Forums

signal processing for real time gas measurements

Started by jalbme January 8, 2008
hello friends,

i am using blackfin adsp bf537 for embedded and signal processing
purpose.
application is multiple gas analyzer and making first prototype of it.i
want to smooth digital signal and to remove noise. so after analog  to
digital converter, i will use bf537. but, i am still puzzling for signal
processing stuff. should i have to use kalman filtering or moving average
filtering for smoothing and noise removal purpose of real time signal? or
can i do this stuff using bf537? please give me your suggestions. if you
want further data on it , let  me know.


On Tue, 08 Jan 2008 06:36:02 -0600, jalbme wrote:

> hello friends, > > i am using blackfin adsp bf537 for embedded and signal processing > purpose. > application is multiple gas analyzer and making first prototype of it.i > want to smooth digital signal and to remove noise. so after analog to > digital converter, i will use bf537. but, i am still puzzling for signal > processing stuff. should i have to use kalman filtering or moving > average filtering for smoothing and noise removal purpose of real time > signal? or can i do this stuff using bf537? please give me your > suggestions. if you want further data on it , let me know.
Eh? Implementing things like moving average or Kalman filters is what a DSP is _for_, where does the question "or can I ... using the DSP" come from? Yes, you should probably choose some filter to accentuate your signal to noise ratio, but you have a long row to hoe at this point. Here are the steps that you need to contemplate: * Understand the nature of the problem -- you suspect that you're going to get a noisy signal, why? * Determine the signal processing needed to get the best signal to noise ratio, then determine if this SNR is good enough -- this is the step that'll tell you if you need a moving average or Kalman filter, or something in between. * Understand the nature of the pieces that you can implement your signal processing on -- this is the step that'll tell you if you can do it on the DSP, or if you need to do something before your signal hits the ADC, or if you need an FPGA between the ADC and the DSP (or multiple DSP chips). * Put it all together. -- Tim Wescott Control systems and communications consulting http://www.wescottdesign.com Need to learn how to apply control theory in your embedded system? "Applied Control Theory for Embedded Systems" by Tim Wescott Elsevier/Newnes, http://www.wescottdesign.com/actfes/actfes.html
hello,

thanks a lot for your thought provoking reply..
well, one thing is sure that i have to use bf537 after ADC. i am going to
use it just for embedded purpose. but, i can do signal processing ,
filtering stuff using bf537. after doing this using bf537, i am not sure
whether do i need further filtering or smoothing of real time signals to
measure gas concentration? i am very keen to decide it right now before
starting my work. because it will be main topic of my master thesis.thats
why i am trying to clear predicted picture of project in next step.thank
you.

Jalpa
>On Tue, 08 Jan 2008 06:36:02 -0600, jalbme wrote: > >> hello friends, >> >> i am using blackfin adsp bf537 for embedded and signal processing >> purpose. >> application is multiple gas analyzer and making first prototype of
it.i
>> want to smooth digital signal and to remove noise. so after analog to >> digital
converter, i will use bf537. but, i am still puzzling for signal
>> processing stuff. should i have to use kalman filtering or moving >> average filtering for smoothing and noise removal purpose of real time >> signal? or can i do this stuff using bf537? please give me your >> suggestions. if you want further data on it , let me know. > >Eh? > >Implementing things like moving average or Kalman filters is what a DSP >is _for_, where does the question "or can I ... using the DSP" come
from?
> >Yes, you should probably choose some filter to accentuate your signal to
>noise ratio, but you have a long row to hoe at this point. Here are the
>steps that you need to contemplate: > >* Understand the nature of the problem -- you suspect that you're going >to get a noisy signal, why? > >* Determine the signal processing needed to get the best signal to noise
>ratio, then determine if this SNR is good enough -- this is the step >that'll tell you if you need a moving average or Kalman filter, or >something in between. > >* Understand the nature of the pieces that you can implement your signal
>processing on -- this is the step that'll tell you if you can do it on >the DSP, or if you need to do something before your signal hits the ADC,
>or if you need an FPGA between the ADC and the DSP (or multiple DSP >chips). > >* Put it all together. > >-- >Tim Wescott >Control systems and communications consulting >http://www.wescottdesign.com > >Need to learn how to apply control theory in your embedded system? >"Applied Control Theory for Embedded Systems" by Tim Wescott >Elsevier/Newnes, http://www.wescottdesign.com/actfes/actfes.html >