rate so as to allow more samples in the
>prescribed period, then simply discarding a number of the highest and lowest
>samples from the average. I appreciate that there are more professional
>methods, but this one worked quickly, and well. I seem to remember that I
>used the values derived on only the second attempt. One key was being able
>to visualize the data by displaying it in a simple excel graph (the PLC was
>used to store it). Although there was no pattern to the noise, I was able to
>determine what percentage of samples were good ones (close enough to the
>curve to not adversely affect performance with the given PID) do come up
>with simple numbers for averaging and discarding. Now I have the lowest cost
>system in my industry group that works as well as my competitors' at about
>20% of the cost. I recall doing something similar with a Stamp program for a
>temporary test circuit also.
I concur with Chris about the effectiveness of that kind of filter
and think the analysis of the statistical distribution of the process
is a really good way to go about constructing it.
A "median filter" is one that stores up a certain number of samples
in a table, adding new ones while discarding the oldest ones. At
each step, the output of the filter is the median value in the table.
That kind of filter is very robust against pop noise. It is often
the first stage of digital filtering. The output of the median
filter can then feed into an averaging or a lowpass filter, which is
less resistant to pop noise.
-- regards,
Tracy Allen
electronically monitored ecosystems
mailto:tracy@e...
http://www.emesystems.com
in the DSP world as a sliding box car filter. It has another useful
feature. When you analyze its frequency response, it has a flat region
extending to DC, but also several side lobes with deep notches at higher
frequencies, as is usual with most LP filters. The useful part is that if
you choose the sample rate and N correctly, you can position one of the deep
notches close to 60 Hz, (or 50 Hz if you live in Europe), and have a very
effective line frequency notch filter, without any added circuitry or
software.
I use this with PICS and Stamps, choosing N to be 16 for easy averaging
without floating point calculations.
Dennis O'Leary
WSR Inc
Pasadena CA
Original Message
From: Chris Loiacono [noparse]/noparse]mailto:[url=http://forums.parallaxinc.com/group/basicstamps/post?postID=nM943wDmBhd6PAZiKMiVwPWq_X3_ZB8_Saia4qIJHGfQn84rPw9y7Z02tBJSJ_LoXZYFT8FR3AsN1A]chris@m...[/urlOn Behalf Of Chris
Loiacono (E-mail)
Sent: Thursday, May 10, 2001 8:40 AM
To: basicstamps@yahoogroups.com
Subject: RE: [noparse][[/noparse]basicstamps] third order filter ?
I know this is a bit delayed - but....
A while back I had to input an inexpensive pyrometer at about a 10 mS rate
into an analog input module on a micro PLC. The PLC manufacturer said it
couldn't be done. The sensor was so noisy that it gave output all over the
map, and the PLC actually followed it. I was pressed for time, so I made a
simple filter by increasing the rate so as to allow more samples in the
prescribed period, then simply discarding a number of the highest and lowest
samples from the average. I appreciate that there are more professional
methods, but this one worked quickly, and well. I seem to remember that I
used the values derived on only the second attempt. One key was being able
to visualize the data by displaying it in a simple excel graph (the PLC was
used to store it). Although there was no pattern to the noise, I was able to
determine what percentage of samples were good ones (close enough to the
curve to not adversely affect performance with the given PID) do come up
with simple numbers for averaging and discarding. Now I have the lowest cost
system in my industry group that works as well as my competitors' at about
20% of the cost. I recall doing something similar with a Stamp program for a
temporary test circuit also.
Chris
>
Original Message
> From: Dave Mucha [noparse]/noparse]mailto:[url=http://forums.parallaxinc.com/group/basicstamps/post?postID=VluuvOgSKaVjzdyH1CWl2Sv24K06AEMd5E_HsRfxLl272QknOHVEbUTqTdgnqqNKemZLb4YyucU]davemucha@j...[/url
> Sent: Wednesday, May 09, 2001 12:41 PM
> To: basicstamps@yahoogroups.com
> Subject: [noparse][[/noparse]basicstamps] third order filter ?
>
>
> Hi,
>
> I am trying to monitor an analog signal
> from a sensor that is very noisy.
>
> I had found that there is a way to derive
> an average of the inputs and get a useable
> value.
>
> it may even be possible to do this in hardware.
>
> Does anybody know where I can get more information
> on a method of getting a stable value from
> a noisy signal ?
>
> Dave
>
>
>
>
> Your use of Yahoo! Groups is subject to
http://docs.yahoo.com/info/terms/
Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/