Filter coefficient calculation In Sigma studio
in [DSP]
|
From: rammya.tv on 8 Feb 2010 23:36 hi.... i'm currently working on Analog Devices Dsp processor ADAU 1701.. Sigma studio is the software we are now on,in that i have some queries in the way they calculate filter coefficient. - main specification i needed to design a filter 1)Type[LPF,HPF, etc] 2)frequency 3)Q 4)Gain eg: I want a 2nd order LPF with Frequency= 1000 Q=0.5 Gain=1 The given below derivation is the one which i got from sigma studio help window. ->ω0 = 2*pi*f0/Fs ->gainLinear = 10^(gain/20) ->Lowpass ->Transfer Function ->H(s)=1/(s^2+(s/Q)+1) ->Coefficients ->alpha = sin(ω0)/(2*Q) ->a0 = 1 + alpha ->a1 = -2*cos(ω0) ->a2 = 1 - alpha ->b0 = (1 - cos(ω0)) * gainLinear / 2 ->b1 = 1 - cos(ω0) * gainLinear ->b2 = (1 - cos(ω0)) * gainLinear / 2 After compiling we'll get coefficients a1,a2,b0,b1,b2 in hex format in CAPTURE WINDOW of software. i did manual calculation using the expression provided above and compare with that of coeffients of software, but it differs. Please help me to get detail calculation with the specification i given above. We know ω0 = 2*pi*f0/Fs -what is the value of pi(180 0r 3.14) -ω0 is the angular representation of requency i think it will be 180 -then in above value of alpha will be zero always. -it means Q doesnt have any importantce in filter design -please suggest a good book which describe in detail about filter design with regards rammya
From: Jerry Avins on 9 Feb 2010 09:40 rammya.tv wrote: > hi.... > i'm currently working on Analog Devices Dsp processor ADAU 1701.. > Sigma studio is the software we are now on,in that i have some queries in > the way they calculate filter coefficient. > > - main specification i needed to design a filter > 1)Type[LPF,HPF, etc] > 2)frequency > 3)Q > 4)Gain > > eg: > I want a 2nd order LPF with > Frequency= 1000 > Q=0.5 > Gain=1 > > The given below derivation is the one which i got from sigma studio help > window. > > ->ω0 = 2*pi*f0/Fs > ->gainLinear = 10^(gain/20) > > ->Lowpass > > ->Transfer Function > ->H(s)=1/(s^2+(s/Q)+1) > > ->Coefficients > ->alpha = sin(ω0)/(2*Q) > ->a0 = 1 + alpha > ->a1 = -2*cos(ω0) > ->a2 = 1 - alpha > ->b0 = (1 - cos(ω0)) * gainLinear / 2 > ->b1 = 1 - cos(ω0) * gainLinear > ->b2 = (1 - cos(ω0)) * gainLinear / 2 I can't see your omega unless I switch to unicode. It seems gauche, but it is the custom here to use 'w' instead. > After compiling we'll get coefficients a1,a2,b0,b1,b2 in hex format in > CAPTURE WINDOW of software. > i did manual calculation using the expression provided above and compare > with that of coeffients of software, but it differs. > Please help me to get detail calculation with the specification i given > above. > > > We know ω0 = 2*pi*f0/Fs > -what is the value of pi(180 0r 3.14) > -ω0 is the angular representation of requency i think it will be 180 > -then in above value of alpha will be zero always. > -it means Q doesnt have any importantce in filter design > -please suggest a good book which describe in detail about filter design It will be 180 what? All angles above are in radians. 'w' (or ω) is for radians/second. For cycles/second (Hz), it's 'f'. One cycle is 2*pi radians. That pi radians equal 180 degrees doesn't make pi (Π) equal to 180. There's no magic that can make it so. Jerry -- Engineering is the art of making what you want from things you can get. ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
From: robert bristow-johnson on 9 Feb 2010 15:38 On Feb 8, 11:36 pm, "rammya.tv" <rammya...(a)ymail.com> wrote: > hi.... >  i'm currently working on Analog Devices Dsp processor ADAU 1701.. > Sigma studio is the software we are now on,in that i have some queries in > the way they calculate  filter coefficient. > > - main specification i needed to design a filter > 1)Type[LPF,HPF,  etc] > 2)frequency > 3)Q > 4)Gain > > eg: > I want a 2nd order LPF with > Frequency= 1000 > Q=0.5 > Gain=1 > > The given below derivation is the one which i got from sigma studio help > window. > > ->Ï0 = 2*pi*f0/Fs > ->gainLinear = 10^(gain/20) > > ->Lowpass > > ->Transfer Function > ->H(s)=1/(s^2+(s/Q)+1)  > > ->Coefficients > ->alpha = sin(Ï0)/(2*Q) > ->a0 =  1 + alpha > ->a1 =  -2*cos(Ï0) > ->a2 =  1 - alpha > ->b0 =  (1 - cos(Ï0)) * gainLinear / 2 > ->b1 =  1 - cos(Ï0)  * gainLinear > ->b2 =  (1 - cos(Ï0)) * gainLinear / 2 > > After compiling we'll get coefficients a1,a2,b0,b1,b2 in hex format in > CAPTURE WINDOW of software. > i did manual calculation using the expression provided above and compare > with that of coeffients of software, but it differs. > Please help me to get detail calculation with the specification i given > above. > >  We know Ï0 = 2*pi*f0/Fs > -what is the value of pi(180 0r 3.14) the latter. > -Ï0 is the angular representation of frequency i think it will be 180 w0 is unitless but would be "radians per sample" and is always less than pi. > -then in above value of alpha will be zero always. > -it means Q doesnt have any importantce in filter design GIGO > -please suggest a good book which describe in detail about filter design it looks like you've been to the cookbook. there *is* actually sufficient detail in the "FYI" at the bottom to show how all the coefs were designed. but look for Zoelzer's book, i guess. the formulae have been used over and over and i'm confident are correct for how the parameters are defined in the cookbook. try reading it carefully and not making any funky assumptions. there are plenty of folks here who can help you. r b-j r b-j
From: rammya.tv on 9 Feb 2010 23:19 rammya.tv hi.... i'm currently working on Analog Devices Dsp processor ADAU 1701.. Sigma studio is the software we are now on,in that i have some queries in the way they calculate filter coefficient. - main specification i needed to design a filter 1)Type[LPF,HPF, etc] 2)frequency 3)Q 4)Gain eg: I want a 2nd order LPF with Frequency= 1000 Q=0.5 Gain=1 The given below derivation is the one which i got from sigma studio help window. ->w0 = 2*pi*f0/Fs ->gainLinear = 10^(gain/20) ->Lowpass ->Transfer Function ->H(s)=1/(s^2+(s/Q)+1) ->Coefficients ->alpha = sin(w0)/(2*Q) ->a0 = 1 + alpha ->a1 = -2*cos(w0) ->a2 = 1 - alpha ->b0 = (1 - cos(w0)) * gainLinear / 2 ->b1 = 1 - cos(w0) * gainLinear ->b2 = (1 - cos(w0)) * gainLinear / 2 I can't see your omega unless I switch to unicode. It seems gauche, but it is the custom here to use 'w' instead. After compiling we'll get coefficients a1,a2,b0,b1,b2 in hex format in CAPTURE WINDOW of software. i did manual calculation using the expression provided above and compare with that of coeffients of software, but it differs. Please help me to get detail calculation with the specification i given above. We know w0 = 2*pi*f0/Fs -what is the value of pi(180 0r 3.14) -w0 is the angular representation of requency i think it will be 180 -then in above value of alpha will be zero always. -it means Q doesnt have any importantce in filter design -please suggest a good book which describe in detail about filter design It will be 180 what? All angles above are in radians. 'w' (or ω) is for radians/second. For cycles/second (Hz), it's 'f'. One cycle is 2*pi radians. That pi radians equal 180 degrees doesn't make pi (Π) equal to 180. There's no magic that can make it so. Jerry -- >Engineering is the art of making what you want from things you can get. >¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯ hi jerry thanx for ur response still i didnt get what u mean?? first off all i'm new to this field. i made another mistake what i meant is in the expresion w0=2*pi*f0 value of pi 1s 180 or 3.14. with regards rammya write
From: Jerry Avins on 10 Feb 2010 08:39 rammya.tv wrote: ... > thanx for ur response still > i didnt get what u mean?? > first off all i'm new to this field. > i made another mistake > what i meant is in the expresion w0=2*pi*f0 > value of pi 1s 180 or 3.14. > with regards According to one passage in the Bible, pi is three. Various legislators at various times introduced ordinances to set the value of pi at some convenient rational fraction. No ordinance can affect the value of a physical constant. Pi remains 3.1415926535897932384626433832795... despite their efforts. 355/113 comes close. Jerry -- Engineering is the art of making what you want from things you can get.
|
Next
|
Last
Pages: 1 2 3 4 5 6 7 Prev: SPI and I2S on SHARC ADSP-21369 Next: Calculation of filter coefficients in Sigma studio |