Dear All, I am working with the DBSRX2 front end and USRP2. I am trying to get the GPS signals. In fact I have got it, I can synchronize with GPS using an external-active antenna. The problem is the C/No (Signal Power/ Noise density) estimated. I am comparing the C/No estimated with the USRP2 and the C/No estimated with a GPS receiver. Both used the same antenna. The point is with the USRP2 I measure 17dB less than with the normal GPS. I known that a GPS receiver has been builded only for GPS system and sure that it must work better than the USRP2. But 17dB is too much. I also have a WBX front end. Using this front end I obtain more or less the same C/No than with the GPS receiver, that makes me think the problem is on the DBSRX2 front end. So I have tried to measure the Noise figure ( DBSRX2 + USRP2) but the obtained results are inconsistent. Also, I have mesured the ouput gain as function of the USRP2 gain. I have done this in order to check if the Gains of the USRP2 works fine. I have two DBSRX2. Figure 1 shows the Gain curve. At top we can see two plots for the two DBSRX2 front ends. As we can see we have not any output gain up to 10dB gain of USRP2. For me, this could be for two reasons: 1) We have some problem with some LNA, amplifier or somethink like this. 2) The input power signal is too low. I think that the second one is impossible because, if I plot the spectrum of the received signal (Figure 2) I can see the imput signal. As you can see, Im using a single tone at 0.4MHz. On baseband we see a spurious. It is normal? Why I dont see any change in the first 10dB? maybe, There are any problem with my front ends? As you can see in Figure 1 at bottom, I get a slope for all the gains using a XCVR2450 front end ( I know that this front end works in another frequency range). There are any test points in the board in order to check if all works fine? NOISE FIGURE: I don't know if im doing something wrong. Im going to explaning the method that I have used in order to measure the noise figure. I used a single tone with power = Pin ( This power is known). So, I can calculate the SNRin SNRin = Pin / Pn Pn = kTB where k is the Boltzmann constant, T is the temperature 293 B is the sample frequency i.e. 100e6/dicimation SNR = PindBm - 173.9325 -10*log(B) In order to calculate the SNRout I have done the FFT of the receiver signal with N samples. X = abs( FFT(X) ).^2 SNRout = max(X)/N / Noise_flat SNRoutdB = 10*log(SNRout) Noise_flat is the value of the measured noise in a bin close to the bin that has the signal. ( I have done the mean value of several bins) at the end: N = SNRindB - SNRoutdB and I have got values around 39 dB. I know that this value is too large. I have tried with different Gains, diferents front ends.. and the value is more or less the same. Am I doing something wrong? in Figure 3 you can see the abs ( FFT (x) ).^2, using a -40dmB input signal, 2^15 samples in the FFT ( without zero-padding). The USRP2 gain is 70dB and the Fs = 5Mhz (i.e. Decimation factor = 20 ) I also have tried with different decimation values and the resolts are the same. ( Also with decimation values multiples of 4). IN SUMMARY: Do you think there is a problem with my front ends? What is the expected noise factor using DBSRX2 + USRP2? Do you know why I get such high values​​? I have the received signal saved... so If you want it in order to calculate the noise figure... I can send it to you. -- Moisés Navarro Ph.D student phone: +34935868009 Q6/2008 Bellaterra Universitat Autònoma de Barcelona

Moses: The DBSRX2 has about 80dB of gain control range, which is what your figure shows. The gain range starts around 10dB. Hence the flat-spot in your curves for DBSRX2. Further, the DBSRX2 has a noise figure of approximately 4-5dB, but only at maximum gain at lower gain settings, the noise figure becomes worse. In variable-gain receivers,noise figure is dependent on gain setting, due to the use of a variable attenuator in front of the 2nd gain stage. I also suggest that you investigate "Y-factor method" for calculating noise figure of a receiver system. Also, make sure you're making your measurements with the gain turned up on the DBSRX2. If you still see significant deterioration of C/No with the gain turned up, then there are a couple of different possibilities: o You have damaged the LNA on both your DBSRX2 boards o the phase-noise of the DBSRX2 synthesizer may be higher in critical areas as compared to the synthesizer on the WBX. With the gain turned up on your DBSRX2 to maximum, you should easily be able to "see" a -140dBm narrowband input signal (from a signal generator) in the FFT display. If not, then it's likely that the LNA has become damaged. They are sensitive to ESD, and to input signals above about -10dBm. -Marcus On Thu, 19 Jan 2012 17:48:31 +0100, MOISES NAVARRO wrote: > Dear All, > > I am working with the DBSRX2 front end and USRP2. > I am trying to get the GPS signals. In fact I have got it, I can synchronize with GPS using an external-active antenna. > > The problem is the C/No (Signal Power/ Noise density) estimated. I am comparing the C/No estimated with the USRP2 and the C/No estimated with a GPS receiver. Both used the same antenna. > > The point is with the USRP2 I measure 17dB less than with the normal GPS. I known that a GPS receiver has been builded only for GPS system and sure that it must work better than the USRP2. But 17dB is too much. > > I also have a WBX front end. Using this front end I obtain more or less the same C/No than with the GPS receiver, that makes me think the problem is on the DBSRX2 front end. > > So I have tried to measure the Noise figure ( DBSRX2 + USRP2) but the obtained results are inconsistent. Also, I have mesured the ouput gain as function of the USRP2 gain. I have done this in order to check if the Gains of the USRP2 works fine. > > I have two DBSRX2. Figure 1 shows the Gain curve. At top we can see two plots for the two DBSRX2 front ends. As we can see we have not any output gain up to 10dB gain of USRP2. > For me, this could be for two reasons: > > 1) We have some problem with some LNA, amplifier or somethink like this. > 2) The input power signal is too low. > > I think that the second one is impossible because, if I plot the spectrum of the received signal (Figure 2) I can see the imput signal. As you can see, Im using a single tone at 0.4MHz. On baseband we see a spurious. > > It is normal? Why I dont see any change in the first 10dB? maybe, There are any problem with my front ends? > As you can see in Figure 1 at bottom, I get a slope for all the gains using a XCVR2450 front end ( I know that this front end works in another frequency range). > > There are any test points in the board in order to check if all works fine? > > NOISE FIGURE: > I don't know if im doing something wrong. Im going to explaning the method that I have used in order to measure the noise figure. > > I used a single tone with power = Pin ( This power is known). So, I can calculate the SNRin > > SNRin = Pin / Pn > Pn = kTB > where k is the Boltzmann constant, > T is the temperature 293 > B is the sample frequency i.e. 100e6/dicimation > SNR = PindBm - 173.9325 -10*log(B) > > In order to calculate the SNRout I have done the FFT of the receiver signal with N samples. > > X = abs( FFT(X) ).^2 > > SNRout = max(X)/N / Noise_flat > SNRoutdB = 10*log(SNRout) > > Noise_flat is the value of the measured noise in a bin close to the bin that has the signal. ( I have done the mean value of several bins) > > at the end: > > N = SNRindB - SNRoutdB > > and I have got values around 39 dB. > I know that this value is too large. I have tried with different Gains, diferents front ends.. and the value is more or less the same. > > Am I doing something wrong? > in Figure 3 you can see the abs ( FFT (x) ).^2, using a -40dmB input signal, 2^15 samples in the FFT ( without zero-padding). The USRP2 gain is 70dB and the Fs = 5Mhz (i.e. Decimation factor = 20 ) > > I also have tried with different decimation values and the resolts are the same. ( Also with decimation values multiples of 4). > > IN SUMMARY: > > Do you think there is a problem with my front ends? > > What is the expected noise factor using DBSRX2 + USRP2? > > Do you know why I get such high values​​? > > I have the received signal saved... so If you want it in order to calculate the noise figure... I can send it to you. > > -- > Moisés Navarro > Ph.D student > phone: +34935868009 > Q6/2008 > Bellaterra > Universitat Autònoma de Barcelona