Hi, I am working on a WiMAX project. My jobs is to handle ADC/DAC interface. I am not familiar to Signal Processing. So i need your experiences about filtering. I have search about WiMAX and decimation a lot. We implement 10 MHz, 1024 FFT WiMAX system. In our design, analog signal is sampled at 40 MHz. Oversampled by 4 times. So in digital world, first we should filter the signal by LPF and then take one sample from every four samples. I am working on TMSC6455 DSK. I have mesaured TI's DSPLib filtering functions and my owns. Since OFDM symbol duration is 103 us and this is a real-time application, the best filter i can design can be 120 tabs. I don't have any experience on Filtering. So I am worry if 120 tabs FIR filter can be used for OFDM-WiMAX application? Is 120 tabs FIR filter enough for 10 MHz WiMAX system? Burhan TÜRKEL R&D - TG03 Nortel Netaş Alemdağ Cad. No:171 34768 Ümraniye İstanbul – Turkey Tel: +90 (216) 522 5632 bturkel@netas.com.tr
WiMAX Decimation FIR Filter
Started by ●September 7, 2010
Reply by ●September 7, 20102010-09-07
On Sep 7, 8:50 am, burhan <turk...@gmail.com> wrote:> Hi, > > I am working on a WiMAX project. My jobs is to handle ADC/DAC > interface. I am not familiar to Signal Processing. So i need your > experiences about filtering. > I have search about WiMAX and decimation a lot. We implement 10 MHz, > 1024 FFT WiMAX system. In our design, analog signal is sampled at 40 > MHz. Oversampled by 4 times. So in digital world, first we should > filter the signal by LPF and then take one sample from every four > samples. I am working on TMSC6455 DSK. I have mesaured TI's DSPLib > filtering functions and my owns. Since OFDM symbol duration is 103 us > and this is a real-time application, the best filter i can design can > be 120 tabs. I don't have any experience on Filtering. So I am worry > if 120 tabs FIR filter can be used for OFDM-WiMAX application? Is 120 > tabs FIR filter enough for 10 MHz WiMAX system? > > Burhan TÜRKEL > R&D - TG03 > Nortel Netaş > Alemdağ Cad. No:171 > 34768 Ümraniye > İstanbul – Turkey > Tel: +90 (216) 522 5632 > btur...@netas.com.trHopefully the person or persons responsible for the processing of the data after you sample it has/have an idea about how strongly out of band or aliased energy has to be rejected and how well does the phase response of the data need to be kept nearly linear? I assume from your reference to an FIR filter implies you are concerned about phase linearity. I would also investigate using recursive filters as 120 taps is a lot of taps! There should be a system level design that specifies what you need. If not, how do you know why it doesn't work when it fails or why it sort of works but not very well? Good Luck, Clay
Reply by ●September 8, 20102010-09-08
On Sep 7, 5:26 pm, Clay <c...@claysturner.com> wrote:> On Sep 7, 8:50 am, burhan <turk...@gmail.com> wrote: > > > > > > > Hi, > > > I am working on a WiMAX project. My jobs is to handle ADC/DAC > > interface. I am not familiar to Signal Processing. So i need your > > experiences about filtering. > > I have search about WiMAX and decimation a lot. We implement 10 MHz, > > 1024 FFT WiMAX system. In our design, analog signal is sampled at 40 > > MHz. Oversampled by 4 times. So in digital world, first we should > > filter the signal by LPF and then take one sample from every four > > samples. I am working on TMSC6455 DSK. I have mesaured TI's DSPLib > > filtering functions and my owns. Since OFDM symbol duration is 103 us > > and this is a real-time application, the best filter i can design can > > be 120 tabs. I don't have any experience on Filtering. So I am worry > > if 120 tabs FIR filter can be used for OFDM-WiMAX application? Is 120 > > tabs FIR filter enough for 10 MHz WiMAX system? > > > Burhan TÜRKEL > > R&D - TG03 > > Nortel Netaş > > Alemdağ Cad. No:171 > > 34768 Ümraniye > > İstanbul – Turkey > > Tel: +90 (216) 522 5632 > > btur...@netas.com.tr > > Hopefully the person or persons responsible for the processing of the > data after you sample it has/have an idea about how strongly out of > band or aliased energy has to be rejected and how well does the phase > response of the data need to be kept nearly linear? I assume from your > reference to an FIR filter implies you are concerned about phase > linearity. I would also investigate using recursive filters as 120 > taps is a lot of taps! > > There should be a system level design that specifies what you need. If > not, how do you know why it doesn't work when it fails or why it sort > of works but not very well? > > Good Luck, > ClayI agree with Clay; there's no point asking if 120 taps is enough without more information, specifially asking the question "Enough for what?" Do you have a target for performance/out-of-band rejection/etc? Can you test the performance of the system using different numbers of taps/different cutoff frequencies/different filter designs (FIR vs IIR being one question)? Even testing a very much idealized system (for example, OFDM with no channel) could give a lot of information and will show up any fundamental flaws (such as the cutoff frequency being inappropriate, ripple being too large, or phase distorion problems, etc.). If you can answer these questions, then you'll answer your original question; however, without more information I don't think it's possible to say whether or not your filter has "enough taps". Frank
Reply by ●September 13, 20102010-09-13
On Sep 8, 12:10 pm, Frank <frank.snow...@gmail.com> wrote:> On Sep 7, 5:26 pm, Clay <c...@claysturner.com> wrote: > > > > > > > On Sep 7, 8:50 am, burhan <turk...@gmail.com> wrote: > > > > Hi, > > > > I am working on a WiMAX project. My jobs is to handle ADC/DAC > > > interface. I am not familiar to Signal Processing. So i need your > > > experiences about filtering. > > > I have search about WiMAX and decimation a lot. We implement 10 MHz, > > > 1024 FFT WiMAX system. In our design, analog signal is sampled at 40 > > > MHz. Oversampled by 4 times. So in digital world, first we should > > > filter the signal by LPF and then take one sample from every four > > > samples. I am working on TMSC6455 DSK. I have mesaured TI's DSPLib > > > filtering functions and my owns. Since OFDM symbol duration is 103 us > > > and this is a real-time application, the best filter i can design can > > > be 120 tabs. I don't have any experience on Filtering. So I am worry > > > if 120 tabs FIR filter can be used for OFDM-WiMAX application? Is 120 > > > tabs FIR filter enough for 10 MHz WiMAX system? > > > > Burhan TÜRKEL > > > R&D - TG03 > > > Nortel Netaş > > > Alemdağ Cad. No:171 > > > 34768 Ümraniye > > > İstanbul – Turkey > > > Tel: +90 (216) 522 5632 > > > btur...@netas.com.tr > > > Hopefully the person or persons responsible for the processing of the > > data after you sample it has/have an idea about how strongly out of > > band or aliased energy has to be rejected and how well does the phase > > response of the data need to be kept nearly linear? I assume from your > > reference to an FIR filter implies you are concerned about phase > > linearity. I would also investigate using recursive filters as 120 > > taps is a lot of taps! > > > There should be a system level design that specifies what you need. If > > not, how do you know why it doesn't work when it fails or why it sort > > of works but not very well? > > > Good Luck, > > Clay > > I agree with Clay; there's no point asking if 120 taps is enough > without more information, specifially asking the question "Enough for > what?" > > Do you have a target for performance/out-of-band rejection/etc? > Can you test the performance of the system using different numbers of > taps/different cutoff frequencies/different filter designs (FIR vs IIR > being one question)? > Even testing a very much idealized system (for example, OFDM with no > channel) could give a lot of information and will show up any > fundamental flaws (such as the cutoff frequency being inappropriate, > ripple being too large, or phase distorion problems, etc.). > > If you can answer these questions, then you'll answer your original > question; however, without more information I don't think it's > possible to say whether or not your filter has "enough taps". > > FrankThanks Frank and Clay. You are right, the information about the system is very limited. I have no experience about filter design. Maybe, someone who have work on OFDM or WiMAX will tell us his/her experiences. Burhan.
