I might have misread your email; I was assuming that you didn't
 change the relative position of the antennas, but only used an
 additional cable to connect them to the USRP.
 can you send us a photo of the two configurations you're comparing?

 Your 104.6MHz signal has a wavelength of about 3m. If you affix
 both antennas directly to the USRP, you don't get wave propagation
 between the two antennas. These two antennas would directly
 couple, since they are in reactive near field, not in far field.

 Best regards,
 Marcus


 On 06/04/2015 12:19 PM, Marcus Müller wrote:

 Hi Siva,

 20dB more gain should really do the trick, considering that with
 50dB you still seem to be deep in a linear range of the RX
 amplifiers. I'm pretty sure the B210 is not to blame here --
 there's something wrong with your cable or antenna:

 if you matched your system well, 5m cable should do not much more
 than delaying your RF signal by
 $\approx\frac{\SI{5}{\meter}}{\frac23
 c}=\frac32\frac{\SI{5}{\meter}}{\SI{3e8}{\meter \per
 \second}}=\SI{25}{\nano\second}$, given you didn't chose wrong
 cabling. Now, $\SI{104.6}{\mega\hertz}$ should be covered even by
 the cheapest coax cabling, so that's probably not the case. What
 kind of cable are you using?

 Though the common mixup of $\imp$ (which should be used, as it is
 the USRP's impedance) and $\wrongimp$ cables only cost about 20%
 of voltage per interface, there /are/ other impedances out there,
 and with a bit of bad luck, you might end up with a cable whose
 impedance is further away fro $\imp$ than it should. But still, a
 $\SI{125}{\ohm}$ cable would lead to a reflection loss of 60%, so
 not the order of magnitudes you're seeing.

 So the next suspect is antenna matching. What kind of antenna are
 you using? What's its wave impedance at the given frequency?

 Best regards,
 Marcus

 On 06/04/2015 11:32 AM, siva sankar via USRP-users wrote:

 Hi list, 

 We are using an Ettus B210 with UHD version 3.8.4. 

 In my previous query to the list for zero lag correlation,
 Marucs suggested to eliminate set_time_now() which has worked
 for us. This is working fine when both the receiver antennas are
 on board(we are getting a peak at zero lag). 

 The idea was if both the receivers are at different distances
 then when we perform correlation between the received signals
 then we expected to see some lag in the peak with which to
 estimate the time difference of arrival of the two signals. 

 While this works marginally well (we see peaks at non zero lag
 roughly 2 out of 10 attempts) when both antennas are on the
 board, when one antenna is connected to the board via a 5m
 cable, the performance drops drastically and nothing meaningful
 can be observed. 

 Attached are screenshots of the signal amplitude for both cases,
 with and without the cable. The performance of the latter
 doesn't improve even at higher gains. 

 USB-3.0
 sample rate 15.36Msps 
 rx-frequency 104e6
 gain 50db(when antennas on board)
        upto 70db (when one antenna is 5m away)
 nsamps=2 million.

 -- 
 Thanks and regards
 Siva Sankar.


 _______________________________________________
 USRP-users mailing list
 USRP-users@lists.ettus.com <mailto:USRP-users@lists.ettus.com>
 http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

 _______________________________________________
 USRP-users mailing list
 USRP-users@lists.ettus.com <mailto:USRP-users@lists.ettus.com>
 http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

Hi Siva, there's several thing wrong here: * you noticed that the Vert2450 is an 2.4GHz ISM band antenna. Why are you using it then? It's not even closely appropriate for a 100MHz signal. When using an antenna that is specified for roughly 24 times the frequency you're sending on, you will get far less power out of the antenna. In your use case without the cable, you do, however, don't use the VERT2450 as an antenna -- I'd rather think of these two pieces of wire (which these monopoles are, really) as a capacitor with some parasitic inductance (which models the part of energy that doesn't directly get absorbed by the RX antenna). * the USRPs and Vert2450 have regular SMA, not Reverse SMA. You should be having one center pin too many at the antenna-side connector, and one too little on the USRP side. * as I said, if you arrange two antennas (no matter which ones, really) in a proximity of less than 1% of the RF wavelength, they will not work as individual antennas -- power will directly couple from one antenna into the other. So your TX antenna doesn't radiate into all directions -- it sinks a majority of its energy directly into the RX antenna. As much as we love SDR, we still bow to Maxwell and Hertz, because we can't change physics. * In your directly-attached use case, only a little part of the TX energy is radiated, whereas in your 5m-distance case, you'll get something that more closely resembles the unidirectional behaviour of a monopole. Notice that this only *very* approximately applies here, as one of the monopoles is still attached perpendicular to a ground plane (the B210 itself), AND both are far too short -- it's pretty impossible to predict radiation pattern in this scenario without precise simulation, AND 5m is still not far-field for these frequencies. * This is real world transmissions, your amplitude will fluctuate, especially in a near-field situation with the wrong antennas. In fact, your body, holding the vert2450 probably increases the efficiency of the VERT2450 for 100MHz significantly. You stop your every move and go into an anechoic chamber, and you'll get a much more consistent amplitude :) * 100 MHz relatively easily passes through walls (which is why you can get FM radio indoors), so anything that moves in your building might be influencing your transmission. Read up on fading. 2/3 c is an approximation of the propagation velocity in a coax cable. There's a bit of EM theory behind this, but basically, for coax, you get $\frac{c_\text{coax}}{c_0} = \frac{1}{\sqrt{\kappa_\text{dielectric}}}$, because the wave propagates through the dielectric medium between the inner and outer conductor of the cable, in which the speed of light is defined by the material's dielectric constant. Solid polyethylene is most often found in low-power, high-bandwidth coax, which has a $\kappa_\text{PE}\approx 2.25$, and hence, EM waves in these cables propagate with $\frac23 c_0$. Best regards, Marcus On 06/05/2015 10:41 AM, siva sankar wrote: > Hi Marcus, > > We did give more 20db gain but its still the same. I used a Reverse > SMA male to female antenna extension cable of 50 Ohms impedance > matched and we are using the ettus Vert 2450 dual ISM band antenna. > > In your first response you have taken 2/3 of speed of light, could you > explain this ? > > We performed the correlation in the second set-up where one of the > antenna is 5m away from the first antenna, the peak of the correlation > is varying and is not constant. Is there any way to sort this ? > > PFA both the configuration set ups. > > PS: while performing the experiment we fixed the position of the > antenna(not holding it with hand). > > Thanks > Siva > > On Thu, Jun 4, 2015 at 4:01 PM, Marcus Müller > <usrp-users@lists.ettus.com <mailto:usrp-users@lists.ettus.com>> wrote: > > I might have misread your email; I was assuming that you didn't > change the relative position of the antennas, but only used an > additional cable to connect them to the USRP. > can you send us a photo of the two configurations you're comparing? > > Your 104.6MHz signal has a wavelength of about 3m. If you affix > both antennas directly to the USRP, you don't get wave propagation > between the two antennas. These two antennas would directly > couple, since they are in reactive near field, not in far field. > > Best regards, > Marcus > > > On 06/04/2015 12:19 PM, Marcus Müller wrote: >> Hi Siva, >> >> 20dB more gain should really do the trick, considering that with >> 50dB you still seem to be deep in a linear range of the RX >> amplifiers. I'm pretty sure the B210 is not to blame here -- >> there's something wrong with your cable or antenna: >> >> if you matched your system well, 5m cable should do not much more >> than delaying your RF signal by >> $\approx\frac{\SI{5}{\meter}}{\frac23 >> c}=\frac32\frac{\SI{5}{\meter}}{\SI{3e8}{\meter \per >> \second}}=\SI{25}{\nano\second}$, given you didn't chose wrong >> cabling. Now, $\SI{104.6}{\mega\hertz}$ should be covered even by >> the cheapest coax cabling, so that's probably not the case. What >> kind of cable are you using? >> >> Though the common mixup of $\imp$ (which should be used, as it is >> the USRP's impedance) and $\wrongimp$ cables only cost about 20% >> of voltage per interface, there /are/ other impedances out there, >> and with a bit of bad luck, you might end up with a cable whose >> impedance is further away fro $\imp$ than it should. But still, a >> $\SI{125}{\ohm}$ cable would lead to a reflection loss of 60%, so >> not the order of magnitudes you're seeing. >> >> So the next suspect is antenna matching. What kind of antenna are >> you using? What's its wave impedance at the given frequency? >> >> Best regards, >> Marcus >> >> On 06/04/2015 11:32 AM, siva sankar via USRP-users wrote: >>> Hi list, >>> >>> We are using an Ettus B210 with UHD version 3.8.4. >>> >>> In my previous query to the list for zero lag correlation, >>> Marucs suggested to eliminate set_time_now() which has worked >>> for us. This is working fine when both the receiver antennas are >>> on board(we are getting a peak at zero lag). >>> >>> The idea was if both the receivers are at different distances >>> then when we perform correlation between the received signals >>> then we expected to see some lag in the peak with which to >>> estimate the time difference of arrival of the two signals. >>> >>> While this works marginally well (we see peaks at non zero lag >>> roughly 2 out of 10 attempts) when both antennas are on the >>> board, when one antenna is connected to the board via a 5m >>> cable, the performance drops drastically and nothing meaningful >>> can be observed. >>> >>> Attached are screenshots of the signal amplitude for both cases, >>> with and without the cable. The performance of the latter >>> doesn't improve even at higher gains. >>> >>> USB-3.0 >>> sample rate 15.36Msps >>> rx-frequency 104e6 >>> gain 50db(when antennas on board) >>> upto 70db (when one antenna is 5m away) >>> nsamps=2 million. >>> >>> -- >>> Thanks and regards >>> Siva Sankar. >>> >>> >>> _______________________________________________ >>> USRP-users mailing list >>> USRP-users@lists.ettus.com <mailto:USRP-users@lists.ettus.com> >>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >> > > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com <mailto:USRP-users@lists.ettus.com> > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com > > > > > -- > Thanks and regards > Siva Sankar.